Measuring Behavior [2 ed.] 0675080789

Citation preview

Measuring Behavior

Second Edition.

John O. Cooper x

The Charles E. Merrill Series on Behavioral Techniques in the Classroom Thomas M. Stephens, Consulting Editor

John O. Cooper

Measuring Behavior (Second

John O. Cooper & Denzil Edge

Parenting: Strategies and Educational Methods

John P. Glavin

Behavioral Strategies for Classroom Management

Stainback, Payne, Stainback, & Payne

Establishing a Token

Thomas M. Stephens

Directive Teaching (Second

Thomas M. Stephens

Edition)

Economy in the Classroom

Edition)

Implementing Behavioral

Approaches in Elementary and Secondary Schools

Thomas M. Stephens

Teaching Children Basic Skills

Thomas M. Stephens

Teaching Skills to Children With Learning and

Behavior Disorders

Measuring

Behavior

Second Edition

John O. Cooper The Ohio State University

Charles E. Merrill Publishing Company A Bell & Howell Company Columbus Toronto London Sydney

Published by Charles E. Merrill Publishing Company A Bell & Howell Company Columbus, Ohio 43216

Copyright © 1981 by Bell & Howell Company. All rights reserved. No part of this book may be reproduced in any form, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher.

A previous edition was published under the title Measurement and Analysis of Behavioral Techniques. Copyright © 1974 by Bell & Howell Company.

Library of Congress Catalog Card Number: ISBN:

0-675-08078-9

80-81787

Printed in the United States of America

12

34

567

8

9

10

/

86

85

84

83

82

81

To Bunny, Chris, Sharon, Greg, and Brian

Contents Foreword Preface

ix xi

Introduction

Part 1

Part 2

Measurement

3

6

]

Direct Measurement of Permanent Products

2

Observational Recording

Graphic Presentations

Techniques for Behavior Analysis

96 98

Appendix

A_

Instrumentation for observational Recording 186

Appendix

B

Preparing for Printing

Glossary

202

Index

207

70

72

Analytic Teaching 4

8

32

Reporting Effects of Instruction 3

Part

1

by Twila Johnson

196

Foreword

As Professor Cooper correctly notes in the preface to this book, recent progress in using applied behavioral technology in classrooms has been

considerable.

Because

source

Those

advances

of his methodical

for those

school

practitioners

approach to their instruction. An

improved

have

technology

been

presentation,

who

incorporated

in this edition.

this text is an indispensable take seriously

of instruction

a data-based

should be of concern

to all

school personnel. Economic considerations are the first grounds for such concern. Spiraling instructional costs demand that we have sound methods

for identifying effective classroom

treatments.

No

longer can

we afford the luxury of instructional fads and wishful thinking about how to accelerate academic learning. Professionally, we never could afford such wasteful

practices. But now,

and in the long-range

future,

we must be accountable for how our limited resources are expended. Professional considerations are also important. Sound teaching can make

a difference

in children’s learning. Teachers

now

have available

to them technology which can modify and reduce the frequency of maladaptive, inappropriate, and incorrect responses. They also have the 1X

X

Preface

power to help students maintain what they have learned. Today, those teachers who are prepared to use the technology presented here will be equipped to teach in ways that were unknown just a few decades ago. Rate of learning is increasingly important in a world where each day an increasing amount of knowledge must be acquired. Social and academic concepts and skills are typically acquired more rapidly through applied behavior analysis than through more conventional instructional approaches. Each

text in the Merrill

Series

on

Behavioral

Techniques

for the

Classroom is devoted to a particular aspect of schooling or behavioral methodology. In this book, tactics and procedures are presented for measuring

and

analyzing

in this

book

school-related

behaviors.

John

Cooper

has

measuring,

and

made an important contribution to teacher training and, thereby, to improving classroom instruction of children and young people. Readers will

find

specific

tactics

for

observing,

analyzing classroom behavior. In the past, teachers have expected others to conduct evaluations of their students’ performances. Even when others performed these services, results were often unusable because by the time evaluations were completed, the evaluated students were often no longer under the teacher's

instructional

care.

Consequently,

instructional

modifications

for those

particular students often did not occur. The approaches presented in this

text

will

enable

sults, and—when

teachers

to

conduct

observations,

evaluate

re-

necessary—modify their instruction. No longer must

teachers wonder, without ever knowing for sure, whether their instruc-

tion is effective. This book will give you sound ideas about evaluating instruction in practical ways. It will inform you, but by itself it cannot change your behavior. For that, you must practice the suggestions it contains. Thomas M. Stephens Consulting Editor

Preface

This text presents tactics for behavioral analysis. Generally the book addresses all school personnel (e.g., principals, school psychologists, counselors, therapists). However, it is designed specifically for persons preparing to become regular classroom teachers and special education teachers. Measurement and analysis are the key to successful teaching. The application of laws and principles of behavior in themselves generates no assurance of behavior change. Many programs in education may be employing specific technologies because of fad or custom rather than verified effect. The application of a technology of teaching should be based on continuous scientific verification at the classroom level by classroom teachers. This book is designed to serve as a basic textbook in measurement and analysis for directive teachers and also for seniors and beginning graduate students in education and psychology. Since the publication of the first edition (Measurement and Analysis of Behavior Techniques), much progress has been made in applying behavioral analysis techniques in the classroom. This revision was undertaken to incorporate these changes. Among the recent developments XI

xii

Preface

included in this revision are: the delayed multiple baseline design, the multiple baseline probe design, the changing criterion design, and the multielement baseline design. In addition, the measurement chapters have been updated to include a discussion of trials to criterion, measurement of the complexity of task items, pre- and posttests, and probes. The topic of measurement and analysis is covered in three parts. Part 1 outlines measurement techniques for the directive teacher. Two types of measurement are presented—direct measurement of permanent prod-

ucts and observational recordings. Part 2 is concerned with graphic presentations for teacher use. Part 3 presents some basic procedures for analyzing teacher effectiveness. I am grateful for the support and assistance of many colleagues, teachers, and students in completing this book. Special appreciation is extended to Thomas M. Stephens, friend and colleague. John O. Cooper

Introduction

A Ph.D. aspirant was once requested to express to his oral examination committee

his views

on how

children

would

be educated

in the year

2000. The student described several innovative approaches to education that he predicted would be common by the year 2000. After the candi-

date finished his discourse, an elder member of the examination committee stated that all instructional approaches espoused by the student

were employed

in his university elementary laboratory school during

the 1930s. Why, he asked the student, do we have to wait until the year

2000 for these innovations to become common in educational settings if they were already applied forty years ago? In light of the committee members’ remarks, the student concluded that educational practices in the year 2000 would probably differ little from those employed in 1970. Risley

(1969)

made

a similar

observation.

“Procedures

which

we

are now using and empirically demonstrating to be effective were used over a century ago, but have been discarded in the interim.” Risley supports his statement by comparing the following: 1. Rousseau’s hypothetical account of rearing Emile with the work 1

2

Introduction

of Hart, Allen, Buell, Harris, Risley, and Mees in 1964;

and

Wolf

in

1964

and

Wolf,

2. Itard’s procedures for teaching the Wild Boy of Aveyron with the work of Lovaas in 1966 and 1967;

3. Seguin’s school procedures for the education of retarded children with that of Sherman in 1965, Baer, Peterson, and Sherman in 1967, and Meyerson, Kerr, and Michael in 1967.

Many other examples in the literature show how procedures for educating children have changed very little from generation to generation. Of

course,

we

have

introduced

audiovisual

instrumentation,

better-

quality printing, modern facilities, and so forth, but the actual instructional interaction between teacher and student has changed little since the time of Socrates. The history of education has not demonstrated a cumulative development in educational approaches. Probably this is a result of basing many educational changes on historical accidents, untested theories, and the opinions of influential individuals. Yet one might wonder, surely some educational change has been based upon sound experimentation and empirical evaluation? Just look at the number of yearly M.A. and Ph.D. theses written in colleges of education that end with the statement: The implications for the classroom teacher are—one, two, three, and four. Has this had no effect on changing our educational endeavors? Look at the number of journals of education reporting research. With the abundance

of research

in education,

has there not been

a cumulative

development of instructional approaches? Why have we predominately used historical accidents,

untested

theories,

and opinions

as bases for

educational change? To find answers to these questions, we must look to the statistical models used in educational measurement and evaluation. The measurement and evaluation tactics used most in education are concerned with groups of students and how one group compares with another or with the relative standing of a student in a particular group. This concern with group analysis is a paradox because a basic tenet of American public education has been the importance of the individual student. Yet most of our measurement and evaluation procedures are designed for groups of students. This approach obscures the unique qualities of the individual and provides the teacher with no relevant instructional information. As Kunzelmann stated, “Useful information to educators is that which demonstrates differences within an individual. Building toward an educational system which is sensitive to individual differences allows for the isolation of environmental influences or instructional opinions which contribute to and thus change classroom performance”

Introduction

3

(p. 17). However, the reader should not infer that group measurement and analysis procedures are not relevant to certain school endeavors. They allow school officials to draw general conclusions, to summarize data in convenient form, to make predictions on how effective a procedure will be for a given group of students, to anaylze the effectiveness of educational institutions, and to provide sources of information for public relations. All of the above outcomes are important for administrative decision making, yet they provide no relevant individual instructional information for teachers. This book addresses the subject of a technology of teaching individual students. It is not new procedures or approaches to education that are setting the occasion for this development but rather tactics for measurement and analysis of individual behavior. Designs for the scientific analysis of individual students have only recently been developed. Mussen, Conger, and Kagen

(1963)

concurred by stating that we have

developed methods for scientifically analyzing group data but have not yet generated tactics for analysis of individual cases. Specifically, the procedures presented in this book will provide the classroom teacher with tools to answer questions such as the following: Was I effective in teaching John two-column addition? Would he have acquired this skill without my intervention? Could another teaching procedure be more effective? What materials set the occasion for increased rates of correct student performance? What effect do different seating arrangements have on academic performance? Is role playing producing changes in the social behavior of my students? The emphasis throughout this text is on the application of measurement and analysis tactics in classroom situations. First, a description of the tactic is presented. Second, an actual applied example of a classroom application is given. Finally, a discussion of the considerations of application follows the applied example. This form of organization is used in the hope that information as to how a measurement or analysis technique is used by teachers will enable the reader to know the purpose of the technique and the circumstances in which it should be used. All techniques presented in this book are described with narrative examples and, whenever possible, with graphic illustrations. Student evaluation exercises are presented in the body or at the end of each chapter. All evaluation exercises should be answered correctly before progressing further in the text. Suggested applied assignments for practice in using the techniques follow the evaluation exercises. In most

4

Introduction

cases, the reader’s knowledge and skill in using the tactics under discussion will be increased if the applied assignments are performed. A final applied assignment is given at the end of the book which allows the reader to integrate most of the applications that have been presented. The reader should plan to spend approximately eight weeks to complete this assignment. It should be noted that in several places in chapter 2 the reader is referred to appendix A. This appendix contains a description of instrumentation used in observation recording. Appendix A is a crucial part of chapter 2 and should not be slighted.

References Baer, D. M., Peterson, R. F.,

& Sherman, J. A. The development of imitation

by reinforcing behavioral similarity to a model. Journal of the Experi-

mental Analysis of Behavior 1967, 10, 405-16. Hart, B. M. Allen, K. E., Buell, J. S., Harris, F. R., & Wolf,

M.

M. Effects

of social reinforcement on operant crying. Journal of Experimental Child Psychology 1964, 1, 145-53. Itard, J.-M. G. The wild boy of Aveyron. New York: Appleton-CenturyCrofts, 1962.

Kunzelmann, H. P. Precision 1970. Lovaas,

O. I., Bererich,

Lovaas,

O.

teaching.

J. P., Perloff,

Seattle:

Special Child

B. F., & Schaeffer,

imitative speech by schizophrenic children. Science I., Freitas,

L., Nelson,

K.,

& Whalen,

C. The

Publications,

B. Acquisition

of

establishment

of

1966, 151, 705-7.

imitation and its use for the development of complex behavior in schizophrenic children. Behavior Research and Therapy, 1967, 5, 17181. Meyerson, L., Kerr, N., Michael, J. Behavior modification in rehabilitation. In S. W. Bijou & D. M. Baer (Eds.), Child development: Readings in experimental analysis. New York: Appleton-Century-Crofts, 1967. Risley, T. R. Behavior

modification: An

experimental-therapeutic

endeavor.

Rousseau, J. J. Emile. Seguin, E. Traitement

London: J. M. Dent and Sons, Ltd., 1948. morale, hygiene et education des idiots. Paris:

Paper prepared for Banff International Conference on Behavior Modification, April 1969. Battiere, 1846.

Sherman,

J. A.

The

use

of reinforcement

and

imitation

to reinstate

J. B. verbal

behavior in mute psychotics. Journal of Abnormal Psychology, 1965, 70, 155-64. Wolf, M. M., Risley, T., & Mees, H. Application of operant conditioning procedures to the behavior problems of an autistic child. Behavior Research and Therapy, 1964, 1, 305-12.

Part

1

Measurement

Until you’ve measured it you don’t know what you're talking about. —Lord Kelvin

Parents, school officials, and state departments of education are requiring classroom teachers to demonstrate the effectiveness of their instruction. In order for teachers to be accountable, it is necessary for them to possess evaluation skills. Evaluation is based on measurement, which is the way to determine student growth. The following case about Miss Lucas and Gary illustrates one way measurement can be used in the classroom. When Gary entered first grade, his teacher, Miss Lucas, observed that his class assignments were highly accurate but almost always incomplete or late. His peers apparently had no problem in meeting most task requirements on time. Miss Lucas recognized that completing classroom work on time was very important for Gary’s future academic success. Therefore, Miss Lucas wanted an objective, quantitative record of the number of assignments completed on time, since subjective impressions of student work may not be accurate. She knew that without reliable measurement she might subjectively feel that Gary had improved even though his performance remained unchanged. So Miss Lucas began daily data collection by tallying the number of assignments made and the number of assignments Gary completed on time. At the end of one week of measurements, objective statements concerning Gary’s work could be made. Miss Lucas gave twenty assignments, out of which Gary completed eleven on time. His performance was not as bad as Miss Lucas had anticipated, but it was obviously a problem behavior. With this baseline information, Miss Lucas planned an educational strategy to help Gary complete his class assignments on time and had data to make objective comparisons concerning Gary’s improvement during the year. By measuring specific student performance at certain points in time, we can determine which academic and social responses have been learned and which behaviors still need to be learned. Also,

through measurement tactics it is possible to analyze the effectiveness of our teaching. Two ways of measuring student performance are (1) obtaining and examining permanent products and (2)

observational recording. Chapter 1 presents techniques for measuring permanent products. This chapter includes a discussion of frequency, rate, and percentage measures, trials to criterion, complexity of task items, pre- and

posttests, and probes. Chapter 2 presents tactics for observational recording, event recording, duration recording, interval recording and momentary time sampling.

Direct Measurement of Permanent Products

Direct measurement of mon measurement tactic of measurement has a examination or written permanent

products.

permanent products is perhaps the most comused by classroom teachers today. This method long history. When teachers grade a written responses in a workbook, they are measuring

Certain types of behaviors

result in products that

can be measured following student responses. Other examples of perma-

nent

products

include

audio-

or

videotapes

of

student

performance,

written arithmetic computation, written spelling words, written alphabet

letters, colorings, completed puzzles, strung beads, and stacked blocks. Figure 1-1 shows a classroom example of a permanent product. The

permanent products in figure 1-1 are the written responses to computational problems. Teachers usually translate permanent products into numerical terms of (1) frequency of occurrence (John worked six of his problems correctly),

(2)

rate of occurrence

(3) percentage correctly).

of occurrence

(John read

(John

100 words per minute), or

worked

90%

of his problems

Direct Measurement

of Permanent Products

9

Mixed Addition and Subtraction Combinations. Add: 9

6

4

5

6

9

5

8

4

12

%

5

3

7

H^

4

Gl

4

9

6

3

5

ı

7

6

8

3

9

7

3

9

IC

6/

9

31

5|

4

11

2

HY

3

3

10

12 5

3$

5

7

6

4

6

8

6 3

6 8

9 8

7 8

3 S

q

WE]

71

5]

13 4

9 8

13 8

5 4

9

14

05

4

15

7

1

14 9

10 9

5 l

0 0

10 5

4

O

5

13

2

6

7

9

12.

7

If

4]

2

5

7S

6

4

8

7

3]

9

7]

6

9 6

7 6

5 7

3

5l

2l

[2.

14 7

12 8

8 1

11 5

T

9

9

E

Subtract

/ 9

Q

9

0

2.

6

pl

6o

07

7 3

2.

62. Date May.

Begin

1-1.

11

4

4

35

17 8

7 3

4 E

7

16

14

Ye

4

2

10

2Oo

1q

i

End

Lesson

FIGURE

12

Lesson

Classroom Example of a Permanent Product

Most teachers and prospective teachers are familiar with frequency,

rate,

and

percentage

measures.

Yet many

times teachers may

not use

those measures to the best instructional advantage. Chapter 1 looks at

10

Chapter 1

1. How to use frequency, rate, and percentage measures in classroom situations 2. Which parameters should be used in selecting a measurement tactic for permanent products

Frequency of Response Frequency of response is the number of times a specific behavior occurs

in some when

period of time. Tallying, or counting,

responses

is possible

the behavior is readily observable and when it is discrete; that is,

when it can be separated from other responses. Applied Example McKenzie,

Egner,

Knight,

Perelman,

Schneider,

and

Garvin

(1970)

employed frequency counts to measure correct addition problems in a first-grade class. Each school day, 20 students were assigned worksheets containing 50 addition problems of two single digits. The same 50 problems were used daily. A bell timer was set for 2 minutes for each work session. Children were instructed to begin computation on cue and to stop at the sound of the bell. Figure 1-2 shows one student’s frequency of correct responses during 8 days of baseline. Note that during baseline the frequency of correct addition responses ranged from 18 to 32. Considerations

The McKenzie study provides an excellent example of the appropriate use of frequency measures. The number of addition problems (opportunity for response) was always 50 per day. Time was constant at 2 minutes per session. With these two variables controlled, the teacher was able

to compare

student

performance

over

time

(e.g., sessions,

days,

weeks). For example, the frequency of the student’s correct addition responses (figure 1-2) was approximately 18, 18, 31, 29, 29, 30, 32 and 27 for 8 consecutive days. These data demonstrated no improvement in addition skills during the last 7 days. Frequency of occurrence should be used as a measure of behavior only when opportunity for response and time are constant. If frequency measures are utilized without reference to opportunity for response and time variables, student performance is left to subjective interpretation. The problem of using frequency measures without controlling opportunity for response and time is illustrated in the following case study.

Direct Measurement of Permanent Products

11

50

2 =

a

40r

®

e

ES9

Ir

Q

$

[: = Z

20 =

10 F E

| 5

10 Days

FIGURE 1-2. Frequencies of Correct Addition Responses by One Student in Daily 2-Minute Periods During 8 Days of Baseline. (Adapted from "Training Consulting Teachers to Assist Elementary Teachers in the Management and Education of Handicapped Children" by H. S. McKenzie et al. Exceptional Children, 1970, 37, 141. Copyright by the Council for Exceptional Children.) Miss Johnson was a teacher of primary-age learning disability chil-

dren.

John,

one

of her students,

tended

to reverse the letters b, d, p,

and q. Miss Johnson was interested in knowing if her instruction was improving John's reversal problem. Therefore, on Monday, Wednesday, and Friday of each week she made frequency counts of the number of reversed b, d, p, and q letters occurring in John's written assignments. At the end of 3 weeks, Miss Johnson had recorded the occurrences of

reversals shown in table 1-1.

