Citation
Social acknowledgement of reading: multiple schedule effects /

Material Information

Title:
Social acknowledgement of reading: multiple schedule effects /
Creator:
Phelan, Pamela
Publication Date:
Copyright Date:
1976
Language:
English
Physical Description:
vii, 115 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Child psychology ( jstor )
Error rates ( jstor )
Experiment design ( jstor )
Flash cards ( jstor )
Learning procedures ( jstor )
Oral reading ( jstor )
Oral surgical procedures ( jstor )
Research methods ( jstor )
Social reinforcement ( jstor )
Tutoring ( jstor )
Dissertations, Academic -- Foundations of Education -- UF ( lcsh )
Foundations of Education thesis Ph. D ( lcsh )
Reading -- Programmed instruction ( lcsh )
Reinforcement (Psychology) ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Bibliography: leaves 108-114.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Pamela Phelan.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
025551668 ( AlephBibNum )
03022061 ( OCLC )
AAU1625 ( NOTIS )

Downloads

This item has the following downloads:


Full Text










SOCIAL ACKNOWLEDGMENT OF READING:
MULTIPLE SCHEDULE EFFECTS










by

Pamela Phelan


A DISSERTATION PRESENTED TO THE
GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY




UNIVERSITY OF FLORIDA

1976
















TABLE OF CONTENTS


Abstract . . . . . . . . .

Chapter

I. INTRODUCTION . . . . . . .

Purpose and Importance of Study . . .

Background of Study: Two Major Research
Projects . . . . . . . .

Schedules of Reinforcement . . . .

The Staats Study . . . . . .

General Statement of the Research Purpose .

Definitions of Terms . . . . . .

Limitations . . . . . . . .. .

II. LITERATURE REVIEW . . . . . . .

Major Developments in Social Reinforcement
Research . . . . . . . .


Page

. . v





2


3

3

5

9

9
. . 1

. 1 2
. . 3

. . 3

. . 5

. . 9

. . 9

. . 13

. . 14


Satiation and Deprivation . . . . .

Subject Characteristics and Their Influence
on Social Reinforcer Effectiveness . .

Experimenter Effects . . . . . .

Informative and Affective Roles . . . .

Relationship between the Response and
Reinforcement Delivery . . . . .

Summary of Historical Developments . . .









Page
II. Literature Review (cont'd)

Schedule Research on Reading . . . . . 31

Schedules of Token Reinforcement . . .. 31

Schedules of Social Reinforcement. . . . 34

Procedural Requirements of a Study of Social
Reinforcement Schedule Effects on Reading. 35

III. METHODS AND PROCEDURES . . . . . .. 42

Research Hypotheses . . . . . .. 43

Experiment I. . . . . . . .. 43

Experiment IE--. . . . . . . . 45

Experiment III . . . . . . ... 46

Subjects . . . . . . . . . . 48

Tutors. .... . . . . .. . . 48

Remedial Reading Program . . . . ... 49

Setting. .. . . . . . . . . 49

Apparatus. ... . . . . . . . 49

Materials. .. . . . . . . . 50

Data Collection. . . . . . . . 50

Data for Data Analysis . . . . . .. 51

Procedures . . . . . . . . . 51

Flash Card Procedures. . . . . . . 51

Oral Reading Procedures. . . . . . 52

Adaptation of the PDI Model. .. . . .. 52

Point Reinforcement Procedures . .. . . 54

Experimental Procedures. . . . . . . 54

Social Reinforcement Procedures. . . ... 55








Page

IV. RESULTS . . ... . . . . . . 59


Experiment I . .

Subexperiment I- .

Subexperiment 1-2

Subexperiment 1-3

Summary of Experiment

Experiment II . .

Summary of Experiment

Experiment III . .

Subexperiment III-1

Subexperiment III-2

Summary of Experiment


I Results



II Results.


III Results


V. SUMMARY AND DISCUSSION. . . . .

APPENDIX A Point Sheet. . . . .

APPENDIX B Daily Data Sheet . . .

APPENDIX C Standard Behavior Charts .

Bibliography . . . . . . .

Biographical Sketch . . . . .













Abstract of Dissertation Presented to the
Graduate Council of the University of Florida in
Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy


SOCIAL ACKNOWLEDGMENT OF READING:
MULTIPLE SCHEDULE EFFECTS

by

Pamela Phelan

August, 1976


Chairman: Donald Avila
Major Department: Foundations of Education

This study investigated the influence of social

reinforcement on the reading behavior of children enrolled

in a tutorial reading center. Four black third grade

pupils from a Gainesville, Florida elementary school served

as subjects. The basic data were daily rates of Flash Card

responses and Oral Reading responses. Programmed instruction

materials were adapted for use in this study. The study

was divided into three experiments, each comprised of two

experimental phases. The basic question being asked was

whether there is a predictable relationship between the

Multiple schedule of social reinforcement and the subjects'

mean rates of response.









Social reinforcement was delivered by university

students who functioned as tutors. Two tutors were

assigned to each subject. The tutors taught subjects on

alternate days in one-to-one sessions over a six-week

period.

Experiment I investigated Multiple schedules of

social reinforcement of correct responses in which the two

Multiple schedule components were a Variable Ratio schedule

and a Fixed Interval schedule. Schedule components

assigned to tutorsin Phase I were switched in Phase II.

Experiment I included three subexperiments. The Variable

Ratio schedule was expected to generate higher rates of

responding than the Fixed Interval schedule in both phases.

Experiment II used a slightly different design called

a contrast design. Multiple schedule components were

equivalent in Phase I and in Phase II one of the components

was changed. Correct responses were reinforced. Phase I

components were a Variable Interval schedule and one of

these was replaced by a Variable Ratio schedule in Phase II.

A contrast effect was expected. Similar rates of responding

were expected in Phase I. In Phase II the rates for the

changed tutor were expected to increase from Phase I to

Phase II and the rates for the unchanged tutor were expected

to decrease from Phase I to Phase II.








Experiment III, similar to that of Experiment I,

investigated the effects of Multiple schedules in which one

of the two components reinforced error responses. The

tutor that reinforced corrects in Phase I reinforced errors

in Phase II, and the tutor that reinforced errors in

Phase I reinforced corrects in Phase II. Experiment III

was comprised of two subexperiments. Reinforcement was

delivered on equivalent Fixed Interval schedules in the

first subexperiment and on equivalent Variable Interval

schedules in the second one. Higher rates of correct

responses were expected to be generated by the tutors who

reinforced corrects in both phases. Likewise, higher rates

of error responses were expected by the tutors who rein-

forced errors in both phases.

The results of the three experiments were not in the

expected direction. Two limitations of this study may have

influenced the results. First, precise measurement of rate

of response was impossible since the time used for reinforce-

ment delivery was included in the total time used to calculate

rate. Second, the verbal statements used by the white tutors

as reinforcers may not have been reinforcing to the black

subjects. This study demonstrates clearly, however, that

social reinforcement research can be conducted on an

educationally relevant behavior in a public school setting.




Chairman













CHAPTER I
INTRODUCTION

All societies develop structures to resolve and

modify conflict among their members. Mechanisms for

allaying a conflict may be formal or informal. Laws and

their enforcement codify norms while folkways and social

standards informally govern behavior through social rewards

and punishment. Social reinforcement is a societal or

cultural tool for shaping and maintaining acceptable

behavior.

The methodological application of social reinforcement

techniques in the management of human behavior is a wide-

spread practice. These management techniques have been

used for training and supporting behavioral appropriateness

in varied settings such as schools, mental health treatment

centers, and homes. Since incorporation into daily

operating procedures, social reinforcement techniques have

become a viable tool for change.

While the need for pragmatic psychological procedures

is recognized and, indeed, pursued by social agencies,

there is little evidence of sustained application of social

reinforcement techniques to cognitive task performance in

education. This lack became increasingly clear as extensive

literature research for the present study progressed.





-2-



Purpose and Importance of the Study

The major goal of this investigation is to examine

social reinforcement effects on cognitive task performance.

More specifically, this study investigates for the first

time the effects of the systematic use of social rein-

forcement on the reading behavior of elementary school

children.

A second aim is to develop research methodology and

procedures for the application of laboratory technology

to the study of social reinforcement effects in reading.

The successful specification and implementation of these

procedures will provide a research strategy for the continuous

assessment of social reinforcement effects on reading

behavior. Furthermore, implementation within the context

of a remedial training program encourages the use of the

natural environment for basic research programs.

It is significant that the literature on social

reinforcement effects on cognitive task performance contains

only a few references of its use over more than a few

sessions' duration. Many current educators and research

psychologists agree that (1) most of the existing studies

are superfluous to an ongoing assessment of social rein-

forcement effects, and (2) that most studies fail to

relate to the classroom teacher in more than an indirect

fashion. For these reasons, the need for basic research in

a context directly relevant to the educational process is

obvious.









Past research germane to reading, using social

reinforcement as the experimental treatment, has been

limited to short term studies of letter discrimination,

sight word vocabulary responding, and/or unit reading

responses. This study investigates the effects of

systematic manipulations of social reinforcement during

long-term reading development.

Background of the Study: Two Major Research Projects

A substantial number of research reports which have

direct and indirect bearing on this study will be reviewed

in Chapter.II. There are, however, two studies which have

such a fundamental impact on the current investigation that

they will be noted at this point. The first, representing

a substantial program of research, explicates the term

schedules of reinforcement. The second describes a study

on schedules of token reinforcement of reading and has a

direct bearing on the present investigation.

Schedules of Reinforcement

B.F. Skinner conducted the foundation research in the

area of schedules of reinforcement. A lengthy work,

Schedules of Reinforcement (Ferster and Skinner, 1957)

summarized the whole of his investigations. From this

work, the term schedule of reinforcement is defined as

a systematic arrangement of response-reinforcement contin-

gencies which vary the proportion of reinforced responses

or the time intervals after which reinforcement is obtained.








Many response-reinforcement contingencies have been

labeled, and derive from two categories of schedules:

simple and complex. The six simple schedules are

Continuous Reinforcement schedule (CRF), Fixed Ratio (FR),

Variable Ratio (VR), Fixed Interval (FI), Variable

Interval (VI), and Extinction (EXT). These are defined

in the Glossary. Complex schedules combine two or more

simple schedules. Some of the complex schedules are the

Multiple schedule, the Mixed schedule, the Tandem schedule,

and the Concurrent schedule.

Although it may be instructive to describe the various

schedules and their pattern effects, it is beyond the scope

of this paper to do so. However, description of the Multiple

schedule is germane to an understanding of schedules in

general and to the current study in specific. The Multiple

schedule was used in both the present study and the token

reinforcement study which will be discussed in the next

section.