From the data of table 1—1, could Miss Johnson conclude that John's

reversal problem had improved? He obviously was emitting fewer reversed b, d, p, and q letters during each consecutive week. However, from the recorded data, she could not know if the problem behavior had improved because she did not take into consideration John's opportunities to make b, d, p, or q letters. For example, on Wednesday, week one, John emitted 20 reversals. Yet he may have had 100 opportunities to write b, d, p, and q. On

Wednesday,

week two, John made

only 2

reversed letters. But perhaps there were only 3 possible opportunities

12

Chapter 1

for reversals. Were the 2 reversals in week two an improvement over the 20 reversals in week one? In order to answer this question, we must look at rate of response. TABLE

1-1.

Number

of Reversed Letters

Days

Weeks 1

2

3

Monday Wednesday Friday

15 20 9

9 2 10

5 7 6

TOTAL

44

21

18

Rate of Response Rate of response is the frequency of occurrence of a behavior during a certain unit of time. Rate is calculated by dividing the total number of correct or error responses by the amount of time spent producing those frequency responses (rate = ). Rate is usually expressed in responses time

per minute. Note that frequency is represented by the number of responses while time is the amount of minutes in which all of the responses occurred. Applied Example Hopkins, Schutte, and Garton

(1971)

reported rate measures for print-

ing and writing responses of 14 first-grade and 10 second-grade students. The first-grade students copied printed assignments that were typically composed of phonetic drills of descriptions of current events. These first-grade assignments averaged 194 letters. Second-grade children used cursive writing to copy excerpts from stories or poems such as Hiawatha. The average length of the second-grade assignments was 259 letters. To derive rate of occurrence, Hopkins

et al. used measures of dura-

tion and frequency of occurrence. Duration was defined as the elapsed time from the teacher’s instruction to begin copying the assignment until the student brought his paper to her desk. Frequency was the actual number of letters printed or written by the students. Figure 1-3 presents mean baseline rates of occurrence of printed letters per minute printed by first-grade children. Each data point represents the mean, averaged over all children for that day. The horizontal

Direct Measurement

of Permanent

Products

13

dashed line represents the average of the daily means over all days. Note in this figure that the first-grade children printed, on the average, approximately 6 letters per minute.

13

Pr

12

=

=

3

11

=

A

10

=

©

= ©

[ -

E

€

Sz

9r 8

=

c

7

Fr

=

6

L

z

o

v"

LN

WN

an Days

FIGURE 1-3. Mean Number of Letters Printed per Minute by a Firstgrade Group. (Adapted from “The Effects of Access to a Playroom on the Rate and Quality of Printing and Writing of First- and Second-grade

Students” by B. L. Hopkins, R. C. Schutte, and K. L. Garton, Journal of Applied Behavioral Analysis, 1971, 4(2), 81. Copyright 1971 by the Society for Experimental Analysis of Behavior, Inc.)

Figure 1—4 gives the mean baseline rates of occurrence of cursive letters per minute written by second-grade children. In figure 1-4, the second-grade children wrote, on the average, approximately seven letters per minute. Again, each data point represents the mean, averaged

over all children for that day.

Considerations Rate of occurrence is considered the basic datum of a science of education (Skinner, 1966). Rate is a preferred measure when evaluating

permanent products such as written material. The amount of teacher

14

Chapter 1

time needed in monitoring rate measures can be a problem for classroom teachers. One way to reduce monitoring time is to have students time their own behavior by asking them to note the time they begin and end a particular task. Instructions for recording may be printed on assigned material or given verbally. For example: Time started—10:00; Time stopped—10:25; Total time—25. The teacher or the students can

calculate

acquired

draw

two

rate

of

response.

Occasionally,

skills in “telling time.” When clocks

without

hands

on

the

students

this occurs, students’

will

not

have

the teacher can

materials.

(Some

teachers have a rubber stamp of a clock face without hands for this 18

|

Mean Number of Letters Per Minute

17 =

Days

FIGURE 1—4. Mean Number of Letters Written per Minute by a Second-grade Group. (Adapted from “The Effects of Access to a Playroom on the Rate and Quality of Printing and Writing of Firstand Second-grade Students" by B. L. Hopkins, R. C. Schutte, and K. L. Garton, Journal of Applied Behavioral Analysis, 1971, 4(2), 83. LE So 1971 by the Society for Experimental Analysis of Behavior , nc.

Direct Measurement of Permanent Products

15

purpose. ) Students could then be taught to copy the position of the big and little hands on the classroom clock when they start and when they stop. For example: 1. Worksheet before lesson

Start

Stop

Rate measures are sentive to efficiency of student performances. As

Kunzelmann (1970) states, "The student who worked 10 minutes and got 80 problems correct with a correct rate of 8 problems per minute

on the 100-item test is much more proficient than the student who took 20 minutes, also got 80 problems correct, but who had a correct rate of 4 problems a minute” (p. 31). (This statement assumes that prob-

lems

were

the

same

or of similar

difficulty.)

Basically,

then,

rate

is

important if data concerning level of student performance is to be meaningful. Also, of all measures of permanent products, rate is most sensitive to the effects of teaching tactics on student responses because it will reveal very small increments of behavior change.

Remember, Miss Johnson asked the question, “Were the 2 letter reversals in week two an improvement over the 20 reversals in week

one?” Miss Johnson cannot answer this question because time and opportunity for response were not recorded. Let’s suppose that Miss Johnson

took

into consideration

(1)

the opportunity for writing letters b,

d, p, and q; (2) the amount of time John spent on the written assign-

16

Chapter 1

ment; and

(3)

the number of nonreversed b, d, p, and q letters as well

as the number of reversed letters. With these three types of information, Miss Johnson can easily demonstrate whether John’s reversal problem did or did not change. Concerning opportunity for response, John’s written assignment was comparable in number of words on the 9 measurement days reported earlier. The

assignments number

smallest number

was

30 and

of b, d, p, and q

the maximum

of b, d, p, and q letters was

was

letters occurring in the

41. The

33. Therefore,

9-day

average

there was only a

small variance in the opportunity for John to write his letters. At no time did he reach the limit on opportunity for response. John’s reversed and nonreversed letters and time worked on the assignment are presented in table 1-2. TABLE

1-2.

Frequency of Reversed and Nonreversed Letters

Days

Reversed Week

Monday Wednesday Friday

Monday Wednesday Friday

5 7 6

2 1 1

3 min. 5 min. 2 min.

5 9 15

2 min. 2 min. 5 min.

15 20 23

5 min. 6 min. 4 min.

2

9 2 10 Week

Time

1

15 20 9 Week

Monday Wednesday Friday

Nonreversed

3

Since the amount of time that John worked on his assignment varied from day to day, Miss Johnson converted John’s frequency of letter writing to rate of letter writing per minute. Miss Johnson was correct in converting to rate of response. If she had not, a day-by-day comparison of John’s letter responses could not be made. By analogy, a high school English teacher might say that last month Bill read at a rate of 250 words per minute, but this month he is up to a rate of 400 words per minute. Bill’s reading performances could not be compared unless the teacher had converted words read to words read per unit of time.

Direct Measurement of Permanent Products

17

Miss Johnson converted to rate of response by dividing the frequency of response by time. For example on Monday, week one, John reversed 15 letters in 3 minutes. This is a rate of 5.0 reversed letters per minute (15/3

= 5). The number

of John’s nonreversed

on Monday, week one, was 2.

b, d, p, and q

letters

This is a rate of 0.7 nonreversed letters

per minute (2/3 — 0.67 — 0.7). John's rates per minute of reversed and nonreversed b, d, p, and q letters are shown in table 1-3. TABLE

1—3.

Rate of Reversed

Day

Reversed Week

Monday Tuesday Wednesday

Monday Tuesday Wednesday

Nonreversed

0.7* 0.2* 0.5

2 5.0 1.0 2.0

Week

Letters

1 5.0 4.0 5.0

Week

Monday Tuesday Wednesday

and Nonreversed

3.0 5.0 3.0

3 1.0 1.0 2.0

3.0 3.0 6.0

Note. The rate of 0.7 (7/10 of a response per minute) could be read as 7 nonreversed responses in 10 minutes. Similarly a rate of 0.2 (2/10 of a response per minute) would be the same as 2 responses in 10 minutes.

Miss Johnson can now state with some assurance that John’s reversal problem is improving. The rate of nonreversed b, d, p, and q letters is

accelerating and the rate of reversed letters is decelerating in occurrence.

Percentage of Response Percentage of response is a ratio that expresses the amount of a behavior as a certain number of responses per every 100 responses. Percentage is usually obtained by dividing the total opportunities for response (e.g., 25 test items) into the number of correct or incorrect responses (e.g., 17 correct answers) and multiplying that result by 100 (17/25 = .68 X 100 = 68%).

18

Chapter I

Applied Example Hopkins et al. (1971) were concerned with the relationships between correct and incorrect letter responses. Each student’s letter responses were scored by drawing circles around incorrect responses. Criteria for incorrect responses were as follows: All omissions of assigned letters Substitutions of letters in place of those assigned Reversals, e.g., printing or writing À instead of K Omitting any part of a letter, e.g., failing to cross a t A failure of any vertical stroke to be within 15 degrees of the perpendicular to the baseline (printing) or failure of written letters to be consistently slanted about 60 degrees from the baseline 6. Short lowercase letters being less than half as tall as the distance between the baseline and the centerline or being more than 50% taller than the distance between the baseline and the centerline 7. Tall lowercase or uppercase letters being less than 75% as tall as the distance between the baseline and top line or being more than 25% taller than this distance

Un

RU.

. . . . .

8. Lowercase

letters, which

should extend below

the baseline, e.g.,

y, failing to extend below the baseline at least 50% of the distance between the base- and centerlines or extending further than 100% of this distance (p. 79)

The total number of errors was tallied and written at the top of each student’s copy paper. Hopkins et al. reported the mean proportion of errors each day. This was computed by dividing the total number of errors by the total number of letter responses. Percentage was then obtained by multiplying the quotient by 100. Figures 1-5 and 1-6 show the percentage of errors during baseline for the first and second grades. Considerations

Ideally, if the total number of responses is less than 100, percentages should not be computed (Guilford, 1965). In the Hopkins et al. (1971) study, the first-grade class averaged 194 letter responses; the secondgrade averaged 259 responses. Guilford states that if a lower limit must be set, it is unwise to compute percentages when the divisor is less than 20. The addition of only one frequency in the numerator will produce a corresponding change of at least 5 percent when the divisor is less than 20. For example, suppose that a student answered 11 questions

Direct Measurement of Permanent Products

19

correctly on a 13-item test, he would have scored 86 percent correct (11/13 X 100 = 86% ). However, had the student answered one addi-

tional

(12/13

item

correctly,

his

X 100 = 92%),

percentage

would

have

been

an increase of 6 percent! When

92

percent

this situation

occurs, it is recommended that a ratio of total correct and total opportunities (12/13) be used, rather than percentage. 22.5

Percentage of Errors

Baseline 20.5

j=

18.5

pP

16.5

»

14.5 12.5

>»

10.5

m

Nr

6.5

Days

FIGURE 1-5. Percentage of Errors in Printing by a First-grade Group. (Adapted from “The Effects of Access to a Playroom on the Rate and Quality of Printing and Writing of First- and Second-grade Students” by B. L. Hopkins,

R. C. Schutte, and K. L. Garton, Journal

of Applied Behavioral Analysis,

1971, 4(2), 81. Copyright

the Society for Experimental Analysis of Behavior, Inc.)

1971 by

One other caution should be noted in using percentages. Frequently teachers incorrectly report a mean or average percentage that was calculated from other percent scores. For example, suppose Jill’s percentage

scores were

as follows:

72%,

88%,

80%,

100%,

and 90%.

It would be incorrect to figure Jill’s average percentage correct score for the five tests as follows:

20

Chapter 1

+

72% 88% 80% 100% 90% 430%

+ 5 = 86%

To calculate the correct mean percentage for the five tests, we must use frequency scores. If the total of possible correct responses for Jill’s

five tests was

150, and she responded

correctly to 132 test items, her

mean or average percentage correct would be 88%. number of correct responses

132

= ——= 150

total possible correct

0.88 x 100 = 88%

average

I

ww

Percentage of Errors

Baseline

LL

2

L

1 4

f

tf 6

8

Days

FIGURE 1-6. Mean Percentage of Errors in Writing by a Secondgrade Group. (Adapted from “The Effects of Access to a Playroom on the Rate and Quality of Printing and Writing of First- and Second-grade

Students” by B. L. Hopkins, R. C. Schutte, and K. L. Garton, Journal of Applied Behavioral Analysis, 1971, 4(2), 83. Copyright 1971 by

the Society for Experimental Analysis of Behavior, Inc. )

Direct Measurement of Permanent Products

21

There are two advantages to using percentages. First, the use of percentages is an efficient way to deal with large numbers of responses, and second, most people easily understand percentages. But there are also disadvantages. First, percentages do not say anything about efficiency; percentage may increase while error rate remains the same or increases. Second, a percentage figure may not be as easy to understand as a simple tally of correct or error responses.

Variations on Frequency, Rate, and Percentage Measures Techniques selected for measuring academic performance must be relevant to the purpose of collecting data. For example, student responses to items on a formal achievement test may not provide sufficient information for planning individual programs of instruction. No matter what technique is selected, measurement must provide data on student responses to the actual materials used during the instructional process, and it must be continual; that is, frequent samples of performance must be obtained. The basic procedures for direct and daily measurement which have been described are frequency, rate, and percentage. These basic procedures can be applied in a variety of ways. Four widely used variations have been described by Cooper and Johnson (1979): (1) trials to criterion, (2) complexity of task items, (3) pre- and posttests, and (4) probes. Trials to Criterion

Trials to criterion measures the number of times response opportunities are presented before the student achieves a preestablished level of competence or. proficiency. For example, the measurement could be reported as 10 trials to achieve 100% accuracy (percentage) or 10 trials to achieve 60 words per minute (rate). In all cases the criterion set will probably require a measure of proficiency during each trial. Measuring trials to criterion is appropriate when making an evaluative comparison of two or more kinds of materials or instructional methods. For example, a teacher might report the number of times the word that was presented orally before the student read it independently and compare that number to the number of times the word that was presented in written form to the student for tracing before he read it independently.

22

Chapter 1

Applied Example

A common technique used in developing math skills is to have students say or write answers to a set number of problems until a specified rate correct has been demonstrated. For example, each day, the students give the answers to 50 multiplication problems. The teacher’s criterion is for the students to complete the 50 multiplication facts in one minute at 100% accuracy. The student responses in this illustration can be measured appropriately as number of trials or opportunities needed to meet the criterion of answering 50 multiplication problems in one minute at 100% accuracy. For instance, the teacher could report that Mary reached criterion in 11 trials and that Bill required 20 trials to reach criterion. Considerations

Measuring trials to criterion can give a teacher useful information for planning the amount of time that should be allotted for instruction. If trials to criterion are counted for several words taught to a student in different ways, the teacher can compare performance to determine if the student masters words better with one of the teaching methods than she does with another.

Complexity of Task Items The response units suggested for previously discussed measures have been whole units (e.g., whole words read or spelled, finished mathematics problems). Measuring responses in whole units, however, does not take into account the complexity of each possible response. A method for measuring academic responses which would take complexity into account would be to count the operations necessary to achieve a correct response. These operations may be reported as frequency, percentage, or rate. For example, in measuring spelling performances,

rather than count

a complete

word

as correct or incorrect

the teacher might want to consider the number of letters that are correct within each word. Applied Example

Helwig (1973) reported mathematics response rate calculated from number of operations sufficient to produce the answer. The student

Direct Measurement of Permanent Products

23

was given 20 multiplication and division problems to work at each session selected at random from a set of 120 problems. The length of time for each session was recorded by the teacher. The 120 problems were of the type a X b = c anda + b = c where the student had to find a, b, or c. The student was asked to find one of the factors—the product, dividend, divisor, or quotient. The problems required from one to five operations to find the answer. For example, to find the answer to the problem 55 X 5 = ? requires the student to perform four operations: First operation 55 __3 (Multiply the ones: 25 5X5 =25) Third operation 55 __5 (Multiply the tens: 255 5 x 50 = 250)

Second operation 2 55 __5 (Carry the two tens from 5 the first operation) Fourth operation 55 __5 (Add the two tens from 275 the second operation)

A correct answer would be scored as four correct responses. When there was more than one possible way to find the answer, a mean number of operations was figured for that problem. (For example, in the problem 4 X c — 164, the answer by multiplication requires two operations, by division four. The mean number of operations would be three.) Each set of 20 problems was checked and the number of correct responses and errors was recorded and reported as rate of response. Pre- and Posttests One tactic used by teachers to measure students' acquisition of academic skills is pre- and posttesting. The pretest is used primarily to provide an assessment of what the student can do before instruction. The posttest is given following instruction. Pre- and posttest measurement can be used in two ways. First, the teacher can report the student's posttest score as a final grade. Second, the teacher can report the student's score as instructional gain from pretesting to posttesting. To illustrate, say that a student scored two correct out of 20 on the pretest, and 10 correct out of 20 on the posttest. Therefore, the teacher would report a gain score or 4096 (8 + 20 = .40, or 40%).

(increase)

of 8,

24

Chapter 1

Applied Example

Johnson and Bailey (1974) employed pre-, mid-, and posttesting in conjunction with direct and continuous measurement of basic arithmetic skills of kindergarten children. A 122-item skill-based arithmetic test was constructed to be used as a pre- and posttest measure and also as a midtest at two-week intervals. The following procedure was employed: Approximately 3 minutes at the end of each session were used for “review

test”;

ie.

the

tutor

simply

went

over

all

of

the

tasks

worked on that day and recorded the student’s responses. The tutor then scored the tests and graphed the data, giving the student immediate feedback to his daily progress. When a student received a perfect score, on any task for three consecutive review tests, teaching of the task was discontinued and a new task was introduced. (p. 226) Considerations

A major advantage of pretesting is that it provides the teacher with a rough estimate of student skills in the instructional area. The pretest allows the teacher to modify his or her teaching topics to cover only skills that the student has not previously acquired. Also using of preand posttests requires very little of the teacher’s time. An obvious limitation of pre- and posttests is that both measures must be applied only to items that represent exactly the same unit of study and that do not represent more complex task requirements. Because only two measures are compared, the teacher must be sure that the pre- and posttests are equivalent in requiring the same kind, complexity, and manner of response. This limitation ceases to be a problem if the teacher combines pre- and posttest measures with some other form of continuous measurement. Probes An important question of instructional effectiveness is whether the teaching of certain specific items has a generalized effect and improves students’ accuracy on similar untaught items. For example, if the teacher specifically teaches some words with a long vowel-consonantsilent e pattern, will the student correctly read other untaught words which follow the same pattern? To answer this question the teacher could periodically present untaught words, or probes, as well as the words being taught, and then record correct or incorrect responses to probes. Probe items must be related in form or function to items taught.

Direct Measurement of Permanent Products

25

Applied Example Clark,

Boyd,

and

Macrae

(1975)

were

concerned

with

teaching

six

disadvantaged youths to complete biographic information on job application forms. After training, students were tested on the three application forms on which they received training. Students were also tested on a fourth application form which had not been used in training (a probe form). Test results were reported in percentage of correct responses on all four application forms. The authors stated, “The tests show that after an item had been taught, it was correctly used in completing application forms on which the youths had been trained and forms on which they had never been trained” (p. 67). Considerations

The advantages of probe measures are similar to the advantages of pretests. They provide a rough assessment of student skills and require very little teacher time. Also, probes tell the teacher whether instruction on selected items has affected the student’s accuracy on similar items that were not taught. If accuracy on untaught responses increases, the teacher can say that the student has mastered the concept and that additional instruction is unnecessary. However, if accuracy on untaught items does not increase, the probe measures can be used as measures of baseline skill prior to teaching those items. Table 1—4 provides a summary of the measurement techniques discussed in chapter 1.

Evaluation : You may wish to evaluate how well you comprehend measurement of

permanent products by responding to the following test items. Cover the

answer column with a sheet of paper; then write your answers in the blank spaces provided. The correct answers can then be checked. written arithmetic computation coloring written spelling words strung beads after (subsequent to)

1. List four examples of permanent products. a. b. c. d. 2. Permanent products can be measured the

student's

behavior.

26

Chapter 1

TABLE

1-4.

Measurement Techniques for Permanent Products

Definition

Application

Data

Evaluation

FREQUENCY Simple count of correct or error responses OCcurring in some period of time

1.

2. 3. 4. 5.

Easily understood when communicating with parents and children Simple procedure Generally does not interfere with ongoing teaching Requires minimum amount of teacher time When used with permanent products, source Of errors may be analyzed for instructional planning

1.

2.

For day-to-day comparisons of data, total number of opportunities for response should not vary with assignments Reflects academic efficiency

RATE Frequency of a behavior during a certain unit of time. To compute, divide the total number of correct error responses by the amount of time spent working on the material.

1.

4. 5.

Most sensitive measure of academic behavior, i.e., most likely to show small changes Excellent as criterion for mastery of skills Some parents and children may have difficultly understanding rate Approximate for measuring all academic behaviors May require more teacher time to measure and compute

1. 2.

3.

4.

Reflects academic efficiency Measures are not comparable if difference in length of sessions is extreme; e.g., rate computed from a 1-minute sample may be higher than that computed from a 30-minute sample for some students and some academic responses Student (not teachers or materials) should control maximum possible rate of response Accurately reflects student efficiency if complexity of tasks does not change between measures

PERCENTAGE =

Amount of behavior expressed as a certain number of responses in every hundred. To compute, divide the number of correct

or

error

Te-

sponses by total number of opportunities for responses, then multiply by

Easily understood when communicating with parents and children

Student performance at sessions offering unequal response opportunities can be compared

Direct Measurement of Permanent Products TRIALS Number of times response Opportunities are presented before student achieves a preestablished criterion

1.

TO

Measurement of the number of operations sufficient for a correct response. When there is more than one algorithm for an answer, a mean number of operations can be reported

before tion

and

of behavior

after instruc-

OF

TASK

1.

AND

Can show efficiency if criterion is stated as rate In all cases the criterion set will probably require a measure for each trial; however, only the number of trials to achieve the criterion is usually reported.

ITEMS

Use when complexity of response varies greatly within one measurement episode or when a fine analysis of response accuracy is desired PRE-

Measurement

CRITERION

Easily understood when communicating with parents and children An appropriate measure for making an evaluative comparison of two or more kinds of materials or instructional methods May give teachers useful information for planning time to be given to instruction

COMPLEXITY

27

Number of operations may be used to compute frequency, percentage,

or

rate

POSTTESTS

Provides a minimal estimate of student skills in an instructional area Requires a minimum of teacher time Pretest used to provide an assessment of what student can do before instruction Teacher may report posttest score as final

Measurement at intervals during instruction is critical if time between preand posttests is weeks or months

grade

Teacher may report instructional gain from pre- to posttesting PROBES Infrequent measurement of responses that have no current instruction Or consequation

Use to assess whether the teaching of specific items has a generalized effect on similar untaught items Requires minimum of teacher time rough Provides a estimate of student skills Can be used to measure baseline skill prior to instruction

Probe

related

items

function taught

in

must

to

form

be

or

items

28

Chapter 1 . Responses

frequency

that generate

permanent

prod-

ucts can be translated into numerical terms

rate

percent

of

;

, or

Frequency

refers to the number of times a specific behavior occurred in some period of time.

opportunity for response; time

. Frequency of occurrence should be used as a measure of behavior only when and

subjective interpretation

. If frequency measures are utilized without reference to opportunity for response and

are

time variables,

constant.

student performance

is left

to

Rate

quency time.

of

frequency time

. Rate

frequency time

, Rate =

is defined as the freoccurrence during a unit of

is calculated

by dividing the

by

minute

10. Rate is usually expressed in responses per

rate of occurrence

11. The basic datum of a science of education

Percentage

12.

20

13, It is unwise to compute percentages when the divisor is less than

Trials to criterion

14.

operations

15. A method for measuring the complexity of academic responses would be to count the necessary for a correct response. 16. tell the teacher whether instruction on selected items has affected a student’s accuracy on similar items that were not taught.

is

Probes

is a ratio that reports data as a given amount in every hundred responses.

measures the number of times response opportunities are presented before the student achieves a preestablished level of competence or proficiency.

Direct Measurement Exercise

1. 2. 3.

4. 5.

29

1-1

In figure 1-1, of 12 or above In figure 1-1, sums of 12 or In

of Permanent Products

figure

1-1,

what is the frequency of reversed two-digit number sums (e.g., student wrote 21 instead of 12)? what is the frequency of nonreversed two-digit number above? what

is the

rate

per

minute

of

all

addition

and

sub-

traction errors? (Note, do not count unanswered problems as errors.) In figure 1-1, what is the rate per minute of all correct addition and subtraction answers? In figure 1-1, what percentage of two-digit number sums of 12 or above are reversed?

Answers for Exercise 1-1 1. 2. 3.

4.

5.

17 reversals 4 nonreversed sums 17 addition errors +

1 subtraction

error

= .9 errors per minute 20 minutes 15 correct addition problems + 17 correct subtractions 20 minutes = 1.6 correct answers per minute 17 reversed sums x 100 = 18% 21 two-digit sums of 12 or above

References Clark, H. B., Boyd, S. B., & Macrae, J. A. A classroom program teaching disadvantaged youths to write biographic information. Journal of Applied Behavior Analysis, 1975, 8, 67-75. Cooper, J. O., & Johnson, J. Direct and continuous measurement of academic

behavior. Directive Teacher, 1979, I, 10-11; 21. Guilford, J. P. Fundamental statistics in psychology York: McGraw-Hill, 1965.

Helwig,

J. Effects

sponse

Hopkins,

rate.

of manipulating

Unpublished

B. L., Schutte,

an antecedent

paper,

Ohio

R. C., & Garton,

and education.

event on

mathematics

State University,

1973.

New re-

K. L. Effects of access to a play-

room on the rate and quality of printing and writing of first and second-grade students. Journal of Applied Behavior Analysis, 1971, 4(2), 79, 81, 83. Johnson, M., & Bailey, J. S. Cross-age tutoring: Fifth graders as arithmetic tutors for kindergarten children. Journal of Applied Behavior Analysis, 1974, 7, 223-32.

30

Chapter 1

Kunzelman,

1970.

McKenzie,

H.

P.

Precision

H. S., Egner,

teaching.

A. N., Knight,

Seattle: M.

Special

Child

R., Perelman,

Publications,

P. F., Schneider,

B. M., & Garvin, J. S., Training consulting teachers to assist elementary teachers in the management and education of handicapped children. Exceptional Children 1970, 37, 137-43. Skinner, B. F. Operant behavior. In W. K. Honig (Ed.), Operant behavior: Areas of research and application. New York: Appleton-Century-Crofts, 1966.

Observational

Recording

Measurement of permanent products is the foundation for most classroom evaluation. Indeed this measurement tactic should always be a major source of data for teachers. Yet, as instructional technology develops

through

behavior

analysis,

many

teachers

now

see

that ob-

servational recording is an equally important measurement tactic. Classroom teachers are concerned with many socially significant behaviors that do not result in permanent products. These behaviors must be observed and recorded as they occur. When teachers look at behavior and produce records of that behavior as it occurs, they are engaging in observational recording.

By way of illustration, let us look at a problem behavior in Mr. Clamer’s fifth-grade class. Mary Jane’s behavior has been a source of disruption in Mr. Clamer’s class for the past three weeks. Whenever she is called on to participate in class discussion—asked a question or given a directive—she defiantly responds by shaking her head no, much to the amusement of her classmates. Obviously, when Mary Jane shakes her head no, she does not leave a permanent product that can be 32

Observational Recording

33

measured one week after it occurred. Observational recording must be used to measure this behavior. | At first, Mr. Clamer ignored Mary Jane’s gestures of no, hoping that they would cease. But eventually he decided that these gestures were increasing and he planned an intervention tactic. He made a daily tally of the number of occurrences of no gestures for 5 consecutive school days before introducing intervention. With these data, Mr. Clamer could determine whether Mary Jane made a no gesture with her head less often during intervention than she had before intervention. Other examples of socially significant school behaviors that require observational recording include verbal responses to teacher questions, cooperation, increases in sentence speaking, teasing, guffaw laughing, aggressive hitting, and noise levels in the lunchroom. Chapter 2 addresses the topic of observational recording. Hall (1971) outlined five major observational recording techniques for teachers:

(1) continuous recording, (2) event recording, (3) duration recording, (4) interval recording, and (5) momentary time sampling. Each of

these observational recording techniques will be presented. After studying chapter 2 you should be able to:

1. Give applied examples of each observational recording technique 2. Discuss how appropriate each technique is for the classroom teacher 3. Select appropriate tactics for observational recording

Interobserver Agreement Mr. Jones taught tenth-grade Spanish. He and Ms. Roberts, the school psychologist, were concerned with Larry’s boredom in school and his lack of enthusiasm for the subject matter of his courses. Their goal was to help Larry develop an appreciation for the subject matter presented in his classes and more positive feelings toward school. A teaching intervention was started. The intervention tactic consisted of a maximum of teacher attention contingent upon improvements in Larry’s behavior. Mr. Jones had heard of the effects of social reinforcement and was convinced that this technique would help Larry. Ms. Roberts felt that social reinforcement would not hurt anything but had reservations concerning its effectiveness in this case. During the next two weeks, Ms. Roberts attended Larry’s Spanish class and made notes on all behavioral improvements. Also, whenever Mr. Jones observed Larry showing appreciation for subject matter or more positive feelings

34

Chapter 2

toward school, he tallied the number of occurrences in a daily progress report. . At the end of two weeks the teacher and school psychologist met to discuss Larry's progress. Mr. Jones reported marked improvement. Ms. Roberts,

on the other hand,

stated

that her observations

showed

ab-

solutely no change in appreciation for subject matter or more positive feelings toward school. When Mr. Jones and Ms. Roberts realized they could not agree upon Larry's behavior, they asked the following question: What exactly was Larry doing that caused them to say that he needed to develop an appreciation for subject matter and positive feelings toward school? The most obvious problem was the fact that Larry was not completing his assignments. Mr. Jones and Ms. Roberts decided, therefore, to count the number of assignments Larry completed in one week. When they conferred the second time, Mr. Jones reported that Larry completed 12 of 13 assignments and Ms. Roberts reported 13 completed assignments. By counting the number of completed assignments, the teacher and psychologist were able to communicate what they had observed. As this illustration demonstrates, behaviors must be defined specifically. They must also be defined objectively; that is, people must agree upon the occurrence or nonoccurrence of behavior. Directive teachers can only work with observable and measurable behaviors. In the case of Larry, Mr. Jones and Ms. Roberts did not at first define an observable

and

measurable

behavior.

However,

the

second

behavior

definition (number of completed assignments) proved to be a scientific definition because they were able to report close agreement concerning what was observed. In directive teaching it is necessary to know that the teacher's observations and recordings are reliably reported. A common way to determine reliability of classroom measurement is through simultaneous measurement by independent observers. Hall (1971) defined interobserver agreement as “the degree to which independent observers agree on what they have observed in the same subject during the same observation session" (p. 6).

Interobserver agreement measures are usually reported as percentage of agreement among two or more independent observers. Percentage of agreement is calculated by dividing the number of agreements by the number of agreements plus disagreements and multiplying by 100. agreements X 100 — percentage of agreement agreements 4- disagreements

To illustrate, Mrs. Gomez was interested in decreasing the number

of times that Jack

talked out in class. She invited Mrs. Anderson,

the

Observational Recording

35

school principal, to function as an independent observer of Jack’s talkouts. Before class, Mrs. Gomez gave the principal a definition of talk-outs and told her to make a tally each time she heard Jack talk out between 9:00 and 9:15 A.M. The definition given to the principal was: “Talking out is defined as a vocalization, comment, or vocal noise initiated by the student. It cannot be in response to the teacher or another peer. Each occurrence will be tallied as one talk-out if it is preceded or followed by a breath, time interval or change of topic." At the end of the 15-minute observation period, Mrs. Anderson had tallied 10 talk-outs. The teacher recorded 12 talk-outs. Mrs. Gomez calculated the reliability of her data by dividing the number of agreements (10) by the number of agreements plus disagreements (10 + 2) and multiplying by 100. Mrs. Gomez’s calculations were: 10 — = 0.833 x 100 = 83% 12 Mrs. Gomez’s agreement measure was sufficiently high that she can accept her observation as reliable and proceed with teaching Jack appropriate classroom verbal behavior. As a rule of thumb, an agreement measure should be on the average above 80% before the teacher continues with instruction. Perhaps the most frequent cause of unsatisfactory agreement measures among observers is a lack of precision in defining behavior. Behaviors must be defined in terms of body move-

ments that result in outcomes that can be seen or heard or felt; that is,

defined behaviors should be observable and measurable. If agreement measures are below 80%, the observers should discuss why agreement was not reached concerning the occurrence or nonoccurrence of behavior and arrive at a new collective behavior definition. This process should continue until an acceptable agreement percentage is reached. However, after acceptable agreement measures have been reached with the original observers, an outside person should be asked to be an observer. The: original observers may only have reached agreement between themselves! Asking a third independent person to serve as observer is the best check a teacher can use to tell if the behaviors selected for observation are generally recognizable. An outside observer should be able to tell when Jack is or is not engaging in the behavior that the teacher wants to change. Some key points to remember concerning reliability are: 1. Other students, other teachers, teacher aides, school secretaries,

school psychologists or counselors, consultant teachers, parents, volunteers, and principals can serve as observers for the classroom teacher.

36

Chapter 2

2. Interobserver agreement measures should be made over several sessions prior to and during each phase of instruction. 3. Observers should not be informed of the teacher’s instructional strategies. “An observer may tend to err without being aware of doing so, by recording results in the anticipated direction. Thus, if one was aware that a reinforcer was being applied, he might be more likely to record an increase in behavior. If he knew that the reinforcer had been withdrawn, his tendency might be to record a decrease in the behavior” (Sulzer and Mayer, 1972, p. 269). Exercise 2-1

During your next lecture period, or assembly meeting with a speaker, arrange with another person to tally the number of occurrences of some behavior emitted by the lecturer or speaker. You could count one of many behaviors, such as hand gestures, jokes, steps taken, pushing eye glasses up, rubbing head,

number

of “uhs,”

and

so on.

Agree

upon

the behavior

to be tallied

before the lecture or assembly meeting. Tally the number of times the behavior occurs during the first 15 minutes of the meeting. Calculate your percentage of agreement. Were your data reliable? If not, what should you do?

Continuous Recording Continuous recording often takes the form of “anecdotal reports’ or “diary records.” The aim of this recording technique is to record all behavior as it occurs. No specific behaviors are pinpointed for observation to the exclusion of other behaviors. This technique produces a written narrative of individual or group behaviors for a specified time period. The conditions or situations under which the behaviors were emitted are also described. The main idea of continuous recording is to produce as complete a description as possible of student behaviors in specified settings. Wright (1960) reported directives for continuous recording: Begin in reporting each observed sequence with a description of the scene, the actors, and the ongoing action. Report throughout in everyday language. Describe the situation as fully as the behaviors. Thus include “everything the child says and does,” but include also “everything said and done to him.” . . . Describe the larger “adaptive actions” of the child, but weave in as well the “hows” of these actions as far as possible. “Nonadaptive aspects of behavior” are important on this account. . . . Do not substitute interpretations that generalize about behavior for descriptions of behaviors, but add such interpretations when they point to possi-

Observational Recording

37

bilities of fact that might otherwise be missed. Segregate every interpretation by some such device as indentation or bracketing. Straight reporting must be left to stand out. (pp. 84-85)

Wright also described several possible procedures for continuous recording: Notes on the scene of observation, which obviously are needed for sufficiently detailed and accurate description, can be kept in improvised shorthand. These field notes can be enlarged in writing immediately after each observation period, or they can serve then as a base for a dictated narration of the observed behavior sequence.

Also,

a co-worker

can

hear

this account

through

and

at

once question the observer where it is thin or unclear. The original dictation plus the interrogation and the observer’s responses can be sound

recorded.

All of the recorded

material

can then

be copied

and revised in an improved account. Observations can be timed to permit various measures of duration. Timing of the field notes at intervals of approximately one minute or even 30 seconds has been found practicable. When long records

are made,

events.

(p. 85)

observers can work

in rotation; and the time of

each observing period can be regulated to minimize effects of fatigue upon ability to see and remember an always fast train of

Regardless of the recording instructions or procedures employed, a number of common elements are found in all continuous recording techniques. 1. A time sequence is given. This sequence may be reported as large units of time (e.g., 9:00-9:50 A.M.) or as smaller multiple time samples, (e.g., 9:00-9:05, 9:05-9:10, 9:10-9:15, and so on).

2. Target behaviors are unselected; no specific behavior is pinpointed. 3. Data take the form of narrative descriptions of behaviors. These narrative descriptions should include a three-term contingency

of (a) events that occur before a behavior is emitted, i.e., antecedent stimuli, (b) the behavior of the child, and (c) the

events or stimuli that occur after the specified behavior. 4. The narrative is plainly worded and easy to read. Applied Example

Bijou, Peterson, and Ault (1968) reported an example of continuous recording with a preschool child named Timmy. The setting was a play yard.

38

Chapter 2 Timmy is playing by himself in a sandbox in a play yard in which other children are playing. A teacher stands nearby. Timmy tires of the sandbox and walks over to climb the monkey-bars. Timmy shouts at the teacher, saying, “Mrs. Simpson, watch me.” Timmy climbs to the top of the apparatus and shouts again to the teacher, “Look how high I am. I’m higher than anybody.” The teacher comments on Timmy’s climbing ability with approval. Timmy then climbs down and runs over to a tree, again demanding that the teacher

watch

him.

The

teacher,

however,

ignores

Timmy

and

walks back into the classroom. Disappointed, Timmy walks toward the sandbox instead of climbing the tree. A little girl cries out in pain as she stumbles and scrapes her knee. Timmy ignores her and continues to walk to the sandbox.

This example of continuous recording is written in a style used by reporters for a newspaper or magazine. Directive teachers, however, need to form a clear impression of temporal relationships among antecedent stimuli, responses, and consequent stimuli. The four-column form shown in figure 2-1 is helpful in delineating time relationships. DESCRIPTIVE

Setting:

OBSERVATIONAL

.'

INFORMATION

FORM

Observation date:

Student: Time

Antecedent Events | Child Responses | Consequent Events

FIGURE

2-1

The episode of Timmy's play yard behavior was transcribed by Bijou et al. into a four-column form, and each behavioral and stimulus event was consecutively numbered. The continuous record for his behavior is shown below. Setting. Timmy (T.) is playing alone in a sandbox in a play yard in which there are other children playing. T. is scooping sand into a bucket with shovel, then dumping the sand onto a pile. A teacher,

Observational Recording

39

Mrs. Simpson (S.) stands approximately six feet away but does not attend to T. Time

Antecedent Event

9:14

Response

Consequent Social Event

1. T. throws bucket and shovel into corner of sandbox. 2. ... Stands up. 3. ... walks over to monkey-bars and stops.

4. ... turns toward teacher. 5. ... Says, “Mrs. Simpson, watch me.” 6. Mrs. S. turns toward T.

6. Mrs. S. turns toward T.

7. T. climbs to top of apparatus.

8. ... looks toward teacher. 9, ... says, “Look how high I am. I’m higher than anybody.” 9:16

10. Mrs. S. says, “That’s good, Tim. You're getting quite good at that.”

10. Mrs. S. says, “That’s good, Tim. You’re getting quite good at that.”

11. T. climbs down. 12. ... runs over to tree.

13. ... says, “Watch me climb the tree, Mrs. Simpson.”

14.