A Multiple schedule of reinforcement is one in which

each of two or more randomly alternating schedules of

reinforcement is signalled by the presence of a different

stimulus. An example illustrating a Multiple CRF/EXT

schedule is selected to clarify the definition. Consider

that a subject is seated in a room with two lights overhead,

one red and one green; with one lever and one receptacle

in front of him. Illumination of the red light signals the

CRF schedule, while illumination of the green light signals








the EXT schedule. For each lever press while the red light

is illuminated the subject received a penny in the receptacle.

Lever presses in the green condition produce nothing. Thus,

it can be seen that the two stimuli, red and green lights,

signal the occurrence of the two schedules, CRF and EXT.

Ferster and Skinner's major finding demonstrated that

patterns of responding emerged unique to each schedule.

For example, the VR schedule typically generated higher

response rates than the VI schedule. Research has replicated

these patterns with different species and reinforcers. The

schedule was considered in control of the subject's

behavior when the pattern emerged in the data.


The Staats Study

Schedule research on educationally important behaviors

is rare. The utilization of social reinforcement in such

investigations is virtually non-existent. A discussion

of a token reinforcement of reading study is chosen to

provide a frame of reference for understanding the present

social reinforcement study. The purpose of the Staats

study (Staats, Finley, Minke, and Wolf, 1964) was to demon-

strate schedule control of human reading behavior. A description

of the methodology and procedures is followed by a

discussion of the results and their implications for the

present investigation.

Each of four four-year-old subjects was assigned to

one of four Multiple schedules of token reinforcement.









These schedules were Multiple CRF/EXT, Multiple CRF/VR,

Multiple CRF/VI, and Multiple VR/VI.

An apparatus presented materials, delivered marble

reinforcers, exchanged marbles for trinkets, and recorded

data. The material presentation panel was mounted on top

of a table in front of the subject. A doorbell-type push-

button was also on the table in front of the subject. Four

plexiglass windows, one centered above three others, were

mounted on the panel. The printed material was viewed

through the top window; a matching stimulus and two others

through the bottom three windows. Beneath each of the

lower three windows was a button which, when pressed,

activated a marble dispensing device if the correct match

was selected. Correct matching responses were reinforced

by the delivery of a marble to a small box on the table.

The subject could immediately exchange the marble for a

small trinket or could place it in a tube for accumulation

toward one of four more costly trinkets. A cumulative

recorder recorded the subject's responses, the point of

operation of the marble dispenser, and the point of delivery

of a toy. The operation of the doorbell, windows, marble

dispenser, exchange equipment, and cumulative recorder was

controlled by electromechanical equipment.

A series of training sessions prepared the subject for

the task of appropriately responding to the written materials.

Initially simple echoic responding was reinforced. The

response required for reinforcement gradually approximated





-7-


that of the experimental response which was vocalizations

of written letter combinations.

The final chain of responses required for reinforcement

was the following: the subject pressed the doorbell which

resulted in the appearance of the reading characters in

the windows; the Experimenter said the name of the character

in the top window which the subject echoed; the subject,

then pressed the button below the correct matching window

which activated the marble dispenser. If the subject

vocalized the correct response prior to the Experimenter's

prompt, the Experimenter activated the marble dispensing

machine. The chain began again when the subject pressed

the doorbell for the display of a new set of characters in

the windows.

Marble reinforcement occurred throughout the training

and the experiment. During training every correct response

was reinforced with the presentation of a marble. When

the subject was ready to begin the experimental response

chain, the Multiple schedule reinforcement contingencies

were placed in effect.

Two procedures controlled the change of component and

stimulus requirements of the Multiple schedules. During

one component, an overhead light in addition to a light

above the apparatus was illuminated. For the other

component, just the overhead light was illuminated. The

change of components of the Multiple schedule was determined





-8-

by the following: the CRF was in effect for 15 reinforcers,

the VR for 5 reinforcers, the VI.for 10 minutes, and

the EXT for 5 minutes.

Response rates were reported for each of the components

of the Multiple schedules. Results were discussed in

terms of their compatibility with basic schedule research.

Reference to this occurred in the previous section of this

chapter. (1) The Multiple CRF/EXT subject produced rates

in the CRF condition which exceeded the rates produced in

the EXT condition. These results were compatible with

basic research. (2) The Multiple CRF/VR subject produced

rates in the VR condition in excess of those in the CRF

condition. These results were compatible with earlier

research. (3) The Multiple CRF/VI subject's production of

higher rates in the CRF condition than in the VI condition

was compatible with basic research results. (4) Unexpectedly

the Multiple VR/VI subject produced approximately equal

rates in both conditions. It had been assumed that the VR

condition would generate decidedly higher response rates

than the VI. In summary, each case in which the CRF

schedule had been the comparison schedule resulted in data

compatible with basic research findings.

These results demonstrated that basic research can be

conducted on a significant human behavior. The laboratory

analysis of token reinforcement effects has wide generali-

zability to both the study of reinforcement schedules and

the development of reading. In this regard, a study with





-9-.


a similar research paradigm which also utilizes social

reinforcement would seem to have even greater applicability

to human behavior and the educational process.


General Statement of the:Research Purpose

This study will demonstrate the effects of Multiple

schedules of social reinforcement upon the rate of verbal

responses to written material by elementary school children.

A study of this nature is without precedence. The

application of laboratory technology to the development of

reading in pupils hrolled in a remedial reading program

increases its significance.

In Chapter II, the historical antecedents relating

to the development of the specific research hypotheses will

be presented and discussed. The review of literature is

intended to place these introductory remarks within the

perspective of an historical overview. The specific

research hypotheses will then be presented in Chapter III

where they are discussed with respect to research design

and methodology.


Definitions of Terms

The following list of terms and their definitions is

provided for the reader in order that they might fully

understand the phraseology used in this work.





-10-


Baseline: the ongoing rate of a behavior prior to the

introduction of the treatment.

Behavioral contrast or contrast effect: the effect on

performance on an unchanged schedule of reinforcement

to which the subject is exposed.

Celeration: basic unit of measurement of behavior change.

Change in frequency per unit of time; movements/

minute/week.

Concurrent schedule: Schedules which are in effect simul-

taneously, or which can follow one another in rapid

succession.

Conditioned reinforcer: a reinforcer whose reinforcing effect

depends upon a history of training. The actual

reinforcer maintaining the rate of the behavior is

the stimulus immediately following it. In the case

of the pigeon, the reinforcement for pecking is the

sound and light accompanying the operation of the

food dispenser. These stimuli in turn set the occasion

on which the pigeon may go to the feeder and eat.

Continuous Reinforcement schedule (CRF): reinforcement of

every instance of a target behavior.

Contrast design: a design which will potentially produce

the contrast effect.







Cumulative record: a graphic record which emphasizes the

rate of performance or its frequency. In a cumulative

record a recording pen moves along the abscissa with

passage of time and along the ordinate each time a

performance occurs. Thus, the rate of the performance

is demonstrated by the slope of the curve.

Cumulative recorder: gives a continuous visual picture of

the subject's rate of response.

Extinction (EXT): the operation of nonreinforcement.

Fixed-Interval schedule (FI): reinforcement is produced

by the first response emitted after a certain time

has elapsed since the last reinforcement.

Fixed-Ratio schedule (FR): the target response is reinforced

upon completion of a fixed number of responses counted

from the preceding reinforcement.

Geometric mean: the appropriate method for obtaining an

average on a ratio scale. See Pennypacker, Koenig,

and Lindsley (1972) for a description of how to

determine the geometric mean on the Standard Behavior

Chart.

Movement: recorded behavioral event. Usually specified in

terms of movement cycle with a beginning, middle, and end.

Multiple schedule: A schedule in which reinforcement is

programmed by two or more simple schedules alternating,

usually at random, each schedule being accompanied by

an appropriate stimulus as long as the schedule is in

force. The simple schedules of a Multiple schedule are

often called components of the Multiple schedule.





-12-

Rate: basic unit of behavioral measurement. Generally,

number of movements per unit of time. Specifically,

movements per minute; the measure found on the

Standard Behavior Chart.

Schedule control: When performance appropriate to a

particular schedule is generated by a stimulus which

is not a primary reinforcer, the stimulus may be said

to exert schedule control.

Schedule of reinforcement: a systematic arrangement of

response-reinforcement contingencies which vary the

proportion of-reinforced responses or the time

intervals after which reinforcement is obtained.

Standard Behavior Chart: a six-cycle logarithmic chart

designed for the graphic display of ratio data such

as rate.

Tandem schedule: A schedule of reinforcement in which a

single reinforcement is programmed by two or more

schedules acting in succession without correlated stimuli.

Target behavior: see movement. Typically, this would be

the dependent variable.

Token reinforcement: usually a metal or plastic disc which

an organism can carry around and exchange for

privileges, food, or other items.

Variable-Interval schedule (VI): the same as FI except

that the time from a reinforcement to the availability

of the next reinforcement is variable.






-13-

Variable-Ratio schedule (VR): the same as FR except that

the number of responses required for each reinforcement

is variable, not fixed.


Limitations
Two limitations of this study may have influenced the

results in a negative way, and one limitation influenced the

generality of the findings. First, precise measurement of

rate of response was impossible since the reinforcement

delivery time was included in the total time used to calculate

rate. It was expected that when compared with an FI schedule

a VR schedule of reinforcement would generate higher rates

of response. However, since the reinforcement density was

greater during the VR than during the FI, there was less

total time in which the subject could respond. This would

affect the rate calculation. Second, it is not clear whether

the verbal statements used by the white tutors as reinforcers

were in fact reinforcing to the black subjects. The subjects

may have lowered their rate of response in order to reduce

the frequency of reinforcement. This may also explain the

negative results. In this regard, the term reinforcer is

used throughout the text, although this is not meant to imply

that the events used as reinforcers were in fact reinforcing.

Third, since there were only four subjects, there is no

intent to generalize the results to any other subjects or

populations.













CHAPTER II
LITERATURE REVIEW

This chapter integrates historical perspectives of

social reinforcement research with the procedures of this

study. The major developments found through review of

the literature are (1) investigations of social reinforce-

ment satiation and deprivation effects on conditionability;

(2) evaluation of subject characteristics and their

influence on the effectiveness of social reinforcers;

(3) examinations of experimenter effect; (4) delineation of

informative and affective roles of social reinforcement;

and (5) systematic manipulations of the relationship between

the response and reinforcement delivery. After indicating

these developments, the literature review examines schedule

research on the social reinforcement of reading and discusses

the procedural requirements of such research.

Major Developments in Social Reinforcement Research

The five major developments represent differing views

of the nature of social reinforcement. The purpose here

is to present studies which most clearly exemplify those

beliefs. In this regard, no attempt was made to summarize

all experiments using social reinforcement. Instead,

studies were selected which illustrate typical practices.