9:18

Mrs. S. turns and walks toward classroom. 16. Girl nearby trips and falls, bumping knee. 17. Girl cries.

15. T. stands, looking toward Mrs. S.

14. Mrs. S. turns and walks toward classroom.

18. T. proceeds to sandbox. 19. ... picks up bucket and shovel. 20. ...resumes play with sand.

Note that a response event (e.g., 5, ... says, "Mrs. Simpson, watch me.") may be followed by a consequent social event (e.g.,

40

Chapter 2 6.

Mrs.

S.

turns

toward

Timmy.)

which

may

also

be

the

ante-

cedent event for the next response (e.g., 7. T. climbs to top of apparatus.) Note too, that the . . . form retains the temporal relationships in the narration. Note, finally, that only the child’s responses are described. Inferences about feeling, motives, and other presumed internal states are omitted. Even words like “ignores” and “disappointed” do not appear in the table. (pp. 78-79) Considerations

Directive teachers should employ continuous recordings primarily (1) as an assessment aid in pinpointing specific behaviors that need to be

learned, (2) to identify possible environmental conditions that set the occasion for student responses, and (3) to identify possible consequent

events that maintain behaviors. Perhaps the major reason for utilizing this technique is that a large range of behaviors can be observed and recorded. Continuous recording is a measurement procedure seldom employed by classroom teachers. The teacher cannot concurrently give instruction and measure behavior with continuous recording techniques. The recording technique is more appropriate for observers other than the directive teacher, e.g., teacher aides, volunteers, school psychologists, consultant teachers, principals, secretaries, students.

Some

have

reported

that continuous

observer agreement measures

(see Wright,

records

generate

good

inter-

1960, p. 86, for references).

Yet one must question high interobserver agreement (80% or better) with continuous recording. It is impossible to record everything as it occurs in time, and it is highly unlikely that two observers would report exactly the same things in the same way. For example, tables 2-1 and 2-2

present

simultaneous

observations

compiled

by two

doctoral

stu-

dents trained in observational recording. These two records do not show high interobserver agreement concerning specific behaviors. Yet, even though they differ in the specific temporal sequences, events, and responses recorded, the two reports provide similar overviews.

Evaluation

At this point, you may wish to evaluate how well you comprehend the information on observational recording by responding to the following

41 Observational Recording

punole 919119 MOIP 3 Jeym [[9} 0) puey sosreJ SLID PAD SMe1p Ápjoin()

punoJe SXoo[ SIIUT)

Xueq A331d ur sgurq; punore 9[21I? MBIP O} PJO} SXooq SAEU SYUIPNIS

UOIJESIIAAUOZ

194989 L

s0:11755:01

oun

IC TIaVL

juopnis YW ‘SUYD

S9In}51d-——11809 wos ZuimpesI dnoj8 [[jews—,,sosed um],

I[Sds noA op Moy suyJ sysy SJUSWIUSISSE [ENPIAIPUL SoAIS IOUOUAL

uo[f) ‘Aqny Jo suonsonb pose

suonsonb syse 1ougoeo[, X004 sıy Je 3urqoo[ 0) YJIeq soo sro

sıamsue S,PITU9 Jayjoue oj suojsi[—b syse louo*o[

1949589 L

ou} ÂI9A9 Spuodseal siu)

aded suin} sIIyJ

A]}001109 spuodsoJ sılyJ

sonunuoo

«QUO SIU} Jamsue U9[O 39] I[,2A,, SÁes 194989 L

os SUIOP Aq spuodsoi sri.

o1Jnjord noge mo SHIR} suu dn uou) UMOP SxXoo[ 'seopr SI9YO SIIUTD

suonsanb

osuodsoI SIq jnoqe uormsonub syse 194989 L

'I9ASUE O} JI? Ul puey YWM—yoeq 3uSr1— PUOSeS 9UO AOpUIA 93} INO SXOO[ SIIUT) spalgo Zurweu Aq spuodsoI sr

«9Ulj,,

sIaMmsue 'sjoopoel I9U989L

noA,

«BUI[IOdS

prey P9XIOM noÁ osne5 -9q S UB noÁ SAIS 0) JUEA ],,

19989) ‘99.19 Surpeoi ur SUIS

L SUIPICS9Y

UONEULIOJUJ

[EUONEAISSIO

' SjuaAg JU9P999JUV

Á?p19jsoÁ 10J FIOM INOA 99S S39T,, —BUIPU9I 0) SOWOS J9U9U9 L

107 Zumem

dng-1ou989

sosuodsow DpIuo

uornobteIloju]

"ZZ

"SUIU OP

(1518 I ‘sAoq Z) sjuopnis € pue Joyovs}—dnoid SUIPe91 [[UUIS ss:0l

QUIL

JA IHV.L

juapnys UWA ‘SUD

Joded smoys °H

popioosw

„SPIOM Zunjjads 9504] edueuqo S}9J—POO3 Sjeu[,,

Áq

1uonbosuo;)

J9A198QO

sjuaAg

g

43 Observational Recording

4, 998

QUI

Joy

(9UO(T,,

,9UO Uo[£) Se IIIM OA,

,9UO SIG} Je Surjoo[ 918 OM—SIIYD noÁ »xueu[,,

9q 3Snf,,

„Aeme

gsuodsel 194989} ON

o3 Ma IMS

sjuoAq Juanbasuo)

uonsonb

suonsanb

SIIMSUV

€ SMEJ(T

puey dn spjoH ap

Á[Je€Q19A SIOMSUE SII)

osuodso1 snooue1]xo JWOS

PIIIO

«994

€

‘yo,

dnoi3 € JaMsue 0} SJ18)S (Surpeoi ,9UO 310210j nOA,, S99CUWUIIS SIIGD au} 3UI}D0II09)

sasuodssy

(300qX1OA E UI) 21910 € MEICT,

gurpea1 Áog

«2194 BUIOP SI} SI FEY oui [[9) "Aqu,

SJU9SAT jUuopooojuy

SO:TT

QUIL

44

Chapter 2

test items. Cover the answer column with a sheet of paper; then write your answers in the blank spaces provided. The correct answers may then be checked. observational recording

. When duce

teachers look at behavior and prorecords of that behavior as it occurs,

they are engaging

verbal responses to teacher questions

crying hitting arriving to class on time

continuous recording event recording

duration recording interval recording momentary time sampling anecdotal reports

in

. List four examples of behavior that require observational

recording.

a.

b. c. d . List the five major observational recording techniques, as delineated by Hall. a. b. c. d. e. . Continuous

recordings

have

diary records. Continuous recording

been

called Or

produces a written narrative for a specified time period of both individual or group behaviors.

A time sequence is given Target behaviors are unselected Data are narrative descriptions of behavior Narrative is plain and easy to read

. What are the four common

as an assessment aid in

. Directive teachers should employ continuous recordings primarily

pinpointing behaviors for instruction to identify possible environmental

in most

continuous

elements found

recording techniques?

a. b. C. d.

a.

Observational Recording conditions that set the

occasion for student responses to identify possible consequent events that maintain behavior

45

b. c.

observed recorded

8. Perhaps the major reason for utilizing continuous recording is that a large range of behaviors can be and

directive teachers

9. Continuous recording is more appropriate for observers other than

Exercise 2-2 To practice continuous recording, select a television program whose characters are familiar to you (soap operas are excellent for this exercise). Make a four-column form such as the one Bijou used to record Timmy’s play yard behavior. For 5 minutes, continuously record the television program. During your first attempts, the temporal relationships of the actor's behavior may give you some problems. If so, return to Bijou's continuous recording of Timmy and study the record given in newspaper reporting style. Then study Bijou's four-column form. After this, try your own continuous recording again and you should have better results.

Event Recording Event recording is a tally or frequency count of discrete student behaviors as they occur. Event recording always takes the form of numerical products. Examples of discrete student behaviors appropriate for event recording include number of correct or error student verbal responses,

number of times a student engages in teasing episodes, number of times

a student is tardy, and so on. Pen and paper are sufficient for making event recordings. However, the following items may facilitate making frequency counts: 1. 2. 3. 4. 5.

Wrist golf counters Hand-tally digital counters Wrist tally boards Masking tape attached to wrist Tally with buttons or paper clips

Examples and descriptions of these items can be found in appendix A.

46

Chapter 2

Applied Example

Christensen and La Barbera (Hall, 1970) used event recording to measure talking between two 8-year-old boys. Even though the teacher separated the boys, they still talked to each other during work periods. These talk-outs interfered with the teacher’s work and distracted other students. During 15-minute work sessions, the number of verbal interactions between the boys was tallied. Christensen and La Barbera reported, “Any time either [boy] would speak to the other an interaction would

be recorded.

Thus,

if [one

boy]

spoke

to the

[other]

and

the

[other boy] replied, two interactions were recorded. Five or more seconds without any verbalization had to intervene between verbalizations in order to count as a separate interaction if the other of the pair made no reply” (p. 57). Considerations

Event recording may not be an appropriate measurement tactic (1) when behaviors are occurring at very high rates or (2) when one class of response can occur for extended time periods. Examples of high-rate behaviors could include rocking; rapid jerks of the head, hands, or legs; running; and tapping objects. Examples of behaviors that occur for long periods of time could include thumb sucking or task-oriented behaviors such as reading or listening. For high-rate behaviors and behaviors that occur

across

time,

duration

recording,

interval

recording,

and

time

sample recording (to be discussed later) are more appropriate. It should be noted that task-oriented behaviors occurring across time are usually not a terminal concern of directive teachers. The process of reading or listening, by itself, cannot be measured. Rate of reading and reading comprehension, not the process of reading, are the important dimensions of reading behavior. Similarly, movements of the student that show the student is utilizing what he hears, not the act of listening, are significant. Event recording may be expressed as number of occurrences or as frequency if opportunity for response and observation time are constant across sessions. If observation time is not constant, rate of occurrence

(rate = frequency/time) is the acceptable datum. Fvent recording is the major observational measurement tactic used by directive teachers. The advantages for using event recording are: (1) it is easily applied in classroom situations, (2) it does not interfere with ongoing teaching, and (3) it produces a numerical product.

Observational Recording

47

Evaluation

You may wish to evaluate how well you comprehend the additional information on observation recording by responding to the following test items. Cover the answer column with a sheet of paper; then write your answers in the blank spaces provided. The correct answers can then be checked.

recording is a cumulative tally or frequency count of discrete events as they occur.

Event

numerical

. Event recording always takes the form of products.

pen and paper golf counters hand tally digital

. List

counters

wrist tally boards masking tape on wrist

five

cording.

items

which

facilitate

event

re-

a. b. C. d. e.

high-rate behaviors responses that occur for extended time periods

. Event recording may not be an appropriate

tapping objects

. Give one example

listening

. Give one example of a behavior that may occur across extended time periods.

opportunity observation

time

measurement

tactic for:

a. b. of a possible high-rate

behavior.

. Event recording may number or frequency

be expressed as of occurrence if for response and are

constant across sessions. rate

. If

observation

time

is

not

constant,

is the acceptable datum.

48

Chapter 2

Exercise 2-3

Select a person that you are with for approximately 30 minutes daily. It could be a spouse, parent, friend, or colleague. For 5 days, 30 minutes daily, tally the number of verbal comments that are initiated and directed toward you by the selected person.

Duration Recording Event recording is perhaps the most common measurement

tactic for teachers. Some

and useful observational

student behaviors, however,

do

not lend themselves to event recording. Duration recording is an appropriate measurement tactic for behaviors that occur at very high rates (e.g., rocking; rapid jerks of the head, hands, or legs; running; tapping objects)

or for behaviors

that occur

for extended

time

periods

(e.g.,

task-oriented behaviors such as reading or listening). Basically, when a teacher is concerned with how long a student or class engages in a particular behavior, duration recording should be used. Measures of duration are usually reported as the percentage of time that a behavior or event continues or lasts. For instance, suppose a teacher was interested in how much time John spent on-task during a seat-work assignment. If John was on-task for 18 minutes during 30 minutes of seat work, the teacher would report that John was on-task 60% of the time. Most teachers use a stopwatch for recording durations of student behavior (see appendix A). As an alternative to the stopwatch, teachers can use a wall clock or wristwatch for duration measures. Duration measures obtained with a wall clock or wristwatch, however, will prob-

ably be less precise than measures obtained with a stopwatch.

Applied Examples

Krauss (1972) used duration recording during a family intervention study. Duration measures were reported as the percentage of time a child was engaged in a certain behavior. The following excerpts describe the objective and procedures of the study. The results are shown in figure 2-2. Behavior Measured. The mother complained that the subject was not "playing alone" when told to do so. The girl either dawdled choosing toys, persistently attempted to get the mother to join her in play, or called out to her mother for attention. This "playing alone" behavior was to be accelerated. . . . Since the father was not in the home during play periods, a reliability check was not made.

Observational Recording

49

Experimental Procedures and Results. During Baseline phase, the subject’s mean “play alone” behavior for 15-minute periods was 95%; the mother not only wanted this at 100% but desired the periods be lengthened to at least 30-minutes each. Television and praise: Beginning with the second week, a bell timer was used to indicate to the child and mother when the 20-minute periods were over. When the child’s nap schedule allowed, an afternoon period was recorded in addition to the routine morning period. If the subject played 100% of the period, the television was immediately turned on to a show determined to be rewarding to her (usually “Electric Co.”). If the subject did not play to criterion, a penalty was assessed: the child was to sit in front of a blank television for 3 to 15 minutes depending on her play behavior. This was necessary 5 times during the 3-week intervention phase. The timer was used also to indicate termination of time-out. Mean desired behavior during this phase was 92%; this was depressed by the initial day of intervention and by a second morning when the child’s routine was thrown off by sleeping late (past the usual play time). Although the mother found that verbal praise during play was disruptive to the child, she did insure that praise was given when the timer went off. Praise only: During this phase, the use of the timer and television were discontinued; this was done partially to discredit the possibility the timer itself was reinforcing play, as posited by the mother. Behavior was at 100% for 5 consecutive days of 30-minute periods.

Van Dyke(1972) also employed duration recording during a family intervention study. Duration measures were reported as the amount of time required for a child to complete a task. The results are shown in 2-3 and 2-4. Behavior

Measured.

Two

behaviors

were

measured,

both

related

to drying dishes. The mother reported, and the father agreed, this behavior literally ‘drove the whole family up the wall” disrupted family unity every evening. The first behavior was duration of time spent in drying the dishes in the evening. second behavior was the number of reprimands given by mother during the process. Reliability checks were provided by oldest brother, in which there was 100% agreement.

that and the The the the

Intervention Strategy and Results. A record was kept by the mother for 8 consecutive days to establish a baseline. In a conference with the subject, mother, and teacher, it was agreed that the client should receive a checkmark for every evening in which the dishes were dried within 40 minutes and in which the number of reprimands did not exceed five. Three contracted rewards were offered from which the subject chose the reward most appealing to her.

50

Chapter 2

Baseline

(20 minutes)

_

[|

t

100 [7N .9

|

Television and praise

ı

(15 minutes)! '

Praise only

(30 minutes) € mm mmm

9o[-

t

>)

(se)

e

70!

e)

601!

2ws A

S

a,

'6 $9 S3

S

8

30 40[—

:

30}

27

|

!

or = 5

!

{

L

|

|

|

|

10

15

20

25

30

[|

HEN 35

_L 40

45

Periods

FIGURE 2-2. Percentage of Time Spent in Playing-Alone Behavior by a 4-year-old Girl. (Adapted from “Increasing Playing-Alone Behavior by Making Viewing of a Favorite Television Show Contingent on Play in a Pre-Schooler” by S. J. Krauss, unpublished manuscript, Ohio State University, 1972.) . . . It is pointed out that the average time required for drying

dishes was slightly less than 52 minutes; the average number of reprimands slightly more than 16 during the baseline period. During the contingency management period of 17 days (subject chose the reward requiring 35 checkmarks and insisted upon it even though it was explained that it would necessitate a longer period of time),

the average

time

was

reduced

to slightly over

33

minutes;

the average number of reprimands was reduced to 5.4. As soon as the contracted reward was received, the average time was 65 minutes and the average number of reprimands was 17 over a period of 4 days. When the positive reinforcement was resumed, the average time was 37 minutes; the average number of reprimands 3.6 over a period of 5 days. Considerations

Duration recording is most frequently used in two ways. One method is used when teachers require data concerning the amount of time a

Observational Recording Contingency management,

Reversal ! Contingency baseline ; management,

(9)

4

1! 6

Lj | L

8

10

12

14

Ig

16

pe

L

ON

1 2

fpe

—l

R

6

ji 8

p

1

4

|_| 2 4

Dr

of

5 times out for illness

contracted reward given

i

©

Frequency of Reprimands

ol

WN

Baseline

51

Days

Reliability check

FIGURE 2-3. Number of Reprimands Required for Dish-Drying by a 9-year-old Girl According to Mother's Standards. (Adapted from "Reduction of Time Spent in Drying Dishes and Number of Reprimands Given by Mother During the Process by Using Positive Rein-

forcement" by M. University, 1972.)

Van

Dyke,

unpublished

manuscript,

Ohio

State

student emits behaviors within specified time periods. For example, if a kindergarten teacher was concerned with “isolate” behavior, the teacher could record the duration of “isolate” behavior that occurred during daily 30-minute free play periods. If “isolate” behavior went on for 25

minutes in a 30-minute period, the behavior as a percentage of time would be figured as follows: duration of “isolate” behavior

duration of free play periods

25 minutes

30 minutes

X

100 = 83%.

Thus in this hypothetical example, a kindergarten student engaged in “isolate” behavior 83% of one free play period. The Krauss study measured duration in this way. Another method is used when teachers

52

Chapter 2

require data concerning the amount of time a student takes to complete a specific task when no minimum or maximum time criteria are specified. For example, if a teacher was concerned with the amount of time a student took to walk from her classroom to the cafeteria for lunch, the teacher could report just time durations (e.g., for four successive days the student’s time in minutes to the cafeteria = 9, 11, 8, 10). The Van

Dyke study measured duration in this way. Baseline

!

Contingency management,

Reversal

100

baseline

95

; Contingency management,

(p

ask

909

75[70[Minutes

65Fr60[55[-

sol-

5 times out for illness

contracted reward given

80

45 = 40 = 35 30

(|

1.4

6

8

(©) Reliability check

1

10

ft

12

L Days

1

14

1

16

|

2

|4

4

L1

2

4

T7

L

2

oo

L1 J | 24 6 8

r=

20

QN

25F

FIGURE 2-4. Number of Minutes Required to Dry Dishes by a 9year-old Girl According to Mother’s Standards. (Adapted from “Reduction of Time Spent in Drying Dishes and Number of Reprimands Given by Mother During the Process by Using Positive Reinforcement” by M. Van Dyke, unpublished manuscript, Ohio State University, 1972.) Evaluation

You may wish to evaluate how well you comprehend the information on duration recording by responding to the following test items. Cover

Observational Recording

53

the answer column with a sheet of paper; then write your answers in the blank spaces provided. The correct answers can then be checked. Duration

recording

1.

is used when teachers are concerned with how long a student or class engages in a particular behavior.

percentage

2. Measures of duration are usually reported as the of time that a behavior continues or an event lasts.

stopwatch

3. The most precise nonautomated instrument for duration recording is a

to record the amount of time a student emits behaviors within specified time periods

4, What are the two ways that duration recording is most frequently used? a

to record data concern-

ing amount of time a student takes to com-

plete

a specific

task

when no minimum or maximum time

b.

criteria are specified

Exercise

2-4

Make arrangements to observe a child watching two or more television programs such as "Sesame Street," "Captain Kangaroo," or "Mr. Rogers' Neighborhood.” With a stopwatch, record the duration of time that the child attends to the programs. Calculate the percentage of attending time for each program. Was there a different duration of attending behavior for the different programs?