-15-


Satiation and Deprivation

In the past, theorists speculated that satiation and

deprivation had opposing effects on social reinforcer

effectiveness. Availability of social stimuli was said to

determine the probable influence of reinforcement. Pre-

sumably deprivation enhanced the value of social reinforce-

ment while satiation reduced the value. If so, then,

reinforcement subsequent to a period of deprivation would

result in a significantly greater production of behavior

than would reinforcement after a period of satiation.

A common method of inquiry was to compare responsivity

to reinforcement after isolation and non-isolation. Typically

an "isolated" subject waited alone in a waiting room for

a period of time prior to performing the experimental task

while an adult remained with "non-isolated" subjects. The

adult provided reinforcement during the task performance.

The performance of the "isolated" subjects would be expected

to be better than that of the "non-isolated" subjects.

Studies like Erickson's (1962) and Gewirtz and Baer's (1958a,b)

provided supportive results.

The unstimulating conditions of institutions for the

retarded have also served as settings for social reinforcement

research. Klaber, Butterfield, and Gould (1969) in a unique

study compared the responsiveness of retarded persons from

two such institutions; one Institution (A) which allowed

social interaction, and Institution (B) which did not.





-16-


Interestingly enough, subjects selected from Institution A

had in the recent past been residents of Institution B.

Children from both institutions were placed in a room

with an adult and some toys. The data showed that

Institution B subjects played for longer periods of time

and maintained less distance from the adult than did

Institution A subjects. Thus, the transferred subjects who

had moved from deprived to less deprived circumstances

appeared to need social interaction less than the subjects

who remained in the low social interaction institution.

One important interpretation of these results might be that

the effects of social deprivation are reversible.

The study which demonstrated most succinctly the

opposing effects of satiation and deprivation was the

Landau and Gewirtz (1967) study. A critical feature of

the design made the clarity of their results unequaled.

Satiation levels were experimentally controlled and their

effects on performance measured. "Well done's" administered

at four frequencies, 4, 12, 30, or 60, consequent to

correct discrimination describes the procedure. Subjects

received one of the four frequencies of reinforcement during

the satiation period and subsequently during the task

performance.

The greater the frequency of reinforcement during the

satiation period the less effective was the reinforcement

on task performance. Furthermore, subjects who had received






-17-

the lower frequencies of reinforcement, 4 or 12, were

conditionable, while the ones who received the higher

frequencies, 30 or 60, were not. These results demonstrated

that the effectiveness of a reinforcer may vary inversely

with the frequency of its prior delivery.


Subject Characteristics and Their Influence on Social
Reinforcer Effectiveness

This section discusses conclusions drawn from the

body of research which investigated individual characteristics.

Experimenters undoubtedly perceived that attributes native

to the person were primary determinants of social reinforcer

effectiveness. Research in this area attempted to establish

which social reinforcer or set of reinforcers modified

behavior most successfully, given an array of particular

characteristics. Categories reviewed are (1) Personality,

(2) Race, (3) Social Class, (4) Intelligence, and (5) Age.


Personality

The many dimensions of personality have been written

about elsewhere (Hall and Lindzey, 1957). A few of these

dimensions are dependence and independence, high and low

self-concept, positive and negative self-image, intro-

version and extraversion, and internal and external locus

of control. Placement along a continuum from one extreme

to another was theorized to govern one's need for affection,

reinforcement, and the like. It is seen that personality

dimensions can in part determine the need system of an






-18-

individual which may relate to a need for particular kinds

of feedback from others..

Crucial inference arose out of the social reinforce-

ment research on personality. Personality, as a construct,

was thought to determine the likelihood that a specific

reinforcer affected behavior. That is, if the social

reinforcer matched the needs of the personality, then it

was likely to have a positive effect on behavior. If,

however, the feedback created cognitive dissonance, improve-

ment was less likely to occur. Thus, if the content,

intensity and timing of the input was discrepant from the

individual's needs at the moment it would be discredited or

noninfluential (Baron, Bass, and Vietzl, 1971; Todd and

Nakamura, 1970).

Race

Family patterns and reinforcement styles may go hand

in hand. Whereas there might be commonality among races,

it seems evident that racial heritage exerts a strong

influence on the development of family patterns (Hess and

Shipman, 1965).

Baron, Heckenmueller, and Shultz (1971) examined black

and white responsiveness to reinforcement styles by comparing

the efficacy of two kinds of praise on performance:

task oriented and person oriented. Task oriented feedback

made reference to the quality of the task performed while

person oriented commented on the "goodness" of the person

performing.






-19-
In their study, blacks and whites responded differentially

to the two forms of praise. Blacks given person oriented

approval performed significantly better on the assigned task

than those given task oriented feedback. Just the reverse

was true for the white subjects: task oriented was more

successful than person oriented. Although this study did

not provide evidence to assume racial differences exist in

family reinforcement patterns, it did support the notion

that blacks and whites differ in their responsiveness to

two forms of verbal feedback.


Social class

Social class has been found to influence health,

intelligence, and personality. Thus, investigators have

assumed there might be some relationship between social

class and one's susceptibility to social reinforcement.

Marshall (1969) explored the effect of delayed and

immediate feedback on the task performance of one hundred

and sixty high and low class subjects. Subjects from both

social classes learned equally as well with immediate

verbal reinforcement. Low class subjects performed better

with immediate than with delayed feedback while the high

class subjects performed equally as well with either.

Reiner (1974) conducted a comparative study of different

reinforcers, social, material, or knowledge of results, on

middle and lower class children. Subjects performed problem

solving and motor learning tasks in each reinforcement





-20-


condition. A dynamic interaction existed between social

class, type of reinforcement, type of task, and grade

level. Reiner concluded that the most potent reinforcers

for particular types of children on specific kinds of

tasks can be determined. These results stimulate thought

about implications for training programs, individualized

educational planning, and exposure to different reinforcement

conditions.


Intelligence

Level of intelligence has been a discriminating factor

in education since Binet developed his scales. In fact,

it is common practice today for placement into a variety

of regular and special education classes to be determined

primarily by one's measured intelligence on group or

individually administered tests. With an objective criteria

for placement established, it seems reasonable for educators

to search for motivational strategies for groups of children

so readily identifiable.

Van de Riet (1963) attempted to find just such a

strategy by researching the effectiveness of praise and

blame on the performance of educationally retarded and non-

retarded children. Blame was found to be significantly more

effective with retarded children while praise worked best

with non-retarded children. Moreover, clinical observation

suggested that praise produced anxiety in the retarded

subjects. It might be concluded that blame created less






-21-

cognitive dissonance and therefore became a more effective

incentive for retarded subjects than praise.

Prominent researchers in the area of social reinforce-

ment and the intellectually retarded, Panda and Lynch (1972),

established strict guidelines for the choice of appropriate

feedback used with retarded pupils in educational settings.

Criteria were that the social reinforcers (1) not be

redundant, (2) be contingent upon behavior, and (3) be

reliably structured. Non-contingent occurrence of feedback

was thought to inhibit performance. Thus, while verbal

support may have been a necessary condition for learning,

its effectiveness was enhanced by the contingent presentation

of diverse social reinforcers which provided information

about the quality of performance.


Age

Much has been written about the intellectual, physical,

emotional and moral development of children (Mussen, 1973).

Developmental psychologists have extensively described the

behavior patterns inimical to particular age stages in a

child's life. A little known area of developmental research

examined changes in perceptions of the meaning of social

reinforcers from childhood to adulthood.

Solomon and Ali (1972) surveyed individuals from

childhood to adulthood on their use of either content or

intonation as the primary determinant of the meaning of a

social reinforcer. Their data yielded distinct age trends.






-22-
Content seemed more important for younger subjects,

whereas intonation was fundamental in the adult subjects'

evaluation of the meaning of the reinforcing stimuli.


Summary

To summarize the section on subject characteristics,

the following tentative conclusions are offered. (1) For

social reinforcement to be effective its content, intensity,

and timing must be appropriate to the momentary needs of

the individual. (2) Racial differences in responsiveness

to some kinds of verbal feedback may exist. (3) Criteria

for the selection of social reinforcers are that the

reinforcer not be redundant but be contingent on behavior

and reliably structured. (4) Developmental trends appear

in perceptions of the meaning of social reinforcers. It

is inferred that the relationship between the subject and

the effectiveness of a social reinforcer is a complex one

whose determinants seem to vary from moment to moment.


Experimenter Effects

The now famous Rosenthal studies examined the influence

of the experimenter on the results of psychological and

behavioral research (Rosenthal, 1966). Several conclusions

can be stated regarding experimenter effect on subject

behavior. These may have implications for a study in which

an adult interacts with a child as the treatment variable.

The conclusions are as follows: (1) Subjects extract

information from the investigator's attributes in order to






-23-


respond in their preconceived "right" manner; (2) The

experimenter as a person and what has happened to him

before or happens during the experiment affects the subject's

responses. The subject's responses subsequently affect

the experimenter's behavior which in turn can alter the

subject's response; (3) Characteristics of the experimenter

influence the subject's behavior in a direct way but sometimes

interact in a subtle fashion with subject characteristics,

task characteristics, or situational characteristics; and,

(4) The principal investigator must rely on the process of

replication in order to control for experimenter effects.

In a unique application of Rosenthal's work, Leventhal

and Fischer (1970) attempted to determine whether differences

in performance could be attributable to characteristics of

the experimenter per se or to his reinforcing statements.

Ninety-six subjects were exposed to experimental conditions

in which the experimenter was seated near and away from

the subject and did not reinforce, or in which he was seated

near the subject but did reinforce. Dependent variables

were rate and preference of placing marbles in holes. The

results showed that the subject's rate differed even before

reinforcement was initiated.

Therefore, it was concluded that reinforcement itself

was not solely responsible for the condition differences.






"24-

It would appear that preference changes
reflected the subject's effort to extract
information from the experimenter's
behavior, while base rates and rate changes
reflected complex alterations in the subject's
emotional state. . It is difficult not to
express some surprise at the unimportance of
reinforcement. (p. 92)

The authors qualify their conclusions by stating that

"perhaps reinforcement effects would be noticeable if

alterations were made in the procedure and experimental

design"(p. 92). The remainder of this review discusses

studies in which reinforcement did make a difference.


Informative and Affective Roles

Social reinforcement has been said to serve two purposes:

to inform about response accuracy and to create a positive

feeling (Panda, 1971). A positive feeling is created when

a person is made to feel good in order to be confident of

making an appropriate choice. Response accuracy occurs

when a person is informed whether their choice was accurate

in order to continue successful choices. Regardless of

divergency of purpose, the objective of both purposes of

reinforcement is to motivate the subject to perform with

greater accuracy.


Informative role

Information theory states that when motivating children

the cueing and signalling aspects of reinforcement are

more important than any feelings aroused. Investigators

have compared the effectiveness of transmitting correct,





-25-


incorrect, or no information to subjects after subject

responses (Siegel and VanCara, 1971; Meyer and Offenback,

1962; Spence, 1966).

Typically information is given.either verbally or with

non-social cues, like marbles, subsequent to each task

response. Experimenters divided subjects into three groups

labeled "Rb," Wb," and"RW." In the "Rb" group subjects

were informed when they were right but not when they were

wrong. In the "Wb" condition only the wrong answers were

consequated. In the "RW" group subjects were informed of

both correct and incorrect responses.

Probably the most important educational application

of information theory was a study by Panda (1970) in which

eighty educable mentally retarded boys assigned to one of

two conditions were tested on a concept learning task.

Subjects received either RW information or RW plus non-

contingent verbal support after each block of four trials.

RW subjects performed with significantly higher accuracy

than did those who received RW information plus verbal

support. Unconditional support seemed to serve as a

distractor. Therefore, it appears that a reinforcer intended

for one purpose, to make the subject feel good, may be

counterproductive for another purpose, to improve the

subject's performance. However, because non-contingent verbal

support was found to be distracting does not imply that the

same interpretation applies to contingent verbal support.





-26-


Affective role

Affective feedback was intended to create either a

positive or negative feeling in the child. In turn the

feeling was thought to motivate the child to perform at a

faster rate or higher accuracy. Drive reduction theory

provides a frame of reference for understanding the role

of affective feedback. Within such a framework, the subject

performs to achieve a positive feeling or to reduce the

presence of a negative feeling. Thus, instead of providing

information the intent was to support or criticize the

person so that performance would improve. Studies, then,

attempted to determine whether praise or blame worked best

for a particular type of subject or task (Binet and

Vaschide, 1897; Leith and Davis, 1969; Kelly and Stevens,

1964; Hurlock, 1924).

Relationship Between the Response and Reinforcement Delivery

Developments thus far reviewed have conceptualized the

nature of reinforcement on a theoretical basis. Inves-

tigators attempted to explain why social reinforcers had

a reinforcing effect on behavior. This section attends to

the empirical conceptualization. Empiricism concerns itself

primarily with the effect. There is considerable emphasis

to determine what environmental event may have precipitated

the effect, but only minimal interest in understanding why

(Sidman, 1960).





-27-



The shift in conceptualization from theoretical to

empirical has correlated with methodological modifications

in research strategy. There was a movement away from

pre/post-like experimental designs to designs which

prioritized the control of a phenomena. Although many

studies exist which utilized group statistical designs,

this section discusses only those which used single

subject designs.

Two basic methods of manipulating social reinforcement

in order to investigate the relationship between the

response and reinforcement delivery will be discussed.

These methods were the presentation and withdrawal technique

and the more complex schedule of reinforcement paradigm.

Presentation and withdrawal

The presentation and withdrawal technique has assumed

some prominence in both research and applied settings.

It is described here as it relates to the modification of

social and academic behavior in the classroom. First,

the research technique will be elaborated upon. Then, a

group of studies which clearly exemplifies the practice

will be examined.

A single subject design, which is a one-to-one

correspondence between the emission of target behavior

and reinforcement, in addition to the baseline technique,

characterized this research strategy. An idiographic model

allowed the investigator to look at the effect of treatment





-28-


variables on the individual subject. A study typically

included more than one subject but results were discussed

relative to each individual's responsivity. Social rein-

forcement was either delivered for each occurrence of the

desired behavior and/or withheld if an undesired behavior

occurred.

A general notion of the technique used to demonstrate

reliable control of a behavioral change will be set forth.

First, data was collected on the frequency of the target

behavior over a period of time establishing a baseline

level of its occurrence against which treatment was applied

and its effect recorded. Third, the original baseline

condition was reintroduced meaning that treatment was with-

drawn. If the final frequency of behavior approximates that

of the original baseline, it is inferred that the experimental

treatment precipitates the effect observed in the treatment

phase. Within this framework, researchers conducted studies

in the classroom.

The modification of aberrant social behaviors has

increased classroom harmony and served to make children

more acceptable to each other and adults (Thomas, Nielsen,

Kuypers, and Becker, 1968). Several studies set as their

goal the elimination of disruptive behavior and the increased

production of on-task performance (Wasik, Senn, Welch,

and Cooper, 1969; Thomas, Becker, and Armstrong, 1968;

Broden, Bruce, Michell, Carter, and Hall, 1970; Hall,

Lund, and Jackson, 1968). Idiosyncratic problem behaviors





-29-


of children such as crying (Allen, Hart, Buell, Harris,

and Wolf, 1964), isolate behavior (Harris, Wolf, and

Baer, 1967), passive play (Hart, Allen, Buell, Harris,

and Wolf, 1964), and non-cooperative play (Redd, 1969) have

been successfully treated as well.

Academic behavior modification has received far less

research attention than the modification of social behavior.

One notable exception, though by no means the only one, a

study by Hasazi and Hasazi (1972), investigated the digit

reversal behavior of an elementary school child. Baseline

observation found that the teacher responded to reversals

and ignored correct digit placement. Not atypically, she

marked reversals incorrect and gave "extra help" on all

sums incorrectly written. For the treatment the teacher

was instructed to socially reinforce correct digit placement

and ignore reversals. The results showed a sharp decrease.

It is concluded from the research on the presentation

and/or withdrawal of social reinforcement that the technique

represents a powerful tool for modifying behavior as well

as a viable research paradigm for the demonstration of

the control of social and academic behavior.


Schedules of social reinforcement

In contrast to simply presenting or withdrawing social

reinforcement contingent upon the occurrence of a behavior,

schedules represent an intricate way of varying the

presentation of reinforcement in relation to the response.







-30-


As previously defined, a schedule is a systematic arrange-

ment of response-reinforcement contingencies which vary the

proportion of reinforced responses or time intervals after

which reinforcement can be obtained. The resulting effect

of such manipulation has been called the patterning effect.

Due to methodological problems indigenous to such

research, there have been few studies (Lahey and Drabman,

1974; Staats et al., 1962; Chadwick and Day, 1971). The

next section provides an overview of the broader topic,

schedule research on reading. This topic includes but is

not restricted to a survey of token schedule research and

a discussion of the procedural application of such studies

to one which investigates social reinforcement schedule

effects. Such review provides a framework for resolving

methodological problems.


Summary of Historical Developments

The historical developments reviewed included satiation

and deprivation effects, subject experimenter influence,

characteristics, informative and affective roles of social

reinforcement, and the relationship between the response

and the delivery of reinforcement. Each of these develop-

ments was presented within the context of their differing

views of the nature of reinforcement. Attention was paid to

the methodological and procedural components of studies

which investigated those views. The next section integrates





-SI-

schedule research of reading with the procedural requirements

of one which uses social reinforcement.

Schedule Research on Reading

Studies which manipulated token and social reinforce-

ment schedules will be discussed in view of the application

of methodological characteristics pertinent to the present

investigation.


Schedules of Token Reinforcement

Token reinforcement is generally used within a system

called the token economy (Ayllon and Azrin, 1968). The

token economy refers to an economic system of payment

for appropriate behavior in which the tokens earned can be

saved and later exchanged for valuables. Tokens, then,

can be exchanged for objects of value much the same as

currency is exchanged for goods in our society. Their worth

is enhanced by their relevance to real currency, by the

fact that the subject can manipulate and transport them, and

by their readily understandable value.

One particular characteristic of tokens make their

use as a reinforcer advantageous in research (O'Leary and

Drabman, 1971). The delivery of the token is a discrete

event which has a definite effect on the environment.

Therefore, unlike food and social reinforcement, it can be

reliably dispensed and counted.

While Staats and his associates have conducted by

far the most schedule research on reading behavior (Staats,






-32-

Staats, Schutz and Wolf, 1962; Staats, Finley, Minke,

and Wolf, 1964; Staats, Minke, Finley, Wolf, and Brooks,

1964) other researchers have made significant contributions

as well. A few studies that have a direct bearing on

the present investigation will be presented.

Lovitt,.Eaton, Kirkwood, and Pelander (1971) investi-

gated the effects of four reinforcement contingencies

on oral reading. Subjects responded to the contingencies

while reading from an experimental reader. Results were

compared with reading rates of the same subjects from

control readers.

The experimental contingencies were designated as

follows. In Experiment I if the pupil's correct rate

exceeded and error rate was less than prescribed limits

he was reinforced on a 25:1 ratio. For every two errors in

Experiment II one token was taken away. In Experiment III

for every correctly read word the pupil received one token.

In Experiment IV the pupil earned one token for every

thirtieth correct response. No tokens were delivered when

the subjects were reading from the control readers.

Procedurally, the pupil read from each book five

minutes per day. The order in which he read the books was

randomly alternated. Errors were handled the same through-

out all Experimental groups. That is, if the pupil mis-

pronounced a word, he was immediately told the word. Tokens

could be exchanged for minutes of free time at the end of

the ten minutes of reading from the control and experimental

readers.






-33-

The results are summarized as follows. In Experiment I,

where the contingency affected both correct and error

responses, both types of responses were influenced. In

Experiments II and III, the contingencies focused on errors,

and errors were more influenced than were correct responses.

In Experiment IV where the contingency was placed on correct

responses, correct responses were influenced in the first

half of the experiment but not in the latter half. In

most cases, both correct and error rates in the control

reader were not as positively influenced as in the experi-

mental reader.

Sibley (1966) investigated reading rate and accuracy

as a function of Fixed Ratio schedules of token reinforce-

ment. Seven subjects of normal intelligence who were

designated as poor readers participated in the experiment.

There were three phases in the experiment. In Phase I

non-exchangeable points were delivered on a Fixed Ratio for

correct reading responses. In Phase II non-exchangeable

points and tokens were delivered simultaneously on a

gradually increasing Fixed Ratio schedule. Tokens were

exchangeable for back-up reinforcers. The final phase was

a return to Phase I conditions.

Points were not found to be initially reinforcing, but

became strikingly so subsequent to being paired with tokens.

Four of the seven subjects showed increased reading rates

as the FR was increased. Three showed the reverse trend.





S-34-


Some evidence was found to indicate that high ability

subjects read faster on the higher Fixed Ratios.

Hoeltzel (1973) investigated the effects of single and

multiple-ratio schedules of token reinforcement on reading

rates. Like the Lovitt et al. (1971) study the Experimental

and Control reader approach was used. Four subjects read

five minutes per day from each book. There were no

contingent phases with the Control reader but with the

Experimental reader contingent and non-contingent phases

were alternated weekly for seven weeks. Generally, the

results indicated increased correct responding during

contingent phases of the Experimental reader as well as an

overall tendency toward increased error acceleration. These

results generalized to the Control reader only when the

contingent phase was in effect with the Experimental reader.