Interval Recording Event and duration recording together provide a composite picture of student behavior during a single time period. But at times the informa-

tion they furnish is not enough. Sometimes knowing the temporal pattern of a behavior can facilitate educational planning. For example,

54

Chapter 2

suppose Bill was engaged in nonstudy behavior 35% of the time. Were Bill’s nonstudy behaviors occurring throughout the entire period? Or predominantly at the start, or middle, or end of the period? If Bill’s nonstudy behaviors occurred primarily at the start of a study period, the teacher could introduce a plan to decrease the amount of time between the start of class and the start of Bill's study behavior. Interval recording provides an estimate of student performance across time intervals which is not possible with event and duration recording. Interval recording

is used

to measure

the occurrence

or nonoccur-

rence of behavior within specified time intervals. The total observation session is broken down into smaller time intervals of equal size. For example, if the total observation session was 10 minutes and the observer was using 10-second interval measurement, the session would be divided into 60 individual 10-second recording units. The interval size selected should provide the observer with sufficient time to observe and record behavior reliably. Simultaneous recording of multiple behaviors and their various rates of occurrence will both influence the interval length selected. However, interval size will usually range from 6 to 30 seconds depending on the behavior to be observed. To record occurrence or nonoccurrence of behaviors, a paper is ruled

into squares to differentiate time intervals, as shown in figure 2-5. The squares may be positioned either horizontally or vertically. To record,

v] w[x

[vx] x] viv] v|

[LX | 10") x

10" 10" 10" 10" 10" 10"

xev arme — [XIXIVIVIXIS]

v.

"

— id

10

x

|

One minute

TT

V

Student on-task

x

X

Student off-task

FIGURE ' 2-5

the observer checks time intervals in which the behavior occurred or did not occur. Behavior is usually recorded only once per interval and reported as percentage of occurrence. Figure 2-5 shows that the student was on-task 50% of the time: intervals on-task

total intervals

_ 6

—

72 — 30 X 100 = 50%

If more than one discrete behavior or more than one student is to be observed and recorded, the observer may add additional rows for each

Observational Recording

55

behavior or student (figure 2-6). The only thing which limits the number of discrete behaviors or students that can be measured concurrently is reliability of recording 10”

On-task

(Mattos,

1968).

10”

10”

10”

V4

10”

10”

v

Verbal Off-task

10”

10”

10”

10”

Viwiw v

d

Motor Off-task

Passive Off-task

viv

d

FIGURE

2-6

Reliability of interval measurement will probably decrease when the observer simultaneously records more than three or four response categories. However, the reliability of recording multiple behaviors or a group of students should increase with training and recording experience. When interval recording is used for a group, observers frequently record only one student per interval (see figure 2-7). John (10°)

. E S

1

Laura (107)

X

2 3

V

V

V

Alicia (107)

Daryl (10")

Mary Ann (107)

Hans (107)

V

X

X

V

X

4 5 6

VY

Attending

X

Nonattending FIGURE

2-7

The interval data sheet for John, Laura, Alicia, Daryl, Mary Ann, and Scott shows that John did not attend during the first 10-second

interval; Laura

attended

during the second

10-second

interval, and so

on. It should be noted that the check mark in this illustration does not mean necessarily that the child attended throughout the entire 10-second interval, just that attending behavior occurred sometime during that interval. Another tactic used in interval recording of multiple categories is to

observe during the first interval, record during the second interval what

56

Chapter 2

was observed in the first interval, observe again in the third interval, record in the fourth, and so on (see figure 2-8).

oS

D

22

8

2

10”

10”

2

On-task

§

2

2

+

©

zx

10^"

10”

©

9

À

2

10”

10"

9

V

Verbal Off-task

Motor Off-task

V

Passive Off-task

A

FIGURE

2-8

Interval recording commonly involves the use of pen, paper, clipboard, and stopwatch. The stopwatch should be attached to the clip of the clipboard. Commercial stopwatch attachment devices for clipboards are available, but the watch can be satisfactorily attached with rubber bands. A disadvantage of using the clipboard and stopwatch is that the Observer must periodically break eye contact with his subjects to observe the stopwatch. This may decrease the accuracy of measurement. Worthy (1968) described a miniature, portable timing and audible signal-generating device to be used for interval recording. He reported that "this device, which is small enough to fit into a shirt pocket, eliminates the need for the observer to visually monitor a stopwatch, etc. which detracts from his observance and recording of behavior." Worthy's technical note is complete enough that a TV- radio repair shop could build the timing device. Appendix A contains descriptions of timing devices. Jorgenson (1972) described a unique timing procedure for interval recording that does not rely upon special instruments. Her procedure is especially appropriate for teachers. She states that teacher-directed presentations (e.g., lectures, activities, songs) can be timed and presentation cues memorized to differentiate time intervals. For example, in the song "Twinkle, Twinkle Little Star"

(at a slow tempo),

intervals are marked by the words “wonder” and “high.”

10-second

Applied Examples

Jorgenson (1972) used interval recording to measure attending behavior of six trainable retarded children in music activity.

Observational

Recording

57

Attending behavior was defined as sitting in a chair at a desk and looking at the teacher and/or in the direction of the education activity. Once attending behavior was defined, a method of measuring the behavior was determined. Because no observers were available during the activities, a tactic was devised to conduct activities and record behavior without assistance from another person. Songs and activities were timed and 10-second intervals within each song and activity noted. Cues for 10-second intervals within each activity were memorized by the therapist. For example, in the song “Twinkle,

Twinkle

(at a slow tempo),

Little Star”

10-second

intervals occur at the words “wonder” and “high.” Each child was observed for a 10-second interval and a check mark ( / ) was placed under each child's name if he was attending during the 10-second period. If he was not attending, an X was placed under his name. Recording continued throughout

the

name without session.

sessions;

stopping

ie.,

was

a mark

placed

the activity throughout

under a

child's

the 30 times per

Table 2-3 is a sample observation chart showing Jorgenson's method of recording attending behavior for six children. Hall, Lund and Jackson (1968) used interval recording to measure nonstudy behavior, study behavior, teacher verbalization directed toward a pupil, and teacher proximity to the student (teacher within three TABLE 2-3. Sample Observation Chart Showing Method for Recording Attending Behavior of Six Trainable Retarded Children in Music Activity. (Adapted from "Use of a Music Activity and Social Reinforcement to Increase Group Attending Behavior” by H. A. Jorgenson, paper presented at the 50th Annual International Convention of the Council for Exceptional Children, Washington, D.C., 1972.) John

"Laura

Alicia

Daryl

1

X

J

V

X

X

V

2.

V

V

X

V

V

X

Note. Each

column represents a 10-second interval.

Mary

Ann

Scott

58

Chapter 2

feet). Their observational recording sheet and symbol structed as shown in figure 2-9. 1 0”

l 0"

1 0”

1 0”

1 0”

1 0”

1 0”

10”

10”

10”

1 0”

key was con1 0"

NININININININISISISININ TITIT

[| Row 1: Row2: Row 3:

N T /

T

7/

/

/

Nonstudy behavior S Study behavior Teacher verbalization directed toward pupil Teacher proximity (teacher within three feet)

FIGURE 2-9. Interval Recording Used to Measure Selected Teacher and Student Behaviors. (Adapted from “Effects of Teacher Attention on Study Behavior” by R. V. Hall, D. Lund and D. Jackson, Journal of Applied Behavior Analysis,

1968, 1.)

Considerations

Interval recording is a measurement procedure seldom employed by classroom teachers when they are instructing students. It is difficult for teachers concurrently to give instruction and to reliably observe and record behaviors with interval recording. This tactic requires the constant attention of the observer. Therefore, interval recording is an appropriate technique for ancillary teacher personnel (e.g., consulting teachers, psychologists, students, teacher aides) or for the teacher when

he or she is not interacting with students. An exception to this general rule is exemplified by Jorgenson’s study (1972), in which the teacher both directed the activity and employed interval recording to measure group attending behaviors. There are two major advantages in using interval recording. First, interval recording provides an estimate of both frequency and duration of behavior. Second, and perhaps most important, interval recording provides an estimate of student performance across time intervals which is not possible with event and duration recording. Interval recording generates information concerning probabilities of when a behavior is likely to occur or not occur. For example, it can tell us whether a behavior is likely to occur or not occur at the start, middle, or end of an

instruction period. Such information is valuable to directive teachers in programming teaching strategies, instructional materials, or schedules of reinforcement.

Observational Recording

59

Evaluation You may wish to evaluate how well you comprehend the presentation of interval recording by responding to the following test items. Cover the answer column with a sheet of paper; then write your answers in the blank spaces provided. The correct answers can then be checked. Interval

recording

1S

used to measure the occurrence or nonoccurrence of behavior within specified time intervals. . Selection

observe record

vide

the

reliably

of an interval

observer

with

size should

sufficient

and

time

pro-

to

behavior.

. Simultaneous recording of multiple behaviors and respective rates of occurrence jointly influence length.

interval

. Interval

size

will

onds depending observed. reliability

. With

interval

to

on

usually the

recording

range

behavior the

only

from to

sec-

be

thing

which limits the number of discrete behaviors or students that can be measured

concurrently.

is

of

re-

cording.

. When group,

one student

interval recording is used for a observers frequently record only _ per

interval.

. A disadvantage of using a clipboard and stopwatch for interval recording is that the observer must periodically break with his subjects to observe the stopwatch.

eye contact

a teacher-directed sentation

was

pretimed

and presentation cues memorized to differentiate time intervals

. What interval timing tactic did Jorgenson use when teaching six retarded children?

60

Chapter 2

instructing students

9. Interval recording is a measurement procedure seldom employed by classroom teachers when they are 10. Interval recording provides an estimate of and both of behavior.

frequency duration across time event duration

intervals

11. Interval recording provides of student performance which is with and

an

estimate

not

possible

recording. Exercise

2—5

Pinpoint

a behavior to observe.

10-second

Rule a paper into squares to differentiate

time intervals for a total of

5 minutes

(30 observation

intervals).

Using a stopwatch, pen, and paper, check the time intervals during which the behavior occurs. Compute the percentage of intervals in which the behavior occurred.

Momentary Time Sampling A major disadvantage of using interval recording is that it requires the undivided attention of the observer. It is almost impossible for teachers to do interval recording while giving instruction. However, momentary time sampling is a measurement tactic that can be used with ease by teachers

while

giving

instruction.

Momentary

time

sampling

is con-

cerned with recording the occurrence or nonoccurrence of behaviors immediately

following

specified

time

intervals.

Interval

recording,

on

the other hand, is concerned with recording behaviors during specified time intervals. Momentary time sampling usually employs time intervals expressed in durations of minutes (e.g., 5-minute or 10-minute intervals) whereas interval recordings usually use time intervals expressed in seconds

(e.g., 10-second or 20-second intervals).

If a teacher wishes to record the occurrence or nonoccurrence of a behavior, a paper is ruled into squares to differentiate time intervals as in interval recording. The squares may be positioned either horizontally or vertically on the paper. To record, the teacher checks time intervals in which the behavior was or was not taking place at the moment the interval ended. Behavior is recorded only once per interval and reported as percent of occurrence. For example, say a directive teacher is concerned with student on-task behaviors during a 30-minute seat-work

Observational Recording

61

assignment. The 30-minute lesson time could be divided into 5-minute intervals. The teacher then observes and records the student’s behavior six times during the 30-minute seat-work assignment. Observation and recording is made only at the end of each 5-minute interval. Regardless of student behaviors

during the 5-minute

interval, the teacher records

only the student’s behavior at the instant the interval timed out. Figure

2-10

shows

that

the

student

in

the

example

was

on-task

6624

percent of the time during his 30-minute seat-work assignment (4/6 X 100 = 66%%).

5|v 10 | v $

S|

V

V

20

| V

5

25|

X

30

X

Viv

=

= >

10

15

V

X

x

20

25

30

Minutes

X

Student off-task

v/

Student on-task

FIGURE

2-10

A teacher can use a wall clock or a wristwatch for time sampling. When giving instruction, however, it is difficult to time intervals without a signal device. Common kitchen timers have been found to be useful as signal devices for time-sample measures. The Lux minute-meter timer is an inexpensive timing device with a one-year guarantee against defects. The timer can be set at minute intervals up to one hour. A description of timers for signaling time-sample intervals is found in appendix A. Applied Example

Bushell, Wrobel and Michaelis (1968) in a preschool class of 12 children.

reported use of time sampling

The four principal observers were seated in an observation room. Each wore earphones which enabled audio monitoring of the class and also prevented inter-observer communication. On a signal at the beginning of each 5-minute period, each observer looked for

62

Chapter 2

the first child listed on the roster and noted that child’s behavior on the data sheet, then looked for the second child on the list and noted its behavior; and so on for each child. All observers were

able to complete the total observational cycle in less than 3 minutes.

During

the 75 minutes

of observation,

the children’s behavior was

described by noting what the child was looking at, to whom he was talking, and what he was doing with his hands. Fourteen daily observations of each child by each observer produced 672 items of data each day. Criteria were established by which each behavioral description on the data sheets could be coded as either "S," indicating study behavior, or “NS,” indicating nonstudy behavior. Behaviors such as writing, putting a piece in a puzzle, reciting to a teacher, singing a Spanish song with the class, and tracing around a pattern with a pencil were classified as "S," if they were observed in the appropriate setting. Descriptions of behaviors such as counting tokens, putting away materials, walking around the room, drinking at the fountain,

looking out the window,

rolling on the floor, and attend-

ing to another child, were classified as “NS.” Singing a Spanish song was scored “S” if it occurred during the Spanish period when called for, but “NS”

if it occurred during an earlier or later period.

Similarly, if one child was interacting with another over instructional

materials

during

the study

teams

period,

the behavior

was

labeled “S,” but the same behavior during another period was classified "NS." If a given child’s behavior was described 14 times and 8 of these descriptions were coded "S," then the amount of study time for that child was 8/14 for that day. The amount of study behavior for the entire class on a given day was the sum of the 12 individual scores, (p. 57) Considerations

Momentary time sampling is a useful tactic for the classroom teacher because numerical estimates of group or student behaviors can be ob-

tained while the teacher is involved in instruction or other activities.

Evaluation

You may wish to evaluate how well you comprehend momentary time sampling by responding to the following test items. Cover the answer column with a sheet of paper; then write your answers in the blank spaces provided. The correct answers can then be checked:

Observational Recording

63

immediately following

1. Momentary time sampling is concerned with recording the occurrence or nonoccurrence of behaviors specified time intervals.

during

2. Interval recording is concerned with recording behaviors | specified time intervals.

minutes

3. Momentary time sampling usually employs time intervals expressed in

seconds

4. Interval recording usually employs time intervals expressed in

timed out (ended)

5. To record with time samples, the teacher checks time intervals to indicate that behavior was or was not taking place when the interval . -

Exercise 2-6

Observe the same behavior that you recorded in exercise 2-5 (interval recording). Rule a paper into squares to differentiate 3-minute time intervals for

a total

of

30

minutes

(10

observation

intervals).

Using

a stopwatch,

pen, and paper, check the time intervals in which the behavior was going on when the interval timed out. Compute the percentage of intervals in which the behavior occurred.

Using Measurement in the Classroom Having reviewed the two basic approaches to measuring behavior (measuring permanent products and observational recording), it is appropriate at this time to address some questions which are frequently raised by teachers who are just beginning to make use of these measure-

ment techniques. Teachers frequently ask the question, What social and academic behaviors should be measured and graphed? Ideally, all student responses that are instructionally relevant should be recorded and graphed. However, a point may be reached when the information gained from measuring and graphing behaviors is slight enough that these activities are not an efficient use of the teacher's time. By analogy, when an individual who is physically well schedules a check-up, the physician will monitor very few physical responses. However, as physical health deteriorates, the physician will increase the monitoring of physical responses. If the individual is in extremely poor health, many bodily functions will be monitored and recorded. These data will then be used for making decisions about medical intervention: Should the operation

64

Chapter 2

be performed now or later? Should the prescribed medication be increased, decreased, changed, or stopped altogether? The same sort of reasoning can be applied in the practice of education. When students are showing a high degree of success in acquiring academic and social behaviors, measuring and graphing their behaviors may not greatly affect the outcomes of learning. The more severe the problems in learning, however, the more worthwhile it is for teachers to use

measurement and graphs in evaluating instructional programs. Thus, a regular elementary teacher may measure and graph fewer behaviors than a teacher of students with mild learning problems. As a general rule, teachers of the mildly handicapped are encouraged to measure and graph all student responses to direct instruction (e.g., reading, math, spelling). This rule excludes student responses to inde-

pendent

activities

such

pendent

activities

and

as seat work,

homework,

work unless the teacher is concerned about some anticipates

an

intervention.

and

learning

center

aspect of the indeThe

teacher,

how-

ever, should always give students feedback concerning their responses to nondirect instruction activities. Teachers of students with severe and profound problems in learning are encouraged to measure and graph all responses to both planned academic and social activities. A second important question for teachers is, What problems can be expected when measurement and graphing techniques are used? This is a difficult question to answer because each teacher and each classroom situation is unique. The measurement and graphing procedures used successfully by one teacher may not be sufficient or appropriate for another teacher or another classroom. Yet even with the variety of situations, common problems can be addressed. Most problems are associated with the development of sufficient teacher skills to successfully measure and graph behavior. Many university teachers, administrators, and supervisors have not recognized that measuring and graphing behaviors is a teacher skill. They have assumed that knowing about measurement is the same as being able to do it. Just because preservice or inservice teachers can describe how to measure and graph behaviors does not mean necessarily that they can apply these techniques in a teaching situation. Even practicum courses or student teaching experiences that give inservice or preservice teachers the opportunity to measure and graph behaviors can only approximate what they will have to do as classroom teachers. A major problem occurs when teachers try to move from a non-data-based instructional program to a data-based program overnight. Most who try to do this will fail because their skills are not developed sufficiently and they find the procedure too laborious to continue. To initiate a databased system requires a transition period of several months for most

Observational Recording teachers.

The

following

suggestions

may

help

teachers

to

make

65 the

transition. First, begin by measuring and graphing just one or two behaviors of one student. As you become comfortable with this, add another student. Measure and graph the same behaviors of the second student as you did with the first student. In this fashion, gradually expand the number of students whose behavior you have an interest in measuring. This early period should also be a time of trial and error in which you experiment with different techniques. Second, measure student responses for the shortest period of time that is sufficient to provide a sample of the student’s behavior. For example, if small-group reading instruction is planned for 20 minutes, take a 1minute sample of each student’s reading behavior rather than try to measure each student’s responses for the full 20 minutes. A 1-minute daily sample of academic behaviors such as reading and math usually provides sufficient data for most classroom situations. In situations where the teacher does not set the occasion for responses to occur (e.g., talking out), the length of time used to measure

the behavior will depend

on

how frequently the student demonstrates the behavior. The measurement period can be short if the behavior is occurring at a high rate. For example, if talk-outs are occurring at about one per minute, a 5-minute measurement sample of the behavior would be sufficient. On the other hand, if a behavior occurred only once or twice per day (e.g., fighting), all occurrences for the full school day should be recorded. Third, utilize adjunct resources such as teacher aides, volunteers, and the students themselves to measure and graph behaviors. Perhaps the best source of help is the students. Students can and should graph their own behaviors. An exception may be the severely retarded. Many teachers use their bulletin boards to display student-made graphs. Finally, it is important to identify a person skilled in measurement and

graphing

to provide

input

during

the

transition

period.

University

teachers, administrators, or supervisors are possible advisors. However,

another classroom teacher who is currently measuring and graphing behaviors in the classroom on a regular basis may be the best person to offer suggestions for implementation and to encourage teachers when they are beginning to get the “hang” of things. A third question which teachers raise is, How are graphed data used in instructional planning and decision making? Eaton (1978) has suggested that at least three days of collecting data are necessary before decisions can be made. She also suggested specific criteria for evaluating the first three days of data and described choices that can be made following the evaluation.