It is seen from the review of token schedule studies

that results generally support the notion regarding

patterning effect of schedules. That is, schedules produce

particular effects which can be observable across reinforcers,

behaviors, and subjects.


Schedules of Social Reinforcement

While there have been few studies which investigated

social reinforcement schedule effects, the Mays (1970)

study used social reinforcement schedules to determine

their influence on error production. This investigation

looked at the effect of schedules of reinforcement on the








elimination of errors in oral reading. Social reinforcement,

"Fine" and "Good," was delivered on either a VI or VR

schedule. The VI schedule was programmed by tape and the

VR by penciled marks in the book being read.

The response unit was a sentence and reinforcement

was delivered if the entire sentence was correct. Although

correct sentences were reinforced, the dependent variables

in the Mays study were rate of error production and general

error strength. Therefore, there was no attempt to

analyze the direct influence of the schedule of social

reinforcement on the response being reinforced.


Procedural Requirements of a Study of Social Reinforcement
Schedule Effects on Reading

The application of procedures from studies of token

reinforcement schedule effects to a tutoring situation in

which social reinforcement is used is complex. This section

will discuss three pertinent procedural aspects which need

to be ascertained before such application can begin. These

three procedural needs are (1) the importance of the

discrete delivery of the reinforcer, (2) clarification of

the nature of the reading response, and (3) the need to

methodologically control the learning-to-learn phenomenon.

The application of these remarks will be presented in

Chapter III.





-.56- '


Importance of the discrete delivery of the reinforcer

Ayllon and Azrin (1968) stated that the following rule

should govern the selection of a conditioned reinforced.

It should "provide a distinctive and tangible stimulus

event to bridge any delay between the desired response and

the delivery of the reinforcer"(p. 77).

Their work, The Token Economy, outlines the advantages

of tokens as conditioned reinforcers. These advantages

are (1) delivery is quantitatively related to the amount

of reinforcement, (2) portability, (3) possession in

infinite amounts, (4) usefulness to operate machines,

(5) durability and tangibility, (6) ease of standardization

of physical characteristics, (7) indestructibility,

(8) non-duplicability, (9) reinforcement of response at any

time, and (10) reinforcement occurrence without interruption.

Social reinforcement possesses some of these advantages

but also has some critical disadvantages. The advantages

are (1) it represents an immediate stimulus following

a response, (2) does not disrupt performance, and (3) allows

the response to be strengthened. The disadvantages include

the following: (1) it is a momentary and, therefore,

intangible event, (2) the qualitative dimensions do not

allow for standardization, and (3) it cannot be used to

operate machines.

Nature of the reading response

There are many definitions of reading. The definitions

range from reading as comprehension to reading as decoding







-37-

(Singer and Ruddell, 1970). Some experts maintain that a

child has not read a word until he has comprehended it.

Others define reading as the act of saying or decoding a

written word. The proponents of the latter definition

partial out the decoding and comprehension components of

reading as separate behaviors. Decoding and the compre-

hension of what was decoded are considered to be two distinct

behaviors. The present investigation embraces this concept

of reading.

It was thought to be critical to the intent of the

present research to narrow the response datum to the

frequency of words read in time. Sidman wrote in Tactics

of Scientific Research (1960) that the response should be

one "whose relevant characteristics are easily recordable,

with minimal interference from the process of observation

itself. [Additionally]. . there should be minimal con-

straint upon the subject's rate of responding" (p. 395).

The repertoire necessary in order that the child can

decode is complex. Words can be counted as discrete

responses in time, but words are not equal in complexity.

Each word decoded represents a lengthy chain of responses.

The differing complexity of the words as units and as they

fit.in the context of the sentence and paragraph make for

possible variability in resulting data. The response can be

recorded cumulatively in time but each is not necessarily

an instantaneous unitary response (Staats, Finley, Minke,






-38-

and Wolf, 1964). Therefore, while the limitation of the

response datum to counting words read simplifies data

collection, the attendant experimental results may not be

as clear were the response a simple bar press.


Control of the learning-to-learn phenomenon

The previous section implied that the nature of the

reading response induces inherent variability and thus

confusion in the data. An additional complication is the

need to continually increase the complexity of the reading

material as skillevel improves. Control of this learning-

to-learn phenomena is important in order that variability

in the data be due to the experimental manipulations and

not to materials or skill level.

Two methods have been devised to assist in controlling

for this variability. The first has to do with the selection

of materials and the second has to do with the presentation

of the materials to the subject. These two methods will be

elaborated in the remainder of this section.

Material selection. Staats, Finley, Minke, and Wolf

(1964) wrote that in order to study experimental effects on

long-term reading, materials were needed which do not affect

the learning curve. The ideal material would be homogeneous

in content, such as programmed reading materials. For this

study the Science Research Associates Laboratory la (SRA

Lab la) reading materials were chosen (Parker and Scannell, 1961).





-39-


Performance Determined Instruction Model (PDI) was

developed by Gray, Baker, and Stancyk (1969) at the

Monterey Institute for Speech and Hearing as an instructional

procedure for obtaining changes in reading performance.

The methodology combines aspects of binary logic, laws of

learning, and programmed instruction. The PDI model was

designed as a method for decision making to discriminate

which procedures to employ with the individual subject.

A host of procedures was developed by the Institute

to help the child read at the desired reading level. One

procedure, related to the acquisition of sight word voca-

bulary, is described here. The application of this procedure

to this investigation will be presented in Chapter III.

This precision instruction approach makes several

assumptions regarding the nature of reading which relate to

the development of the training procedures. First, reading

is considered to be a learning task and, therefore, subject

to the laws of learning. Second, decoding the word is

antecedent to comprehension and therefore should be

taught first. Third, a training system should provide fre-

quent and immediate reward for successful performance.

Fourth, characteristics such as visual perception, intelligence,

visual motor coordination, visual and auditory sequencing

and the like are considered not to affect differentially the

prognosis for skill mastery unless the deficiency is of

such a marked and gross degree that it results in reduced

functional ability in nonreading activities.





-40-


The procedure for the acquisition of sight vocabulary

incorporates a computer logic system known as TOTE (Test-

Operation-Test-Exit) which was originally conceived by

Miller, Galanter, and Pribram (1960). Basicly the steps

within a TOTE system occur similarly to the following

sequence. First, the behavior in question is measured against

some standard. Second, if the behavior is discrepant to

some specified degree from the standard an operation is

performed to help the behavior reach the goal. Third, the

behavior is tested again. If the behavior reached the

standard, the program is terminated. If, however, the

behavior did not reach the standard, the behavior is treated.

The TOTE system provides a structure for decision making.

More specifically the TOTE system was applied in the

PDI model to improve sight word vocabulary performance.

This application will be described in some further detail.

A target performance of 92 percent accuracy plus or

minus 2 percent was set as the standard against which to

measure performance. Performance was measured against the

standard and operations were chosen if the performance was

either above or below the accuracy range. Questions were

asked and based on the answer decisions were made regarding

whether to terminate, to reduce task difficulty, or to

increase task difficulty.

The following questions or steps were devised to

control the performance accuracy of sight word vocabulary

performance. First, was the performance accuracy within





-41-



the range prescribed for three successive sessions? If

the answer was yes, Question 2 was asked. Was the reading

age level up to the target age level? If yes, training

was terminated. If the answer to Question 1 was no,

Question 3 was asked. Was the performance above the range;

that is, were three successive sessions above 94 percent?

If the answer was yes, then, the subject was advanced to

the next level of difficulty. If the answer to Question 3

was no; that is, three successive sessions were below

90 percent, then a decision was made to move to the level

on which performance was last within the acceptable

accuracy range.

The next chapter will describe how the TOTE system

and PDI model for sight vocabulary building was applied in

this study. As indicated, it was considered necessary to

develop procedures which would control the learning-to-learn

phenomenon. Procedures maintaining accuracy level within

a particular range were necessary to control variability in

performance to such a degree that treatment effects might

be more readily observable. It was important to describe

the procedures here in detail as a frame of reference

for understanding the unique procedures designed for the

current study.













CHAPTER III
METHODS AND PROCEDURES

The purpose of this chapter is to present those methods and

procedures selected in order to investigate whether there is

a predictable relationship between Multiple schedules of

social reinforcement and subjects' mean rate of responses

on Flash Card and Oral Reading tasks. Chapter I defined a

Multiple schedule as one in which two or more simple schedules

randomly alternated, and in which the occurrence of the

simple schedules was accompanied by a stimulus. This study

used tutors as the stimuli associated with the different

simple schedules.

Three experiments were designed to investigate the

relationship between Multiple schedules and rate of response.

In each experiment the Multiple schedule consisted of two

simple schedules. Comparisons were made between the data

derived from the subjects' rate of response in the presence

of the two simple schedules. Experimental procedures will be

discussed in more detail in the final section of this chapter.

The contents of Chapter III will be presented in the

following sequence: (1) Research hypotheses, (2) Subjects,

(3) Tutors, (4) Remedial Reading program, (5) Setting,

(6) Apparatus, (7) Materials, (8) Data Collection, (9) Data

for Data Analysis, and (10) Procedures. This latter section


-42-







-43-

includes those procedures common to all three experiments.

These procedures are (1) Flash card procedures, (2) Oral

reading procedures, (3) Adaptation of the Performance

Determined Instruction model, and (4) Point reinforcement

procedures. Finally, the chapter concludes with an overview

of the experimental procedures unique to each experiment.

Research Hypotheses

A brief introduction to each experiment will be followed

by a list of research hypotheses. The introduction will

include the nature of the two simple schedules compared

and a description of the experimental design used in each

experiment. This information should provide a framework in

which to understand the research hypotheses. Like the Staats

(Staats, Finley, Minke, and Wolf, 1964) study reviewed in

Chapter I, the research hypotheses will state which schedule

is predicted to generate the higher rate of response.

Experiment 1

Experiment I examined Multiple schedules of social

reinforcement in which both simple schedules reinforced

correct responses. Three subexperiments make up Experiment I.

There were two experimental phases in each subexperiment

with approximately 16 sessions in each phase. The simple

schedules, a Variable-Ratio (VR) schedule and a Fixed-

Interval (FI) schedule, assigned to tutors in Phase I were

switched in Phase II.








-44-


Research hypotheses

This experiment tested six hypotheses relevant to

Multiple schedule effects on Oral Reading and Flash Card

responses.

1. Subexperiment I-1 Flash Card Phase I and II

mean VR correct rates will be higher than the

mean FI rates.

2. Subexperiment I-1 Oral Reading Phase I and II

mean VR correct rates will be higher than the

mean FI correct rates.

3. Subexperiment 1-2 Flash Card Phase I and II mean

VR correct rates will be higher than the mean

FI correct rates.

4. Subexperiment 1-2 Oral Reading Phase I and

II mean VR correct rates will be higher than the

mean FI correct rates.

5. Subexperiments 1-3 Flash Card Phase I and II

mean VR correct rates will be higher than the

mean FI correct rates.

6. Subexperiment 1-3 Oral Reading Phase I and II

mean VR correct rates will be higher than the

mean FI correct rates.










Experiment II

A contrast design is used in Experiment II to

demonstrate the effects of Multiple schedules of social

reinforcement. In this two phase experiment, the following

contrast design is employed: (1) the two Phase I simple

schedules are identical and (2) one of the simple schedules

is changed in Phase II. This change is said to produce a

contrast effect defined in Chapter I. Phase I simple

schedules were both a Variable-Interval (VI) schedule. One

of these schedules was changed in Phase II to a VR schedule.

The tutor with the lower average rate in Phase I changed

to a VR schedule in Phase II.

Research hypotheses

This experiment tested four hypotheses relevant to

Multiple schedule effects on Flash Card and Oral Reading

responses.

1. The median of the differences between correct

rates of the two schedules in Phase I for
Flash Card and Oral Reading is zero (p = .05).

This null hypothesis will be tested by the sign

test (Siegel, 1956).






-46-

2. Experiment II Phase II mean VR rate will be higher

than the mean VI rate for Flash Card and Oral

Reading.

3. Experiment II Phase II mean VR rate will be higher

than the mean VI rate for that tutor in Phase I

for Flash Card and Oral Reading.

4. Experiment II Phase II mean VI rate for the

unchanged tutor will be lower than the mean VI

rate for that tutor in Phase I for Flash Card and

Oral Reading.


Experiment III

Experiment III investigated the effects of Multiple

schedules of social reinforcement in which both simple

schedules were the same but one reinforced correct responses

and one reinforced error responses. Two subexperiments make

up Experiment III. The simple schedules used in

Subexperiment III-1 were both an FI schedule. In Sub-

experiment III-2 both schedules were a VI schedule. Each

subexperiment was divided into two experimental phases.

If assigned correct responses to reinforce in Phase I the tutor

reinforced errors in Phase II and vice versa.


Research hypotheses

This experiment tested nine hypotheses relevant to

Multiple schedule effects on Oral Reading and Flash Card

responses.






-47-

1. Subexperiment III-1 Flash Card Phase I and II

mean correct rates for the tutor that reinforced

corrects (RC) will be higher than the mean correct

rates for the tutor that reinforced errors (RE).

2. Subexperiment III-1 Flash Card Phase I and II

mean RE error rates will be higher than the mean

RC error rate.

3. Subexperiment III-1 Oral Reading Phase I and II

mean RC correct rates will be higher than the

mean RE correct rates.

4. Subexperiment III-I Oral Reading Phase I and II

mean RE error rates will be higher than the

mean RC error rates.

5. Subexperiment III-2 Flash Card Phase I and II

mean correct rates in the RC component will be

higher than the mean correct rates in the RE

component.

6. Subexperiment III-2 Flash Card Phase I and II

mean error rates in the RE component will be

higher than the mean error rates in the RC

component.

7. Subexperiment III-2 Oral Reading Phase I and II

mean RC correct rates will be higher than the

mean RE correct rates.

8. Subexperiment III-2 Oral Reading Phase I and II

mean RE error rates will be higher than the

mean RC error rates.









Subjects

Four eight-year-old third grade pupils from an

elementary school in Gainesville, Florida served as subjects.

All subjects were black, two were female and two were male.

The four subjects who participated in the study were

selected from a pool of 60 subjects. Two third grade

teachers were asked to select pupils from their classes who

in their estimation were able to read at the mid-first grade

level. The teachers picked seven pupils. This number was

reduced to the final four necessary for the study. In order

to be selected, the following criteria had to be met:

(1) a grade level performance between 1.3 and 1.6 on the

Lee-Clark Reading Test, and (2) 90 to 95 percent oral

reading accuracy on a 1.5 grade level SRA Lab la story. Each

of the final four met these criteria.

Tutors

Seven undergraduate University of Florida students

served as tutors. Two were male and five were female.

The students received university course credit for their

participation in the study.

Tutors took part in a ten hour training program prior

to the beginning of the experiment proper. The program

included a didactic overview of methods and procedures,

supervised practice of the tutoring of other tutors and







-49-


of the tutoring of nonexperimental pupils. Practice

administering the procedures described in the latter portion

of this chapter was supervised by the experimenter.

Remedial Reading Program

The subjects attended a remedial reading program

directed by the experimenter. Daily hour long sessions

were divided into 40 minutes of task performance and 20

minutes of play. The tasks were administered in the

following sequence: flash card drill, oral reading, written

comprehension exercises, and phonetic drill. Points

earned during these tasks could be exchanged after forty

minutes had elapsed. The flash card and oral reading

performance provide the data for this study.


Setting

The setting was a room in Kirby-Smith Elementary

which had been divided into seven tutoring stations. These

were separated from each other by plywood partitions. In

the remainder of the room were a play area and a materials

center. Activities available in the play area included

games, art materials, and comic books.

Apparatus

Apparatuses included timing devices and a point

dispenser. Each tutoring station was equipped with either

a standard stopwatch or an electromechanical timing device.








-50-

These were used to clock task duration precisely and to

aid in delivering points and social reinforcers at the

appropriate times. The point dispenser was a self-inking

rubber stamp which fit into the eraser end of a pencil.

Points were placed on Point sheets specially designed for

this study. An example of a Point sheet is shown in

Appendix A.


Materials

Oral reading and flash card supplies were the

materials used for this study. Reading selections from

the Science Research Associates Reading Laboratory la

(SRA Lab la) served as the oral reading materials. SRA

Lab la contained seven levels of reading difficulty with

20 stories per level. The levels ranged from the 1.2

grade level to the 3.0 grade level. A flash card unit

was made from words appearing in a particular level of

stories. Each word in the reading level was represented

once in its corresponding flash card unit.


Data Collection

The Daily Data Sheet (DDS) was designed to allow for

the collection of data during the task performance. Space

was provided for the tutors to note the time and also to

note correct and error responses as they occurred during

each task. Thus, the data collected could be easily








-51-

corrected into the frequency of responses in time or rate

of responding. An example of the DDS is shown in Appendix B.


Data for Data Analysis

The basic datum in this study was rate of responding.

Rate was determined by dividing the frequency of correct

or error by the duration of the task. These rates were

recorded daily on the Standard Behavior Chart .(SBC) and

these raw data are shown in Appendix C. The results are

expressed in terms_of the average rates of response per

schedule per phase. Since these data are ratio data,

the averages used are geometric means.

Procedures

Procedures common to the three experiments will be

described in this section. These procedures are (1) Flash

card procedures, (2) Oral reading procedures, (3) Adaptation

of the PDI model, and (4) Point reinforcement procedures.


Flash Card Procedures

The tutor administered the flash card unit which

corresponded with the oral reading level assigned for that

day. Flash cards were administered for ten minutes. The

following standard procedures were developed. If the word

on the card was correctly read within three seconds, the

tutor placed the card in the correct pile and presented










the next card. If, however, the word was not read within

three seconds or was read incorrectly, the tutor pro-

nounced the word for the subject to repeat. The card

was, then, placed in the error pile. When five of these

error cards had accumulated, they were shuffled and represented.

The above procedures were in effect until no more than

two error cards remained. A new card was then presented.

Oral Reading Procedures

The subject read one reading selection orally in its

entirety per session. Error responses were treated in

a manner similar to those which occurred during the flash

card performance. That is, if the subject did not read

the word within three seconds or read the word incorrectly,

the tutor pronounced the word for the subject to repeat.

Each error was noted on the DDS as it occurred. Total

number of correct responses was calculated by subtracting

the total number of errors from the total number of words

in the story.


Adaptation of the PDI Model

The rationale for the utilization of the PDI model

was discussed in Chapter II. This section explicates

those procedures which were adapted from the PDI model

for the oral reading task of this study.








-53-

Each 20 story level in SRA Lab la was divided into

four five-story units. Stories one through five, six

through ten, eleven through fifteen, and fifteen through

twenty made up the five units.

Like the PDI model, and accuracy level of 90 to 95

percent was established. A series of questions were asked

of the data in order to decide whether to increase or

decrease the level of difficulty of stories selected. If

the subject read the first five stories in a level at an

average of 90 to 95 percent accuracy, he continued to the

next unit. If, however, the subject read the first five

stories below an average of 90 percent accuracy, he re-read

those that he had read the least accurately until he had

attained an average of 90 to 95 percent for five consecutive

days. If the subject read the stories in a unit at an

average accuracy above 95 percent, the next unit was skipped

unless the unit to be skipped was the first unit of a level.

These procedures did not apply to the flash card

task. As long as the subject read stories at a particular

level, he responded to the flash cards which corresponded

to that level.







-54-

Point Reinforcement Procedures

Point procedures included both their delivery and

exchange. All subjects received points on a Fixed Interval

90 second schedule. That is, during each task of the 40

minute work session subjects received a point every 90

seconds immediately after the first correct response.

Times for point delivery were listed in the upper left hand

corner of the DDS. Point exchange occurred at the end of

the session. The points could be spent for minutes of

time in the free play area, for pieces of candy, or for

polaroid pictures. Exchange ratios were 1 point per 1 minute

of free play, 5 points per piece of candy, and 10 points

per polaroid picture. Subjects could elect to save

points. They were allowed to save any number of points

and were allowed to buy no more than 15 minutes of free

play, 2 pieces of candy, and 1 polaroid picture per day.

Experimental Procedures

Three experiments were designed to demonstrate the

effects of Multiple schedules of social reinforcement on

rate of correct and/or error Flash Card and Oral Reading

responses. The previous section described those procedures

which were common to each experiment. This section describes

the social reinforcement procedures unique to each experiment.






-55-

Social Reinforcement Procedures

Descriptions of the kinds of verbal statements used

as social reinforcers, the delivery procedures, and the

monitoring of their delivery is presented here.


Kinds of statements used with the schedules

Four groups of verbal statements were selected on

the basis of their face validity and grouped into categories

labeled High (H), Low (L), Consoling (C), and Disapproving (D).

The first two groups were used to consequate correct

responses and the latter to consequate errors. Samples of

these statements are shown in Table 1. Table 2 shows the

schedules and the kinds of statements used in each experiment.