66

Chapter 2

Eaton has given the following criteria for a decision to discontinue, change, or maintain a program of instruction following 3 days of data: 1. Discontinue or change the program of instruction if the data points have a zero slope (i.e., behavior neither increases nor decreases). This condition is illustrated in figure 2-11. 2. Discontinue or change the program of instruction for a target behavior to be increased if over 3 days each day is lower than the previous day (see figure 2-12). 3. Discontinue or change the program of instruction for a target behavior to be decreased if after 3 days each day is higher than the previous day (see figure 2-13). 4. Discontinue or change the program of instruction when a target behavior meets or exceeds a mastery criterion for 3 consecutive days (see figure 2-14). When the student demonstrates mastery of a skill, the teacher should start work on a new skill. Postchecks of the student’s maintenance of the mastered skill should be scheduled. 5. Maintain the instructional program if the student’s performance does not fit one of the four conditions listed above. 100

7

90 = 9

80

8.

70

c LP)

KO or hi

gs

rm

s

^r

S$

20[l-

&

10

$

(o

5

sor

ob

I

——

1

2

pi; 3

|

P

|

])

yg

|»

gy

|

Days

FIGURE 2-11. No Change in Target Behavior in First 3 Days of Instruction Indicates Instruction Should Be Discontinued or Changed. (Adapted from “Data Decisions and Evaluation” by M. D. Eaton in N. G. Haring, T. C. Lovitt, M. D. Eaton, and C. L. Hansen (Eds.), The Fourth R Research in the Classroom, Columbus, Ohio: Charles E. Merrill, 1978.)

Observational Recording

67

100 £7

90 =

S$

sr

& x

nr N

=

in

= 5

20h

re

10l-

c

|»

B©

$ Q

sf

sop OF

4.34

pty

1

13

1

14

14

14»

1»

1 |I

Days

FIGURE 2-12. Decrease of a Behavior Targeted to Be Increased in First 3 Days of Instruction Indicates Instruction Should Be Discontinued or Changed. (Adapted from “Data Decisions and Evaluation" by M. D. Eaton in N. G. Haring, T. C. Lovitt, M. D. Eaton, and C. L. Hansen (Eds.),

The Fourth

R

Research

Charles E. Merrill, 1978.) 100

in the Classroom,

Columbus,

Ohio:

C

Percentage of Correct Responses

90

80 = =

70 60

=

50

=

L4

Days

FIGURE 2—13.

Increase of a Behavior Targeted to Be Decreased in

First 3 Days of Instruction Indicates Instruction Should Be Discontinued

or Changed. (Adapted from "Data Decisions and Evaluation" by M. D. Eaton in N. G. Haring, T. C. Lovitt, M. D. Eaton, and C. L. Hansen

(Eds.), The Fourth R Research in the Classroom,

Charles E. Merrill, 1978.)

Columbus,

Ohio:

68

Chapter 2

90 =

$ 6 &

M

8070h

o

Æ

our

©

-

Ô3

50

=

40

=

&

30r

3

20

m

©

10

=

0

--- Criterion

]

LI

ı 3

1 4

l 5

d

ld 7 8

6

1

IL

LL.

Days

FIGURE 2-14. Target Behavior At or Exceeding Criterion for 3 Consecutive Days Indicates Instruction Should Be Discontinued or Changed. (Adapted from “Data Decisions and Evaluation” by M. D. Eaton in N. G. Haring, T. C. Lovitt, M. D. Eaton, and C. L. Hansen (Eds.), The Fourth R Research Charles E. Merrill, 1978.)

in the Classroom,

Columbus,

Ohio:

References Baer,

D.

M.,

Wolf,

M.

M.,

&

Risley,

T.

R.

Some

current

dimensions

of

applied behavior analysis. Journal of Applied Behavior Analysis 1968, 1, 91-97. Bijou, S. W., Peterson, R. F., & Ault, M. H. A method to integrate descriptive and experimental field studies at the level of data and empirical concepts. Journal of Applied Behavior Analysis, 1968, 1, 175-191. Bushell, D., Jr., Wrobel, A., & Michaelis, M. L. Applying "group" contingencies to the classroom study behavior of preschool children. Journal of Applied Behavior Analysis 1968, 1, 55-61. Christensen,

A., & La Barbera,

J. The use of a feedback-reinforcement

cedure to decrease talking between cationally handicapped. In R. V.

pro-

two boys in a class for the eduHall (Ed.), Managing Behavior

(Part 3). Lawrence, Kansas: H & H Enterprises, 1971. Eaton, M. D. Data decisions and evaluation. In N. G. Haring, T. C. Lovitt, M. D. Eaton, & C. L. Hansen (Eds.), The fourth R research in the classroom. Columbus, Ohio: Charles E. Merrill, 1978.

Hall, R. V. Managing behavior (Part 1). Lawrence, Kansas: prises, 1971.

H & H Enter-

Observational Recording

69

Hall, R. V., Lund, D., & Jackson, D. Effects of teacher attention on study behavior. Journal of Applied Behavior Analysis 1968, I, 1-12. Jorgenson, H. A. Use of a music activity and social reinforcement to increase

group

attending

behavior.

Paper

presented

at the

50th

Annual

International Convention of the Council for Exceptional Children, Washington, D.C., 1972. Krauss, S. J. Increasing playing-alone behavior by making viewing of a favorite television show contingent on play in a pre-schooler. Unpublished manuscript, Ohio State University,

1972.

Mattos, R. L. Some relevant dimensions of interval recording procedures. Working paper #186. Parsons Research Center, Parsons, Kansas, 1968. Sulzer, B. and Mayer, G. R. Behavior modification personnel. Hinsdale, Ill.: Dryden Press, 1972.

Van

Dyke, M. reprimands forcement. Worthy, R. C.

procedures

for school

Reducation of time spent in drying dishes and number of given by mother during the process by using positive reinUnpublished manuscript, Ohio State University, 1972. A miniature, portable timer and audible signal-generating

device. Journal of Applied Behavior Analysis 1968, 1, 159-60. Wright, H. F. Observational child study. In P. H. Mussen (Ed.), Handbook

of research methods in child development. New York:

Wiley,

1960.

Reporting Teaching

Effects One hundred one look.

rumors

are

not

comparable

to

—Ancient Chinese inscription

After reading and applying the measurement techniques presented in part 1, you may have been thinking, “Now that I have pinpointed and measured a behavior for specified periods of time, how will I use this information in teaching?” Part 2 of this book, and later part 3, will

answer that question. Part 2 is concerned with reporting teaching effects through the use of graphs. Graphic presentation of classroom behaviors is a meaningful way for a teacher to communicate the effects of his or her teaching to students, parents, and administrators. Graphs

also tell the teacher how

well students are learning. The effectiveness of graphic displays for communication is attested by the wide use of graphs in government, business, and science. Graphs are a convenient means for summarizing and communicating large amounts of precise information. Chapter 3 discusses the use of (1) noncumulative graphs, (2) cumulative graphs,

(3) multiple concurrent graphic presentations, and (4) graphic presentations of central tendency and range scores.

71

Graphic Presentations

Directive teachers keep written records of student behaviors that indi-

cate when behaviors were measured, when and what intervention tactics

were applied, and the amount of behavior observed at baseline and during intervention. With this information, directive teachers can compare

(1) behaviors across instruction periods, (2) differential effects of instruction, and (3) magnitude of behavior change. Graphic presentation

of such data enables the teacher to make a visual comparison of behaviors across time and to communicate teaching effects to others. Graphic presentation is a much neglected activity in teaching. The main advantage of good graphing is that it allows the teacher to judge student performance and thus teacher performance accurately. Standard graphic arrangements employ two axes, a horizontal axis (x-axis)

and a vertical axis (y-axis), drawn

at right angles

(see figure

3-1). In directive teaching, the y-axis represents the amount of behavior

(e.g., frequency, rate, percent, proportion, duration), and the x-axis represents time (e.g., minutes, sessions, days, weeks) (see figure 3-2). 72

73

y-axis

Graphic Presentations

X-axis

FIGURE 3-1

Graphic Conventions in Directive Teaching Noncumulative Graphs

Amount of Behavior (percentage, rate, frequency, proportion, duration)

The most common form of graphic presentation in directive teaching is the noncumulative graph (line graph). To construct a line graph, a pair of axes is drawn on graph paper. Points representing amount of behavior (y-axis) are plotted opposite the points along the x-axis representing the times at or during which the behavior occurred. After points have been placed, they are connected by a line. For example, say a teacher was concerned with a student who emits inappropriate talk-outs during a 40-minute fifth-grade reading class. For 5 consecutive days, the teacher recorded the number of inappropriate talk-out responses during reading class. The number of inappropriate talk-outs per class were 7, 9, 3, 10,

Time (sessions, minutes, days, weeks )

FIGURE 3-2

714

Chapter 3 12

Baseline

11m

10

O0

D |

a

—

UA

ww

m

—_

BR

Frequency of Talk-outs

or

Days

FIGURE 3-3. Noncumulative Frequency of Talk-outs Emitted by a Student During a 40-Minute Fifth-grade Reading Class. and 11. Figure 3—3 is a graphic presentation of the talk-outs. All types of quantitative measurement (e.g., rate, frequency, percentage) can be reported with line graphs. Exercise

3-1

Construct noncumulative graphs from the following 3—3 as an example of a noncumulative graph. 1. 2.

Larry

completed

secutive weeks:

the following

number

data. Refer

of assignments

during

to figure 5 con-

11, 13, 10, 9, 12.

Lamont emitted the following number of inappropriate talk-outs during 5 observation

sessions

of 15-minutes

each:

10, 30, 25,

18,

11.

Cumulative Graphs Cumulative graphs are also used extensively in directive teaching. These graphs are constructed exactly like noncumulative line graphs with one difference. Cumulative graphs differ from noncumulative line graphs in that the amount of behavior observed and recorded for the first session is added to the amount of behavior recorded for the second session. The sum of the first two sessions is then added to the amount of behavior occurring in the third session, and so on.

Graphic Presentations

75

ME

t TI ror

WwW © N

e

Frequency of Talk-outs

WwW A

S |

Baseline

|

|

2

1 Days

|

3

|

FIGURE 3-4. Cumulative Frequency of Talk-outs Emitted by a Student During a 40-Minute Fifth-grade Reading Class. For example, if the talk-out data

(7, 9, 3, 10, and

from the hypothetical fifth-grade reading class were tive frequencies, the graph would be constructed as Compare the presentation of these same data in cumulative graph. The data points for figure 3—4 graphing the 7 talkouts for day 1. On day 2, the

11 occurrences)

reported as cumulashown in figure 3-4. figure 3—3, the nonwere derived first by student talked out 9

times. Therefore, a cumulative data point for 16 occurrences (7 [day 1]

+ 9 [day 2] = 16) was placed on the graph at day 2. On day 3, the student inappropriately talked out 3 times, which gave a cumulative total of 19 occurrences

(16 [days 1 + 2] + 3 [day 3] =

19), and so on. Figure

3—4 shows that during the 5 days of recorded data the student emitted 40 inappropriate talk-outs. Exercise 3-2

Construct cumulative graphs from the data presented in exercise 3-1. Refer to figure 3-4 as an example of a cumulative graph. Detailed Records with Cumulative Graphs

Cumulative graphs are excellent for maintaining detailed records of how behaviors are occurring across time and what responses were correct or incorrect. Suppose a teacher administered spelling tests. If this teacher recorded spelling scores in percentages or frequencies, she would have no record of what words the student had spelled correctly or incorrectly unless she kept all tests on file. But filing all test responses would be

76

Chapter 3

cumbersome, and precise comparisons of performance on specified words over extended time periods would not be possible. A more efficient procedure would be to cumulatively graph performance on each spelling test word by word and then file only the master copy of the test. To illustrate, suppose the teacher gave the following spelling test labeled October 13 and the student responded as indicated in the right-hand column below. The student’s spelling responses could be cumulatively graphed as presented in figure 3-5. It is easy to look at the graph and immediately see that 10 spelling words were given and that the student misspelled the 4th, 8th, and 10th spelling words on the list. Spelling Test for October 13 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

boy girl house horse dog cat bat chair table eleven

1

jj 2 3

Student’s Responses October 13 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

boy girl house house dog cat bat char table el

13 12 _ Il

10r

e

9m

g

7h

iaE

8 =

2 sp on

6 8

=

z

5P 4-

3f

)L 1

|

0r

1] jf 4 5 6

LLL 7 S8 9

| 10

d 11

| 12

| 13

| 14

| 15

Number of Spelling Words

FIGURE 3-5. Cumulative Graphic Presentation of a Student's Responses to Spelling Words After One Test.

Graphic

Presentations

77

Suppose that on the next day, the teacher gave another spelling test and the student responded as follows: Student’s Responses October 14

big

NNUMPBWNN

SO LHNAINNEHPRE

Spelling Test for October 14

horse up

down June

July

eleven chair

OR

. boy

p

u

. little

big little horse up down June July eleven chair

boy

The graphic presentation (figure 3-6) shows that on October 14 the student spelled all 10 words correctly and that he has spelled a total of 20 words out of which 3 were incorrectly spelled. The student’s spelling 13 Fr 12 [7 11 »

5

sa)

10 [F-

sk

e

8T

©

oF

=

g E3 Z

sr 4r 3

"$—9—9-—9—0-—9—9-——9-——9-——9

pe

2f ij

ol

123

1 LE

4

5

6

October 13

E

7

8

pitti ttrtititryiyt J

9

10,11

12

13 14

15 16

17

18

19 20

October 14

Number of Spelling Words

Cumulative Graphic Presentation of A Student’s ReFIGURE 3-6. sponses to Spelling Words After Two Tests.

78

Chapter 3

graph would continue as presented in figure 3-6. Note that the increments along the x-axis represent cumulative opportunities for response, and the increments along the y-axis represents cumulative error responses. The dotted vertical lines divide responses into test sessions.

Exercise

3-3

Construct a detailed cumulative graph from the following data (see figure 3-6). Rita was learning the Dolch word list. For one of the words Rita’s oral responses were recorded in the following way: Session Response

1

Correct

Incorrect

1.

V

2.

V

3.

V

4,

V

5.

V Session 2

Response

Correct

2.

v

LS
, Sr 9,

Positive teacher consequences

9

m

3m

35 fj

9 E3 c

30 I-

"E

$

©

sy£

20 >

2

15 =

7

10 = 5

2

3

Days

4

5

—

VAN

ı

|

MEN

1

2

3

! 4

5

Days

FIGURE 3-9. Teacher Consequences and Student Disruptive Behavior During 15-Minute Morning Periods. (Adapted from “Project Breakthrough Report” by B. Arnett and A. K. Dremann, unpublished manuscript, Ohio State University, 1972.)

Graphic Presentations

83

effectiveness of the training sessions in accomplishing the objectives of the course. Behavior. Teacher positive and negative consequences, and student disruptive

behaviors

were

defined

[see table 3-1]. These

behaviors

were concurrently recorded using a tally during three 5-minute intervals in a half hour during the school morning. A reliability check taken by two independent observers resulted in 86.5% agreement. Procedure. Teacher and students were observed for 5 days previous to the in-service training sessions, constituting pretest data. Nine training sessions followed, covering behavior and academic assessment, instructional tasks and behavioral objects, terminal behaviors and criterion levels, and evaluation procedures. Following these sessions, 5 days of posttest data were taken.

TABLE 3-1. Definitions of Positive and Negative Teacher Consequences and Student Disruptive Behaviors. (Adapted from “Project Breakthrough Report” by B. Arnett and A. K. Dremann, unpublished manuscript, Ohio State University, 1972.) Teacher Consequences 1. 2.

Verbal contact indicating approval or commendation to an individual, either loudly or softly (“that’s good,” “fine,” “good job,” etc.) Verbal reprimand to an individual. This can include critical comment such as yelling, scolding, threatening (such as in mentioning consequences if certain acts happen or fail to happen) or reprimand (“sit down,” “get quiet,” etc.) Physical contact and close physical proximity to a child when the teacher’s intent is to deal with the child in a positive way (patting, embracing, holding hand, helping with work, etc.) Physical contact and close physical proximity to a child when the teacher’s intent is to deal with the child in a negative or punitive way (shaking, spanking, grabbing, standing over while lecturing, etc.) Looking at child and smiling, nodding, winking or giving other indication of

approval

Looking at child and disapproval

frowning

or eyeing down

|

or giving other indications of

Student Disruptive Behavior 1. 2. 3.

Vocal: any audible behavior (talking, deliberate yawning, belching, whispering, whistling, etc.) |. uu Noise: any audible behavior, without permission, other than vocalization (tapping pencil, tapping of feet, moving desk, etc.) | Out of chair: any observable movement of the child out of chair, or out of a normal working position (walking about room, stretching out over the top of

the desk, etc.)

|

|

|

|

Play: any disruptive action to divert attention from the assigned task (playing with a toy, comb, cleaning fingernails, drawing or reading of non-task ma-

terials, etc.)

Nu

]

Aggression: any threatening action toward another person, (hitting, leaning or otherwise touching student, pulling chair away, taking papers, books or pencils, etc.)

84

Chapter 3 Results. The target teacher showed a decrease in her use of tive consequences from a pretest average of 8.4 occurrences minutes, to a posttest average of 1.2. Disruptive behaviors classroom also decreased from an average of 76 incidents minutes to an average of 5.2 during posttesting.

negain 15 in the in 15

The directive teacher may be concerned with both duration and frequency of behavioral events (e.g., the duration of minutes in the timeout area and the number of times placed in the time-out area). In such cases where both frequency and duration can be graphed together, a combination line and bar graph is a useful tool for showing behavior change. To illustrate, suppose a student demonstrated high durations of nonattending behavior during daily 30-minute seat-work periods. The teacher’s strategy for decelerating nonattending behaviors was to reinforce attending behaviors that were incompatible with nonattending responses. The number of attending behaviors that were reinforced was

gradually reduced as the target behavior came under control of the contingent reinforcement. Duration of nonattending behavior and number of reinforcers presented are given in table 3—2 for 7 days of intervention (reinforcement of attending behaviors). Figure 3-10 gives a graphic representation of the minutes of nonattending behavior per session and the number of reinforced attending responses. Intervention,

Number of reinforced

SS

attending responses

SS

Duration of nonattending

behaviors

|

BA

QN

CA

pma

—" ft

[/

G2

—

nd

t3

—

e

jt

MM

© |

Minutes of Nonattending and Number of Reinforced Attending Responses

29 -

Study Period Sessions

FIGURE 3-10. Duration of Nonattending Behavior and Number of Reinforcers Presented During 7 Days of Intervention.

Graphic Presentations

85

TABLE 3-2. Duration of Nonattending Behavior and Number of Reinforcers During 7 Days of Intervention. Study Period Session

Duration of Nonattending Behavior (minutes)

Number of Reinforcers Presented for Attending

1 2 3 4 5 6 7

20 18 15 7 5 2 2

15 15 8 6 4 1 1

Reporting Central Tendency and Range Scores Directive teachers typically instruct groups of students. Often group data will be used to describe class performance. In describing a group per-

formance, directive teachers should indicate the central tendency, that 1s,

a representative value for the entire class’s performance, and the range of the data, that is, the spread of scores. Central tendency and range can be depicted with graphic methods. Common

Measures of Central

Tendency: Mean and Median

The mean, the most common measure of central tendency, is the arith-

metic average of a set of data. If, for example, a teacher wished to deter-

mine the average score of a group of students on an addition test, she would add up the individual scores and divide that sum by the number of measurements. This procedure generates an average score which is called the mean. As an illustration, say 11 students emitted the following number of’correct responses on an addition test: 10, 15, 15, 13, 9, 7, 5, 11, 12, 10, 10. The arithmetic mean would be calculated as follows:

sum of the numbers number of measurements

The median is another middle score in a ranked into two groups of equal derive the median point

—

117

—

107

11

measure of central tendency. The median is the distribution. It is the point that divides a group number—a lower half and an upper half. To of the addition test scores for the 11 students

above, first place the scores in rank order: 5, 7, 9, 10, 10, 10, 11, 12, 13, 15, 15. Second, find the middle score that divides the measurements into

two equal groups (see figure 3-11a). The middle point, or median, is

86

Chapter 3

10, with five scores occurring below and five scores occurring above the median.