Explanations of the schedules were provided in the Definition

of Terms section of Chapter 1.


Delivery procedures

Tutors delivered social reinforcement according to a

schedule. The schedule to which they were to adhere on a

particular day was written in the upper left hand corner

of the DDS. If the tutor was to reinforce on a time based

schedule, a list of the times when reinforcement was due

was provided. If, however, the tutor was to reinforce on a

response based schedule, the list contained the number of

responses per reinforcement. Space was provided on the DDS

for the tutors to note the occurrence of social reinforcers.








-56-


Table 1. Verbal Statements Used as Reinforcers


High (H)

Three in a row right
Great
What a good reader
Yeah, you got it
That's a girl (boy)
Hurray


Consoling (C)

That was close
Almost
Half right is better than nothing
Boy are these tough
Maybe next time
They've got to get easier
You'll get the next ones
There's another rough spot
Parts of this story are pretty rough
Better luck next time
Everyone makes mistakes


Low (L)

Good
Fine
Okay
That's right


Disapproving (D)

Wrong
Don't miss it again
Not again
I just told you that one
Try harder






-58-

Monitoring of social reinforcement delivery

The experimenter monitored social reinforcement

delivery by observing the tutors and giving them feedback.

Two of the tutors were selected at random and observed

daily: one during Flash Card and one during reading. A

determination was made of the expected number of reinforcers

to be delivered on a given task. Actual number of reinforce-

ments was, then, counted during the observation. The tutors

received written and oral feedback regarding the difference

between the actual frequency of reinforcers and the expected

number. The feedback was delayed until the end of the entire

work session in order that the tutor not know that he or she

had already been observed.


In summary, this chapter has provided a description

of the methods and procedures used in the investigation.

The remedial reading program as well as a description of

the sample and data collection process were provided. A

detailed description of the procedures common and unique to

each experiment was presented. The next chapter will

present the results of this investigation with respect to

the questions asked by the hypotheses outlined in this chapter.














CHAPTER IV
RESULTS

In this chapter the results of the study will be pre-

sented. The experiments and their respective subexperiments

are grouped in units. A brief introduction to each subexperiment

is followed by a presentation of its results. Included in

these results sections are statements of the acceptance or

rejection of the hypotheses relevant to each subexperiment..

Results are presented in tabular form in this chapter and

Standard Behavior Charts of the raw data are to be found in

Appendix C. The basic question being asked is whether there is

a predictable relationship between the Multiple schedule of

social reinforcement and the subject's mean rate of responses.

Table 3 provides an overview of the results which will be

presented in detail in the remainder of this chapter.



Experiment I

Experiment I examined Multiple schedules of the social

reinforcement of correct Flash Card and Oral Reading responses.

Three subexperiments were each comprised of two experimental

phases. The schedule components assigned to the tutors in

Phase I were switched in Phase II. The means of the correct

rates by component and phase are presented for Flash Card and

Oral Reading tasks. Error rate data was collected; but, it

is not reported here since this experiment manipulated correct

responses.
-59-






Table 3. Acceptance or Rejection of Hypotheses


Sub- Accept/
Experiment experiment No. Hypothesis Reject

I-1 1 Flash Card mean VR correct rate will be higher Reject
I than the mean FI correct rate in both Phases

2 Oral Reading mean VR correct rate will be Reject
higher than the mean FI correct rate in both
Phases

1-2 3 Flash Card mean VR correct rite will be higher Accept
than the mean FI correct rate in both Phases

4 Oral Reading mean VR correct rate will be higher Reject
than the mean FI correct rate in both Phases

1-3 5 Flash Card mean VR correct rate will be higher Reject
than the mean FI correct rate in both Phases

6 Oral Reading mean VR correct rate will be higher Reject
than the mean FI correct rate in both Phases

II 1 The median of the differences between correct Reject
rates of the two schedules in Phase I for
Flash Card and Oral Reading is zero (p=.05)

2 The Phase II mean VR rate will be higher than Reject
Sthe mean VI rate for Flash Card and Oral Reading

3 The Phase II mean VR rate will be higher than Accept
the mean VI rate in Phase I for that tutor for
Flash Card and Oral Reading

4 The Phase II mean VI rate for the unchanged tutor Accept
will be lower than the mean VI rate for that
tutor in Phase I









Subexperiment I-i

An eight-year-old female served.as the subject (S3).

Tutors 1 and 2 were two females. The Multiple schedule

components were VR3H and FI 10 minute L (FI10'L).


Results

Hypothesis 1 stated that the mean VR rate will be

higher than the mean FI rate within both phases of the

Flash Card performance. Table 4 shows the correct rates

and the means of those rates by component and phase for

the Flash Card performance of S3. In Phase I the mean VR

rate of correct responses was 4.98. The mean FI rate was

7.11. In Phase II the mean VR rate of correct responses

was 5.27, and the mean FI rate was 5.56. These data show

that the mean VR rate during the Flash Card performance was

lower than the mean FI rate within both phases. Therefore,

Hypothesis 1 is rejected.

Hypothesis 2 stated that the mean VR rate will be

higher than the mean FI rate within both phases of the

Oral Reading performance. Table 4 shows the Oral Reading

correct rates and the means of those rates. In Phase I

the mean VR rate was 35.27 whereas the mean FI rate was 30.5.

In Phase II the mean VR rate was 29.47, and the mean FI

rate was 33.92. These data show that (1) the mean VR rate

was higher than the mean FI rate in Phase 1, and (2) the

mean VR rate was lower than the mean FI rate in Phase II.







-63-


Table 4. Mean Rates of Correct Responses in
EXPERIMENT I


Phase I i Phase II
VR FI VR FI

Flash Card 4.98 7.11 5.27 5.56

I-i Reading 35.27 30.50 29.47 33.92


Flash Card 11.46 11.14 9.44 8.69
1-2
Reading 29.41 31.75 28.79 31.56

Flash Card 8.16 8.50 8.29 10.02

1-3 Reading 32.96 36.91 31.13 40.53








-64-

Therefore, Hypothesis 2 is rejected because the mean VR

rate was higher than the mean FI rate in Phase I alone.


Subexperiment 1-2

A nine-year-old male served as the subject (S4).

Tutors 1 and 2 were a female and a male respectively. The

Multiple schedule components were VR3H and FI 60 seconds L

(F6I0"L).


Results

Hypothesis 3 stated that the mean VR rate will be

higher than the mean FI rate within both phases of the

Flash Card performance. Table 4 shows the correct rates

and the means of those rates for the Flash Card performance

of S4. In Phase I mean VR rate was 11.46, and the mean

FI rate was 11.14. In Phase II the mean VR rate was

9.44, and the mean FI rate was 8.69. These data show that

the mean VR rate was higher than the mean FI rate within

both phases of the Flash Card performance. Therefore,

Hypothesis 3 is accepted.

Hypothesis 4 stated that the mean VR rate will be

higher than the mean FI rate within both phases of the

Oral Reading performance. Table 4 shows the correct rates

and the means of those rates for the Oral Reading performance

of S4. The mean VR rate in Phase I was 29.41, and the

mean FI rate was 31.75. In Phase II the mean VR rate was

28.79, and the mean FI rate was 31.56. Thus, the mean VR








-65-

rate was lower than the mean FI rate within Phase I and

within Phase II. Thus, Hypothesis 4 is rejected.


Subexperiment 1-3

S4 participated in Subexperiment 1-3 subsequent to

his participation in Subexperiment 1-2. Tutors 1 and 2

were both females. Multiple schedule components were

VR3H and FI 10 minute L (FIIO'L).


Results

Hypothesis 5 s-ated that the mean VR rate will be

higher than the mean FI rate within both phases of the

Flash Card performance. Table 4 shows the Flash Card

correct rates and the means of those rates. In Phase I

the mean VR rate was 8.16, and the mean FI rate was 8.50.

In Phase II the mean VR rate was 8.29, and the mean FI rate

was 10.02. Therefore, the mean VR rate was lower than

the mean FI rate within both phases. Thus, Hypothesis 5

is rejected.

Hypothesis 6 stated that the mean VR rate will be

higher than the mean FI rate within both phases of the

Oral Reading performance. Table 4 shows the Oral Reading

correct rates and the means of those rates. The mean VR

rate in Phase I was 32.96, and the mean FI rate was 36.91.

In Phase II the mean VR rate was 31.13, where the mean FI

rate was 40.53. These data show that the mean VR rate

was lower than the mean FI rate within both phases.

Therefore, Hypothesis 6 is rejected.





-66-

Summary of Experiment I Results

Experiment I examined Multiple schedules of social

reinforcement in which one of the two components was a

VR schedule and the other was an FI schedule. The results

of the three subexperiments were discussed relative to the

six hypotheses presented in Chapter III.

The data clearly demonstrate that (1) Hypothesis 3

was accepted; (2) Hypotheses 1, 2, 4, 5, and 6 were rejected;

(3) whereas Hypotheses 1, 4, 5, and 6 were rejected, the

FI component maintained a consistently higher mean rate

than did the VR component within Phases I and II; and

(4) Hypothesis 2 results were in the expected direction in

Phase I, but they were not in the expected direction in

Phase II. The implications of these results will be

discussed in Chapter V.


Experiment II

Experiment II examined Multiple schedules of social

reinforcement across two phases in which (1) the two

Phase I schedule components were equivalent, and (2) one

of the Phase II schedule components was changed. The

data for the experiment are the Flash Card and Oral Reading

performance of S3. Correct rates and the means of those

rates are reported.

S3 participated in this experiment after her partici-

pation in Subexperiment I-1. Tutors 1 and 2 were a female

and a male respectively. Both Phase I Multiple schedule





-67-


components were VI 60 seconds L (VI60"L). Tutor 2

changed to a VR3H schedule in Phase II.

Results

Hypothesis 1 stated that the correct rates of the

two Phase I components will be similar for Flash Card

and Oral Reading performances. The null hypothesis,

the median of the differences is not zero, will be tested

by the sign test. Flash Card and Oral Reading results

are discussed separately.

Table 5 shows the Flash Card Phase I data points and

the direction of the difference between the matched pairs.

For the data in Table 2, x equals the number of matched

pairs who showed differences which equals ten. For N equals

10, an x less than or equal to 1 has a two-tailed probability

of occurrence under the null hypothesis of p = .022.

This value is in the region of rejection for p = .05;

therefore, the alternative hypothesis, that there is a

significant difference between the correct rates of the

two Flash Card components is accepted.

Table 6 shows the Phase I Oral Reading data points

and the direction of the difference between the matched

pairs. For the data in Table 3, x equals the number of

fewer signs which equals 3, and N equals the number of

matched pairs who showed differences which equals ten. For

N equals 10, an x less than or equal to 3 has a two-tailed

probability of occurrence under the null hypothesis of







-68-

p = .344. This value is in the region of rejection for

p = .05; therefore, the null hypothesis is rejected and

the alternative hypothesis, that there is a significant

difference between the correct rates of the two Flash Card

components is accepted.