If the distribution is an odd number

of scores, the median

is

usually easy to determine. However, suppose the teacher only had the following 10 scores from the addition test: 7, 9, 10, 10, 10, 11, 12, 13, 15, 15. In this case, when the distribution is divided into lower and

upper groups, there is no score to divide the groups (see figure 3-11b). The last score of the lower group is 10 and the first score of the upper group is 11; therefore, a median point is placed midway between 10 and 11, at 10.5, even though no student had an addition score of 10.5 (see fiure 3-11c). A median point figured in this way is sometimes called an interplanted median. Lower group 12 3 4 5

Median

(a)

5,

10,

7,

9,10,10,

Upper group 12 34 S

oO

Lu © o n

11,12, 13, 15,15

(b) 8 ® Lower group H =

12345 7,

Upper group

22

9,10, 10, 10

12345

11, 12, 13, 15, 15

ee 2

S c

(c)

Bs

23 #212345

12345 7,

9,10, 10, 10,

(10.5)

11, 12, 13, 15, 15

FIGURE 3-11 The mean

(10.7)

and median

(10.5)

for the hypothetical

addition

test data were very similar. This is not always the case, however. Ordinarily the mean is the best index of central tendency. But if a few scores are exaggerated—extremely high or extremely low—in comparison to the group, then the median

would

be a better index

of central

tendency. Suppose a teacher wished to know typically how long it took his class to “settle down” to on-task behavior once they were back in the classroom after morning recess. Individual student scores for seconds

Graphic Presentations

87

of time that had elapsed until each student was on-task were: 10, 10, 13, 15, 15, and 180, which gave a mean of 37.6 seconds. It is obvious

that the 37.6-second mean does not provide an adequate picture of the “typical” student behavior. The single exaggerated score of 180 seconds inflated the mean score. In this case the median, 14 seconds, gives a better description of the total class behavior. Range

The range score is the difference between the lowest and highest score in a set of measurements. The range score of the hypothetical addition test data used above is 10 (see table 3-3). 2r

WW

YUM

o -3 AWA tA À

WE

@—@

SO

PI

AMA

MM

Mean or Median

Range

m

ND

wo

Nb

LEI TIL LLLLLLLLL LET LLLLLLL U LLLLL LU OO

WIN I

© o

Amount of Behavior

LLL ELLE LLL LLL

21r20 [-

Time

FIGURE 3-12.

Graphic Presentation of Central Tendency and Range

88

Chapter 3

TABLE 3-3.

Calculation of Range Score

Addition Test Scores

Calculation of Range

10 10 10 11

15 (highest score in distribution) — 5 (lowest score in distribution) —— 10 (range score)

Only two scores are used to calculate the range. Therefore, range scores do not show how individual scores are spread in a distribution. Range scores are used in directive teaching only to show the upper and lower scores of a class. Usually, in presenting the range of a class on a graph, both the upper and lower scores are graphed rather than just the range score. Evaluation You may wish to evaluate how well you comprehend reporting teaching effects by responding to the following test items. Cover the answer column with a sheet of paper; then write your answers in the blank spaces provided. The correct answers can then be checked. behaviors across instruction periods differential effects of instruction magnitude of behavior change

1. When directive teachers maintain written records of student behaviors that indicate when behaviors were measured, when and what instructional methods were applied, and the amount of behavior observed at baseline and with intervention, they can compare: a. b. c.

y-axis

2. Standard graphic arrangements employ two axes drawn at right angles. Identify these axes below. X-axis

Graphic Presentations

89

amount of behavior

. Usually in directive teaching, the units of measurement reported on the y-axis represent

time

. Usually the units of measurement reported on the x-axis represent

student performance

. The main advantage of good graphing is that it clearly allows the teacher to judge and

thereby teacher performance. noncumulative

. The most common form of graphic presentation in directive teaching is the graph.

directive teaching

. Cumulative sively in

Cumulative graphs

central tendency range

graphs

also

are

used

exten-

are excellent for maintaining detailed records of how behaviors are occurring across time and what responses were correct or incorrect. . In

describing

rective

a group

teachers

the

performance,

should

di-

the and

of the data.

mean

is 10. The average of a set of data.

median

11. The in a ranked

indicate

an

arithmetic

is the middle

score

distribution.

mean median

is the 12. Ordinarily the best index of central tendency, but if a few scores are extremely high or low the should be used.

range

score is the differ13. The ence between the lowest and highest scores in a set of measurements.

Points to Remember in Making Graphic Presentations Verbal descriptions written on the graph should be brief yet complete encough that the graph tells what the pinpointed behavior, teaching tactic, and results are without requiring the reader to refer to other

90

Chapter 3

sources of explanation. To illustrate, look at figure 3-13. Little information is given in figure 3-13. All we can say based on the graph is that some type of answers emitted by a 7-year-old student were recorded, that something changed during treatment (but we do not know

what), and that the data were accumulated over some 28 days. Baseline

Intervention

30 + 28 -

Number of Occurrences

26 =

Days

FIGURE 3-13. Number of Answers of a 7-Year-Old Student. (Adapted from “ Ohio Learning Disability Project, ” unpublished manuscript, Ohio State University, 1972.)

Now consider figure 3-14. Notice that the phases of the study are accurately labeled. The key below the graph describes each data line. The caption is complete and concise. Figure 3-14 provides information which can be used by the student’s teacher or other interested personnel

Graphic Presentations

30 -

Baseline

91

Self-recording of volunteered answers and teacher praise contingent on answers

g

9

>

=

9

© ©

Gi

$a

2

=

Z

3

6

9

13

16

19

22

25

Days o- — — — +

Number of teacher questions

9— © —

Number of times John volunteered and answered questions by raising his hand

FIGURE 3-14. Answers of a 7-Year-Old Boy in Response to Teacher’s Questions During a 10-Minute Language Arts Period. (Adapted from “ Ohio Learning Disability,” unpublished manuscript, Ohio State University, 1972.) (e.g., other teachers, parents,

ancillary school personnel)

to make

in-

structional decisions. Let’s look at the specific information the graph provides. First, we see that the pinpointed behavior was volunteering and answering teacher questions by raising hand. Second, we see that during baseline the teacher asked, on the average, approximately 17 questions during 9 daily 10-minute segments of a language arts period. During this condition, the student did not answer any questions. Third, when the teacher

92

Chapter 3

intervened by having the student self-record the number of times he volunteered to answer questions and the teacher praised him contingent upon his answers to her questions, the student’s behavior changed. Under this condition, the teacher asked, on the average,

19 questions during 18 daily 10-minute segments period. The student’s pinpointed behavior increased during baseline to a daily average of 17 answers. demonstrate a substantial increase in the pinpointed

approximately

of a language arts from zero answers Fourth, these data behavior and indi-

cate that the teacher’s intervention tactic was, at least, not ineffective.

Clearly, the presentation in figure 3-14 is more acceptable than that

in figure 3-13. Figure 3-14 meets the criteria for an acceptable graphic

presentation by describing the pinpointed behavior, the teaching tactic, and results of the teaching tactic without forcing the viewer to make recourse to other sources of information. Some final points to remember when preparing or reading graphic presentations are:

1. Vertical lines on the graph divide, or show changes in, teaching tactics (see figure 3-15).

2. Consecutive

data points

in different teaching

phases

(i.e., on

opposite sides of a vertical line) are not joined (see figure 3-15). 3. Postchecks are graphed when a behavior is periodically measured after the teacher has formally terminated training for that particular behavior. When graphing postchecks, do not join the data points. Graphic presentation of postcheck data is presented in figure 3-16.

Amount of Behavior

Baseline,

/

Self-recording,

!

Baseline,

|

Self-recording,

N

\ Wrong!

ht! D Data points ght!

are not joined

Data points should not be joined between conditions

between conditions

Time

FIGURE 3-15. Joining Data Points and Use Graphic Presentations.

of Vertical Lines in

Graphic Presentations Baseline,

Intervention,»

Baseline,

Intervention,

93

Postcheck

pe

e

e

5

®

= E tà

Wrong! Data

Right! Postcheck

=

joined

are not connected together

I

© E.

points

/V\

data points

| | | | L Days

FIGURE 3-16.

Postcheck Data in a Graphic Presentation

Evaluation You may wish to evaluate how well you comprehend reporting teaching effects by responding to the following test items. Cover the answer column with a sheet of paper; then write your answers in the blank spaces provided. The correct answers can then be checked. pinpointed behavior teaching tactic results

1. Verbal descriptions written on the graph should be brief yet complete enough so that the graph conveys information concerning the ,

and

without recourse to other sources of explanation. teaching tactics

. Vertical lines on the graph divide, or show changes in,

joined

. Data

joined

as

4-

points

indicated

between

by

vertical

teaching lines,

phases,

are

not

Data points in the postcheck phase are not

94

Chapter 3

References Arnett,

B.,

&

Dremann,

A.

K.

Project

breakthrough

report.

Unpublished

manuscript, Ohio State University, 1972. Crisp, E. M. Ohio learning disability project. Unpublished manuscript, Ohio King,

State University,

1972.

J. Project breakthrough

University,

1972.

report. Unpublished

manuscript,

Ohio State

Analytic Teaching

Instead

of speaking

of “control

groups,”

we

speak of “control procedures.”

—Donald L. Whaley and Richard W. Malott

Directive teachers are concerned with developing the technology of teaching. With this concern, it is not sufficient just to know that children have acquired certain skills during a course of instruction. The crucial question directive teachers must ask is: “What did I do as a teacher that caused the behavior change?” This question can be answered only in the classroom with analytic teaching. Analytic teaching is the educational process of demonstrating what student behavior is emitted in both the presence and absence of the teacher’s instruction procedures. Further, it is desirable to demonstrate that the teacher’s instruction procedures are responsible for the occurrence or nonoccurrence of specific student behavior. Chapter 4 addresses the topic of tactics for analytic teaching. The following designs are presented: (1) AB design, (2) reversal (ABAB) design, (3) variations on the reversal design, (4) multiple-baseline design,

variations on the multiple-baseline design, (6) changing criterion design, and (7) multielement baseline design.

97

(5)

Techniques for Behavior Analysis

In the introduction to part one, we said that “evaluation is based on measurement, which is the way to determine student growth.” This statement is true because measurement procedures document how much a student’s behavior changed. Also, measurement tells us when teaching is ineffective. But measurement procedures alone cannot prove that teaching is effective. As Risley (1969) notes, “Measurement can tell us that the student has improved, but even reliable measurement of improvement does not by itself indicate what caused the improvement.” In order to demonstrate teaching effectiveness, that is, a cause-and-effect relationship between a teaching tactic and a student improvement, an experimental design is required. Demonstrating what teaching procedures caused a student to improve has been difficult for teachers. Unlike measurement

tactics, which have

a well-developed methodology already established, designs that demonstrate causality with individual students or individual groups have only recently been introduced in public school education. All designs presented in chapter 4 use a baseline logic for evaluating the effectiveness of 98

Techniques for Behavior Analysis

99

classroom instruction on student behavior. Wolf and Risley (1969) have explained baseline logic as follows: Baseline logic comprises one question: tion substantially affect the baseline rate To “affect a behavior” means not only havior occurs but also that we have attribute that change in behavior to

(p. 3)

“Does a treatment condiof a student’s behavior?” that a change in the besufficient information to our treatment condition.

AB Design Perhaps the first classroom applications of the baseline logic used a procedure called the AB design. The AB design has two conditions: 1. Pretest or baseline period (repeated measures before intervention) 2. A period of intervention or instruction Figure 4—1 provides a graphic prototype of the AB design. Intervention

|

Amount of Behavior

Baseline

LLLI

!

[II 1 L 1 L I Time

FIGURE 4-1.

Prototype of the AB Design

Directive teachers use the following steps when design.

employing an AB

1. Assess social and academic skills. Teachers not familiar with classroom assessment should read Stephens (1974, 1976). 2. Pinpoint and define the behavior to be modified. For example:

Talking out is defined as a vocalization, comment, or audible vocal noise initiated by the student. It cannot be in response to

100

Chapter 4

the teacher or another peer. An occurrence will be tallied as one talk-out if it is separated from another occurrence by a breath, time interval, or change of topic. 3. Specify the terminal behavior. For example: Given a 10-minute segment of language arts, social studies, class discussion, or seatwork assignments, the student will emit 50 percent (or fewer) of the number of talk-outs emitted prior to the implementation of the instruction variable. 4. Select an appropriate measurement tactic. For example: event, duration, interval, or time sample recording.

5. Measure the occurrence or nonoccurrence of the target behavior for, usually, five or more measurement periods (baseline). Fol-

lowing five or more measurement periods, the directive teacher should proceed to step 6 (intervention) only if the baseline measurement is stable. That is, when you look at your graph, there is no major up or down trend in the data. During baseline, however, a student’s behavior may be consistently improving or regressing. In such cases, the teacher may not want to start instruction until there is an asymptotic trend in the data. Hall (1971)

John's

has commented on this trend:

rate

Therefore,

his

math

correctly,

of

working

if new

proficiency, the

math

problems

correctly

is

increasing.

[teaching] procedures were introduced to increase

even

. . . [teacher]

though

would

he

not

did

work

be

able

more

to

problems

determine

if

the . . . procedures were responsible for the change. . . . Experimental procedures can sometimes be started when a baseline is ascending, if the intent is to decrease the strength of the behavior, or, conversely, to begin experimental procedures when a baseline is descending, if the intent is to increase the strength of the behavior. This is often done in cases where it is desirable to reverse the trend of the behavior—for example, when a child is hitting his peers at an obviously increasing rate and the desire is to decrease the hitting behavior as quickly as possible (pp. 13-14).

6. Apply the instructional strategy (intervention). 7. Discontinue instruction after behavior has reached the terminal criteria. 8. Following the formal termination of instruction, occasionally

measure the pinpointed behavior to see if the skill is still approximating the terminal criteria (postcheck).

Applied Example

The following case study (Wilson, design (see figure 4—2).

1972) is an application of the AB

Techniques for Behavior Analysis

101

Subject. The subject was a 12-year-old male in fifth grade. He was reading on a third-grade level and had not made progress for the previous year and a half. The child was selected for this study because he demonstrated very slow academic growth and very poor study habits. Behavior. On-task behavior was defined for this directed to the activity for 15 minutes without down, covering work, or talking to others. Using cumulative amount of time directed toward the recorded during a 15-minute period. Reliability

study as attention putting his head a stopwatch, the assigned task was was taken by the

aide in the classroom, resulting in reliability figures of 80%,

90%

and

100%

100%,

on the 3rd, 4th, and 6th days of the study.

Procedures. Following baseline observations, a contingency management system was instituted in which the subject received 5 minutes of the teacher’s undivided attention, contingent on 15 minutes on-task behavior. Results. During baseline, the subject averaged 10 minutes on-task. After the contingency was instituted, the subject averaged 14.5 minutes on-task, earning the reward on 12 occasions. Baseline

20 bm

Teacher’s attention contingent on on-task behavior

18 =

Minutes Attending to Task

16} 14 = 12 r10 >» 8 — 6

—

4

—

2p

L1.

2

4

1!

6

1

8

]

10

L_

12

1

14

16

18

1

20

l

22

Days

FIGURE 4-2. Minutes Attending to Assigned Activities of a 12-yearold Fifth-grade Student During 15-minute Periods. (Adapted from Project Breakthrough Report by G. Wilson, unpublished

State University, 1972.)

manuscript, Ohio

102

Chapter 4

Considerations

The teacher can make statements concerning the ineffectiveness of instruction when using an AB design, but she cannot make statements concerning the causality of a behavior change or the effectiveness of instruction. For example, in the previous study by Wilson, variables other than contingent teacher attention may have accelerated on-task behavior. Perhaps the activities were more interesting or the parents intervened in some way. Numerous explanations could be proposed. Even though the student met the teacher’s terminal criteria, the teacher could not say that her teaching tactic was definitely the source of the improved performance. However, if the student’s behavior remained unchanged in both the baseline and teaching conditions, the teacher would be able to say that her instruction tactic was ineffective. To illustrate, suppose that data for a design like Wilson’s were reported as in figure 4-3. In this hypothetical example, the teacher would Baseline

20r-

Teacher’s attention contingent on on-task

behavior

18 [—16 [%

14

E

22k =

3

=

2 < Ö

2

z

10r

GT 6

—

4 2

—

| 1

3

5

|

|

7

l

9

11

| 13

] 15

l 17

| 19

}

21

Days

FIGURE 4-3. Hypothetical Example of the AB Design Demonstrating Ineffective Instruction.

Techniques for Behavior Analysis

103

know that her instruction tactic was ineffective since duration of ontask behavior was not increased. Data have been collected which the teacher can use to make decisions on changing instruction methods. The AB design is very useful in providing information on when to change instruction. Most data in applied behavioral analysis show that if a change in behavior is going to occur, the change comes in the first few training sessions. The implication for classroom teaching is that if a student has not acquired or improved upon the skill to be taught in one week of instruction the teacher should change his teaching procedures. Perhaps the instruction materials should be modified, or the task should be presented in smaller steps, or the consequence (reinforcement) of behavior should be changed. Basically then, the AB

design can establish whether

or not the stu-

dent’s behavior changed and can indicate the magnitude of change. This design cannot prove that teaching was effective for it does not provide controlled demonstrations of causality. The student’s behavior might have changed without the implementation of the instruction procedures.

Evaluation You may wish to evaluate how well you comprehend the AB design by responding to the following test items. Cover the answer column with a sheet of paper; then write your answers in the blank spaces provided. The correct answers can then be checked. baseline intervention

1. What are the two conditions associated with AB design? a. b.

defined

2. Before selecting a measurement behavior must be

tactic the in ob-

servable terms.

baseline

3. Measurement before vention is called

instruction

or intermeasure-

ment.

five stable

(asymptotic)

4. Baseline is ongoing usually for or more measurement periods. 5. The teacher should not introduce new teaching procedures until baseline data is

104

Chapter 4

where it is desirable to reverse the trend of the behavior

. When may it be desirable to introduce new teaching procedures even though baseline is not stable?

false

. The

AB

ineffectiveness

. The

teacher

changed magnitude

ness

of

design

can

demonsirate

instruction.

can

make

effective-

(true/false)

statements

ing the when using an AB design.

of

concern-

instruction

. The AB design can establish whether or not the student's behavior can indicate the

——

and of

change.

Exercise 4—1

Graph a hypothetical example of an AB design.

Reversal Design The AB design is incomplete for a behavioral analysis in that it does not provide information concerning what the student's behavior would have been had the teacher not introduced his teaching procedure. One evaluation design that makes a reasonable estimate of what the student's be-

havior would have been had the teacher not intervened is the reversal,

Techniques for Behavior Analysis

105

or ABAB, design. With this design, the teacher can demonstrate that teaching was effective. The reversal design (ABAB)

has four conditions:

1. Pretest or baseline period

(measurement

2. A period of intervention or instruction

before

intervention )

3. Return to baseline conditions 4. Reinstatement of instruction

Figure 4-4 provides a graphic prototype of the reversal design. Baseline,

Intervention,

D

|

Intervention;

|

Amount of Baseline

Baseline,

Time

FIGURE 4-4.

Prototype of the Reversal (ABAB)

Design

Directive teachers use the following steps when employing a reversal design.

e

Un

4

W

t

=

. . . . .

Assess social and. academic skills. Pinpoint and define the behavior to be modified. Specify the terminal behavior. Select an appropriate measurement tactic. Measure the occurrence or nonoccurrence of the target behavior for, usually, five or more measurement periods (baseline). Apply the instructional strategy. Intervention; should be op-

erative only until a trend toward the terminal criteria is observed, usually three to five training sessions.

106

Chapter 4

7. Return to baseline conditions. The purpose of baseline» is to demonstrate that in the absence of the intervention; procedure, behavior will again approximate the magnitude of behavior recorded in baseline,. With this operation the teacher is demonstrating causality. That is, he may demonstrate an increase in the probability that the behavior change was a result of his instruction rather than other variables. This condition is operative for a very short period of time (possibly three to five sessions). 8. Reinstate instructional strategy (interventions).

Discontinue in-

struction after behavior has reached the terminal criteria.

9. Following the formal termination of instruction, occasionally measure the pinpointed behavior to see if the skill is still approximating the terminal criteria (postcheck).

Applied Example The following case study (Priser, 1972) illustrates the reversal design. Subject. A 12-year-old student, who frequently talked out in class, was chosen for this study. He exhibited mimicking behaviors and disrupted the class. Behavior. Talking out was defined as a vocalization, comment or audible vocal noise initiated by the student. It could not be in response to the teacher or another peer. Each occasion was tallied as one talk-out, if they were separated by a breath, time interval, or change in topic. This behavior was recorded from 9:30 to 9:40 A.M. Reliability checks were taken by two independent observers; all resulted in reliability figures between 90 and 10046. Procedure.

After

baseline,

the teacher

instructed

the class,

"Let's

see how many can talk only after raising their hands and being called upon." She then chose a student exhibiting the desired

behavior,

and told him, “Good,

, you raised your hand

before saying anything." She then immediately chose the target student for the next class contribution,

and

verbally

forced the target child intermittently repeated Baseline conditions instructions and praise

praised

him.

From

then

on,

she

rein-

on all occasions of the desired behavior, and the class instruction. were reinstated. After the second baseline, were again in effect.

Results. Following a high rate of talking out during baseline, instruction and praise brought about a much lower rate. The behavior recovered to its high rate during baseline, conditions, but was again brought under control when contingencies were in effect.

Techniques for Behavior Analysis 57

L

Baseline,

Instructions and praise,

Baseline,

107

Instructions

and praise,

sir

|

D -J

Number of Vocalizations

45 =

Do

IN Pet

fit

2

11

6

J 7

1:1 8 9

f 10

TN

fit 11 12

1.10 13 14

Days

FIGURE 4-5.

Number of Classroom Talk-outs of a 12-year-old Stu-

dent. (Adapted from Project Breakthrough Report by J. Priser, unpublished manuscript, Ohio State University, 1972.)

Variations on the Reversal Design Three variations on the reversal design logic have appeared in the literature. Each of the variations exhibits common features. First, each varia-

tion can show whether a given stimulus consequent to a behavior had or had not affected that behavior. Second, once a consequential stimulus is presented (reinforcement), the analysis does not call for the removal of that stimulus condition but rather a rearrangement of the contingency.

Third,

each

variation

can

demonstrate

causality.

each design employs a different control condition: 1. Noncontingent reinforcement 2. Contingent reinforcement of a behavior previously reinforced behavior

To

this end,

incompatible with a

3. Differential reinforcement of other behavior (DRO)

Noncontingent Reinforcement This type of reversal design is excellent for demonstrating causality if the intervention tactic consists of contingent application of social reinforcement, e.g., standing near the student, smiling, laughing, verbal

108

Chapter 4

praise.

A frequent

hypothesis

to explain

advanced

behavior

change

generated from social reinforcement is that the behavior change resulted

from an improved student-teacher relationship rather than the contingent response-reinforcement arrangement. It is argued by some that it would not matter how the teacher’s praise and attention were given just as long as the student was in a warm, loving, accepting environment. The design prototype in figure 4-6 shows the contingent and noncontingent phases of the design. = —

Baseline

Amount of Behavior

—

Noncontingent iContingent ‘Noncontingent**; reinforcement — ireinforcement (given on (given contingent variable time on a specific or fixed time

intervals regardless of student behaviors)

on m -

Ibehavior) *

Contingent

N

-—

=

CN L

ect LLL

oe

LLL.

LL

LL

3!

tf 1

d

L_1 1

Time * Reinforcement in this condition should be equal to or less than the amount of noncontingent reinforcement presented in the noncontingent component. ** The amount of noncontingent reinforcement in this condition should equal the amount given in the preceding contingent phase.

FIGURE 4-6.

Prototype of the Noncontingent Reinforcement Design

Applied Example

Baer and Wolf (1970) have used a noncontingent reinforcement design to report an increase in cooperative play by a preschool child. Their study is summarized below and data for the study appear in figure 4-7. [The teachers first collected] baselines of cooperative and other related behaviors of the child, and of their own interaction with the child. Ten days of observation indicated that the child spent about 50% of each day in proximity with other children (meaning within 3 feet of them indoors, or 6 feet outdoors). Despite this frequent proximity, however, the child spent only about 2% of her day in cooperative play with these children. The teachers, it was found, interacted with this girl about 20% of the day, not

Techniques for Behavior Analysis

109

all of it pleasant. The teachers, therefore, set up a period of intense social reinforcement, offered not for cooperative play but free of any response requirement at all: the teachers took turns standing near the girl, attending closely to her activities, offering her materials, and smiling and laughing with her in a happy and admiring manner. The results of 7 days of this noncontingent extravagance of social reinforcement were straightforward: the child’s cooperative play changed not at all, despite the fact that the other children of the group were greatly attracted to the scene, Baseline

[ = cp E

sr

E2 o ©

Contingent reinforcement,

5

E A

Contingent reinforcement,

Z

à

0

2

S ©

©

$

sor

o

20 -

.E = 2 rf Un 00

S =

9

9 aw 10 I-

A tel L 1

| 5

"ad 10

15

I 20

L 25

| 30

J 40

Days

FIGURE 4-7. Example of a Reversal Design Using Noncontingent Reinforcement: Cooperative Behavior of a Preschool Child. (Adapted from “Recent Examples of Behavior Modification in Preschool Settings” by D. M. Baer and M. M. Wolf in C. Neuringer and J. L. Michael (Ed.), Behavior Modification in Clinical Psychology, ©

printed by permission Jersey.)

of Prentice-Hall,

Inc.

1970, pp. 14, 15. Re-

Englewood

Cliffs, New

110

Chapter 4 offering the child nearly double the cooperatively. These 7 days having the teachers then began their planned behavior. They defined cooperative served categories, subdivided into activities,

with

forcement

in

which

a team

chance to interact with them produced no useful change, reinforcement of cooperative behavior in four easily obnine classes of very specific

of observers

achieved

reliabilities of

92% and better. Contingent social reinforcement, used in amounts less than half that given during the noncontingent period, increased the child’s cooperative play from its usual 2% to a high of 40% in the course of 12 days of reinforcement. At that point, in the interests of certainty, the teachers discontinued contingent reinfavor

of

noncontingent.

In

the

course

of

4

days,

they lost virtually all of the cooperative behavior they had gained during the reinforcement period of the study, the child showing about a 5% average of cooperative play over that period of time. Naturally, the study concluded with a return to the contingent use of social reinforcement, a recovery of desirable levels of cooperative play, and a gradual reduction of the teacher’s role in maintaining

that behavior.

(pp.

14—15)

Contingent Reinforcement of an Incompatible Behavior Incompatible behavior, as used here, refers to pairs of behaviors that are opposites or that cannot coexist, such as in seat—out of seat, writing was for saw, talking-nontalking, hand up—hand down. To reinforce an incompatible behavior as a control procedure use the following steps:

o3

NO

1. Establish baseline on the target behavior (e.g., number of times students raise their hands to be called on to answer teacher’s questions). . Reinforce the target behavior (e.g., hand raising). . Discontinue reinforcement of the target behavior and apply the reinforcement contingency to the incompatible behavior (e.g.,

hands down). 4. Return to reinforcement of the target behavior (e.g., hand rais-

ing) and havior.

discontinue

reinforcement

of the incompatible

be-

Reinforcement of incompatible behavior is an appropriate tactic when Specific behaviors continue to persist after the teaching procedure is withdrawn. With this tactic the teacher may demonstrate that she can exercise control over events that increase the probability of the occurrence or nonoccurrence

improved

of that behavior.

in the reinforcement

condition

That is, student behavior

but

approximates

is

baseline

Techniques for Behavior Analysis

111

levels during reinforcement of incompatible behavior. Figure 4-8 is a prototype of the reinforcement of an incompatible design. Baseline

Reinforcement ! Reinforcement

of target

behavior

behavior

behavior

|.

Amount of Behavior

=

i Reinforcement

of incompatible ! of target

LL

LL.

LL

1

ld

1 1 1!

Time

FIGURE 4—8. Prototype of Design Using Reinforcement of an Incompatible Behavior. Applied Example The following case study

compatible behavior.

(Cooper,

1970)

uses reinforcement of an in-

Subject. An 11-year-old male student who was enrolled in a learning disability classroom persisted in reversing double numbers in his computation assignments (e.g., wrote 21 for 12; 31 for 13; 41 for 14). An example of his classroom work is presented in figure 1—1, p. 9, of this book. Baseline.

The teacher selected five number reversals which the student

presently reversed:

13, 15, 17, 18, and 19. Each of the five numbers was

printed on 5 separate 4" x 5" note cards, making 25 response cards. On

each

of an

additional

15 cards,

one number,

word,

or letter that the

student did not reverse was printed. A total of 40 response cards was used to obtain baseline data. During training the teacher and student were alone in a small tutor room. The teacher sat at a table facing the student. Prior to the session, the teacher shuffled the 40 response cards to achieve a random ordering. To initiate the session, a pencil and an 8V2" x 11" piece of ruled paper were given to the subject. After pencil and paper were given to the student, the teacher looked at the response cards in sequence and asked the

112

Chapter 4

student to write the word, number, or letter on the card. At no time did

the student see the writing on the cards. For example, the teacher looked

at the first card and said, “Write the number

13.” Then the teacher laid

the card face down on the table and proceeded in identical manner with the remaining cards. No praise for correct responses or instruction for reversed responses was given during baseline.

Model

and

Reinforcement.

The

model

and

reinforcement

condition

was the same as baseline with two exceptions. First, whenever the student wrote 31 for 13 or 81 for 18, the teacher placed the corresponding note card of the correct model of the reversed number in front of the student and asked him to write the number again next to his reversed number. The next word, number, or letter in the random sequence was given following the student’s response to the model. Second, for each 13 or 18 response that the student wrote correctly, without a visual model setting the occasion for the response, the teacher colored one rectangle of a series of 15 small (approximately 4%” x V2"^) vertical rectangles. When the rectangle series was completely colored the student was given a baseball (reinforcement). After meeting reinforcement criteria one time, no other reinforcement was programmed for the study. Incompatible

Model.

During

this condition

an

incompatible

model

was presented (1.e., 31 was substituted for 13, 81 for 18). Whenever the student wrote 13 for 13 or 18 for 18, the teacher placed the note card

with the corresponding incompatible model in front of the student and asked him to write the number again. The next word, number or letter in the random sequence was given following the student’s response to the incompatible model.

Postcheck. Postchecks were made under original baseline on the 7th, 15th, and 31st days after termination of training. Figure 4—9 gives a graphic presentation of the student’s sponses as a percent to the numbers 13, 15, 17, 18, and 19 following conditions: baseline,, model and reinforcement, model of an incompatible behavior, and follow-up.

conditions correct reunder the baselines,

During the first 6 days of baseline, the student emitted almost no correct (nonreversed) responses to 13, 15, 17, 18, and 19. When the

model and reinforcement condition was applied to the numbers 13 and 18, the student’s correct responses to all five numbers changed from zero in the last session of baseline, to 64% in the first session of model and reinforcement and 100% in the second session. Following this behavior change the teacher removed the model and reinforcement contingencies and returned to baseline conditions to see if the behavior would approximate the previous baseline level. The student did not reverse his double

Techniques for Behavior Analysis Baseline,

! Correct model ! Baseline,

i

Percentage of Nonreversed Responses

Incorrect

model

and reinforcement

Correct

| Postcheck

model,

e

e

100 I-

113

©

o

90 == 80 |— 70 I—60 >

50 40 j= 30 [20 —

days after die

10 — oL

termination of study

A

L

| 1

6

8

| 10 School Days

l

l

12

13

L'111 16

71531

FIGURE 4-9. Example of Design Using Reinforcement of an Incompatible Behavior: Nonreversed Written Responses to the Numbers 13, 15, 17, 18, 19 by an 11-year-old Student in a Learning Disability Class-

room.

(Adapted from Eliminating Letter Reversals with Modeling and

Reinforcement Procedures, unpublished doctoral dissertation, University of Kansas, 1970.)

numbers for two sessions in baselines. Each double number was written

correctly 100%

of the time. Since the student did not write number re-

versals in baselines, the next condition was started;

a model of an incom-

patible behavior (e.g., 31 for 13) was presented whenever the student made a correct response of 13 or 18. Percent of nonreversed numbers decreased immediately following onset of the incompatible model contingency. All double numbers

(13, 15, 17, 18, 19) were reversed in the

second session of the incompatible model contingency. At this point, the correct model was presented contingent on a reversed 13 or 18 response. Correct number responses increased immediately. In the first session of this contingency, 65% of the student’s double number responses were correct, 92% of the student’s responses

were correct in the second session, and all responses to 13, 15, 17, 18,

114

Chapter 4

and 19 were correct in the third and fourth sessions. The teacher terminated formal training on reversed double numbers at this point. However, follow-up postchecks were conducted to ascertain if the behavior change was being maintained. The behavior was checked with baseline conditions on the 7th, 15th, and 31st days after termination of training. In each case 100% of the double number responses were correct.

Differential Reinforcement of Other Behaviors (DRO) The use of differential reinforcement of other behaviors in applied analysis 1s similar to the incompatible behavior design. However, with differential reinforcement of other behaviors (DRO) any behavior except the target behavior may be reinforced. To apply the DRO analysis the following steps are taken: 1. Establish baseline on the target behavior. 2. Reinforce the target behavior. 3. Differentially reinforce the occurrence of any behavior except the target behavior. In this condition, reinforcement is usually given on fixed or variable time schedules contingent on the nonoccurrence of the target behavior. For example, if the target behavior was working on-task, the teacher could at the end of each 3 minutes reinforce any behavior that was occurring except working on-task. 4. Return to reinforcement of the target behavior. Figure 4-10 is a prototype of the DRO design. Applied Example

Reynolds and Risley (1968) reported a study designed to increase the frequency of talking of a 4-year-old disadvantaged preschool girl. The girl exhibited an extremely low frequency of verbal behavior even after a normal adaptation of the preschool setting and activities.

Baseline. During baseline, the percentage of 10-second intervals during which the student talked was recorded. The percentage of verbalizations during baseline averaged 11% during the first 129 days of school. Intervention,.

As

lowing conditions:

intervention

Reynolds

and Risley imposed

the fol-

(1) social interaction by the teachers was contingent

upon student verbalizations;

(2) whenever the student requested mate-

rials, access to them was contingent upon the student’s responding to teacher questions. The percentage of student verbalizations, during 73

Techniques for Behavior Analysis Baseline

|

Reinforcement,

|

Reinforcement,

|

|

Amount of Behavior

E

DRO

115

LL

I

1

LH

I

LI

L

LL

JI

Time

FIGURE 4-10.

Behavior

(DRO)

Prototype of the Differential Reinforcement of Other

Design.

days of the intervention condition, increased from an average of 11% during baseline to 75% of the 10-second intervals. Differential

Reinforcement

of

Other

The percentage of decreased to an average

Behaviors.

student verbalizations during 6 days of DRO of 6% of the 10-second intervals.

Since the frequency of teacher attention was now higher, it became necessary to investigate whether the increased instance of teacher attention per se or the contingency of presenting teacher attention was maintaining the verbal rate. It might be said that the child verbalized at a higher frequency simply because the teachers were attending and talking to her more, indicating that the higher incidence of teacher attention rather than its contingency of following the child’s verbalizations was maintaining this frequency. Therefore, the teacher attention was maintained at as high a rate

116

Chapter 4

but was now made contingent upon nonverbalization by the child. Typically the teachers would attend to the child, praising her and providing her with materials while she was silently engaged in activities. For example, if the children near the child were asking for water and she picked up a cup, the teacher would reinforce the child’s behavior of not asking by pouring water into her cup and keeping it filled as long as she was silent, and praising her for pouring from her cup, working hard, and keeping busy. Teachers removed their attention and the supplying source of materials for 15 to 30 seconds immediately following a verbalization by the child. (This procedure is often described as differential reinforcement of other behaviors [DRO] since teacher attention is presented contingent upon any behavior except the behavior being measured, in this case talking.) (p. 259)

Interventions. Teacher attention was again given contingent upon the verbal responses of the student. The percentage of student verbalizations, during 4 days of this condition, increased to an average of 51% of the 10-second intervals. Data reported from the Reynolds and Risley (1968) study appear in figure 4-11.!

Features of the Reversal Design Often teachers ask, “Why must I collect baseline data? Why not start directly with instruction?” The major reason a teacher concerned with analysis collects baseline, data is to establish a basis for predicting future student behavior. For example, suppose a student receives 80%, 82%, 75%, 81%

and 80%

on consecutive math worksheets. With these data,

made

the initial baseline.

the teacher could predict that if no major changes occur in instruction or in the physical or environmental life of the student the scores probably will continue to be approximately 75% to 82% on similar math worksheets. Prediction alone, however, is not the primary aim of analytic teaching. A directive teacher must verify that the baseline, prediction probably was accurate. Baseline, is used to verify the prediction of behavior from

If the

student’s

behavior

in baseline.

starts to approximate what it was in baseline,,then the teacher can have a high degree of confidence that the initial prediction was accurate. Another advantage of reversal designs is that they provide for replication of effect. In education, a teaching strategy that can produce an

* These data do not represent their complete study: “Further experimental analysis demonstrated that social interaction per se was not the reinforcer which maintained the increased verbalization; rather, for this child, the material reinforcers which accompanied the social interaction appeared to be the effective components of teacher attention” (p. 253).

Techniques for Behavior Analysis

100 F^ =

90

Or

=

70 =

e

>

©

60

=

©

50

p

$

40}

=

A

o D

DRO

Intervention.

30 F-

S = o

20

=

m

10

=

0

—

=

Intervention,

m

=

E

Baseline

117

11%

1

129

Days DRO—Teacher attention is presented any behavior except talking.

contingent

on

Intervention, „„.„—Teachers’ social attention contingent upon verbalization; student requests for materials

granted contingent upon student verbal teachers’ questions about those materials.

response

to

FIGURE 4-11. Average Percentage of 10-Second Intervals of Talking by a 4-year-old Preschool Child. (Adapted from “The Role of Social and Material Reinforcers in Increasing Talking of a Disadvantaged Preschool Child" by N. J. Reynolds and T. R. Risley, Journal of Applied Behavior Analysis, 1968, I, 259.)

effect only one time is not a major development. It is important to show that the instructional strategy is controlling the occurrence or nonoccurrence of student behavior. A reinstatement of the intervention following baseline, provides for a replication of effect. The more replications, the more convincing the effect. For instance, behavior changes shown with an ABABAB design would be more convincing than those shown with an ABAB design. All designs employing baseline logic to evaluate classroom instruction must supply sufficient information to attribute a change in behavior to the teaching strategy. Sufficient information includes (1) prediction, (2) verification of prediction, and (3) replication of effect. All three aspects must be present in order for the design to show a cause-and-effect, or functional, relationship. If, for example, behavior in baselines approxi-

mates the recorded behavior in baseline,, the teacher has demonstrated

118

Chapter 4

a higher probability that the behavior change occurred because of her teaching technique. She can state with some assurance that it was her manipulation of instruction tactics that produced the change in student behavior. For an illustration of the features of reversal designs see figure 4-12.

=

Baseline,

Intervention,

Intervention,

Baseline,

|

=

CN =

®

»

-

intervention,

-

eo

6

,

2

[7

Baseline data

Verification

E

["

behavior might

made from

E