Hypothesis 2 stated that the Phase II mean VR rate will

be higher than the Phase II mean VI rate for Flash Card

and Oral Reading performance. Flash Card and Oral Reading

data are discussed separately.

The Flash Card data of S3 shown in Table 7 indicates

that the Phase II mean VR rate was 7.82 and mean VI rate

was 9.86. These data show that the mean VR rate was

lower than the mean VI rate. Therefore, Hypothesis 2 is

rejected for the Flash Card performance.

Oral Reading results as displayed in Table 7 show that

the Phase II mean VR rate was 30.13 and mean VI rate was

32.86. These data demonstrate that the mean VR rate was

lower than the mean VI rate; therefore, Hypothesis 2 is

rejected for the Oral Reading performance.

Hypothesis 3 stated that the Phase II mean VR rate

will be higher than the mean VI rate for that tutor (Tutor 2)

in Phase I. Flash Card and Oral Reading are discussed

separately.

According to the Flash Card data presented in Table 7

the Tutor 2 mean VR rate was 7.82 and mean VI rate was 5.09.

This indicates that the mean VR rate was higher than the

mean VI rate. Therefore, Hypothesis 3 is accepted for the

Flash Card performance.






--69-






Table 5. Data Points and the Direction of
Differences Between the Matched Pairs for
the Phase I Flash Card Performance in
EXPERIMENT II

Direction of
Tutor 1 Tutor 2 Differences

1 9.5 6.8 +
2 13.0 3.0 +
3 8.4 6.2 +
4 11.3 4.6 +
5 13.8 3.6 +
6 17.2 8.3 +
7 10.4 6.0 +
8 12.0 5.5 +
9 6.7 9.6
10 12.8 8.8 +






-70-


Table 6. Data Points and the Direction of
Differences Between the Matched Pairs for
the Phase I Oral Reading Performance in
EXPERIMENT II


Direction of
Tutor 1 Tutor 2 Differences

1 39.5 39.1 +
2 56.0 18.0 +
3 33.0 38.0
4 28.0 26.2 +
5 31.0 32.5
6 38.5 28.6 +
7 40.8 29.0 +
8 31.7 23.0 +
9 30.3 26.3 +
10 29.2 34.6








-71-



Table 7. Mean Pates of Correct Responses in
EXPERIMENT II


Flash Card

Reading


Phase
(Tutor 1)
VI

11.72

36.51


I
(Tutor 2)
VI

5.09

29.43


Phase
(Tutor 1)
VI

9.86

32.86


II
(Tutor 2)
VR

7.82

30.13







-72-

The Oral Reading data shown in Table 7 reveals the

following: the Tutor 2 mean VR rate was 30.13 and mean

VI rate was 29.43. These data indicate that the mean VR

rate was higher than the mean VI rate. Therefore,

Hypothesis 3 is accepted for the Oral Reading performance.

Hypothesis 4 stated that the Phase I mean VI rate for

the unchanged tutor (Tutor 1) will be higher than the

mean VI rate for that tutor in Phase II. Flash Card and

Oral Reading will be discussed separately.

The Flash Card data (Table 7) for Tutor 1 revealed

the following: in Phase I the mean VI rate was 11.72 and

in Phase II the mean VI rate was 9.86. These data show

a decrease in Tutor l's mean VI rate from Phase I to

Phase II; therefore, Hypothesis 4 is accepted for the

Flash Card performance.

Table 7 shows the Oral Reading performance. Tutor 1

Phase I mean VI rate was 36.51 and the Phase II mean VI

rate was 32.86. These data also represent a decrease in

mean rate from Phase I to Phase II; therefore, Hypothesis 4

is accepted for the Oral Reading performance.

Summary of Experiment II Results

Experiment II examined Multiple schedules of social

reinforcement and used a contrast design to demonstrate

the effects. These results are summarized as follows:

(1) Hypothesis 1 was rejected, (2) Hypothesis 2 was rejected,

(3) Hypothesis 3 was accepted, and (4) Hypothesis 4 was







-73-

accepted. Although there was a significant difference

between the correct rates in Phase I, the introduction

of the VR component in Phase II increased the mean rate

for the changed tutor and decreased the mean rate for

the unchanged tutor. An interpretive analysis of these

results will be presented in Chapter V.

Experiment III

Experiment III examined Multiple schedules of social

reinforcement in which at least one of the two components

manipulated error Yessponses. Two subexperiments were

each comprised of two experimental phases. Schedule

components across phases of each subexperiment were identical.

The components assigned to tutors in Phase I were switched

in Phase II. Correct and error rates and the means of

those rates are reported since this experiment manipulated

both correct and error responses.

Subexperiment III-1

Sl, an eight-year-old female, participated in

Subexperiment III-1. Multiple schedule components were

FI 30 seconds L (RC) and FI 30 seconds C (RE). Tutors

reinforced responses at the same frequency; one reinforced

correct responses and the other reinforced error

responses. Tutor 1, a female, reinforced correct responses

and Tutor 2, a female, reinforced error responses in Phase I.

The tutors switched components in Phase II; therefore,








-74-
Tutor 1 reinforced error responses and Tutor 2 reinforced

correct responses in Phase II.

Results

Hypothesis 1 stated that the Flash Card mean correct

rates in the RC component will be higher than the mean

correct rate of the RE component in Phase I and in

Phase II of the Flash Card performance. Table 8 shows

the correct and error rates and the means of those rates

for the Flash Card performance of Sl. The Phase I mean

RC rate was 8.03 and the mean RE rate was 6.37. These

data show that the mean RC rate was higher than the

mean RE rate. The Phase II mean correct rate for the RC

was 6.03 and for the RE was 5.32. These data show that

the mean RC rate was higher than the mean RE rate. The

results from Phase I and Phase II support the acceptance

of Hypothesis 1.

Hypothesis 2 stated that the Flash Card mean error

rates in the RE component will be higher than the mean

error rates of the RC component in Phase I and in Phase II.

Table 8 provides the data for this Hypothesis. The Phase I

mean RE rate was 2.80 and the mean RC rate was 1.35.

These data show that the mean RE rate was higher than

the mean RC rate. The Phase II mean RE rate was 2.04

and the mean RC rate was 3.02. These data show that the

mean RE rate was lower than the mean RC rate in Phase II.

Since only the Phase I results were in the expected

direction, Hypothesis 2 is rejected.







-75-


Table 8. Mean Rates of Correct and Error Responses
in EXPERIMENT III


Phase I
RC RE


Flash Card

Reading


- Correct
Error
- Correct
Error


8.03
1.35
41.30
0.77


6.37
2.80
39.96
3.07


RC RE


Flash Card Correct
Error
Reading Correct
Error


4.93
1.53
17.66
1.78


4.52
2.31
17.81
2.08


Phase II
RC RE


6.03
3.02
34.63
2.19


RC RE


2.46
2.01
15.51
2.12


III-1


III-2


5.32
2.04
42.55
2.03


3.10
1.88
16.26
1.05








-76-

Hypothesis 3 stated that the Oral Reading mean correct

rates in the RC component will be higher than the mean

correct rates of the RE component in Phase I and in

Phase II. Table 8 shows the correct and error rates and

the means of those rates for the Oral Reading performance

of Sl. The Phase I mean RC rate was 41.3 and the mean

RE rate was 39.96. These data show that the mean RC

rate was higher than the mean RE rate in Phase I. The

Phase II mean correct rate for the RC was 34.63 and for

the RE was 42.55. These data show that the mean RC rate

was lower in Phase II than was the mean RE rate. Since

only the Phase I results were in the expected direction,

Hypothesis 3 is rejected.

Hypothesis 4 stated that the Oral Reading mean error

rates in the RE component will be higher than the mean

error rates of the RC component in Phase I and in

Phase II. Table 8 provides the data for this hypothesis.

The Phase I mean RE rate was 3.07 and the mean RC rate

was 0.77. These data show that the mean RE rate was higher

than the mean RC rate in Phase I. The Phase II mean RE

rate was 2.03 and the mean RE rate was 2.19. These data

show that the mean RE rate was lower than the mean RC

in Phase II. Since only the Phase I results were in the

expected direction, Hypothesis 4 is rejected.








-77-


Subexperiment 111-2

S2, an eight-year-old male, served as the subject

in Subexperiment 111-2. The components of the Multiple

schedule were a VI 60 seconds L (RC) and a VI 60 seconds D

(RE). Tutors 1 and 2, a female and a male switched

schedules from Phase I to Phase II.


Results

Hypothesis 5 stated that the Flash Card Phase I and

Phase II mean correct rates will be higher in the RC

component than in the RE component. Table 8 shows the

correct and error rates and the means of those rates for

the Flash Card performance of S2. In Phase I the mean

RC correct rate was 4.93 and the mean RE correct rate

was 4.52. The RC mean correct was, therefore, higher

than the RE mean correct in Phase I. In Phase II the mean

RC rate was 2.46 and the mean RE rate was 3.10. These

data show that the mean RC rate was lower than the mean RE

rate in Phase II. Therefore, since only the Phase I

results were in the expected direction, Hypothesis 5 is

rejected.

Hypothesis 6 stated that the Flash Card Phase I

and Phase II mean error rates will be higher in the RE

component than in the RC component. Table 8 provides

the data for this hypothesis to be tested. In Phase I

the RE mean error rate was 2.31 and the RC mean error

rate was 1.53. These data show that the RE mean error

rate was higher than the RC mean error rate in Phase I.







-78-


In Phase II the RE mean error rate was 1.88 and the

RC mean error rate was 2.01. These data show that the

RE mean error rate was lower than the RC mean error

rate in Phase II. Since only the Phase I results were

in the expected direction, Hypothesis 6 is rejected.

Hypothesis 7 stated that the Oral Reading Phase I

and Phase II mean correct rates will be higher in the

RC component than in the RE component. Table 8 shows

the correct and error rates and the means of those rates

for the Oral Reading performance of S2. The Phase I

data reveals that the mean RC rate was 17.66 and the

mean RE rate was 17.81. Thus, the mean RC rate was lower

than the mean RE rate which was not expected. In Phase II

the mean RC rate was 15.51 and the mean RE rate was 16.26.

These results are not in the expected direction. That is,

the mean RC rate was not higher than the mean RE rate in

Phase II. Therefore, Hypothesis 7 is rejected.

Hypothesis 8 stated that the Oral Reading Phase I

and Phase II mean error rates will be higher in the RE

component than in the RC component. Table 8 provides the

data for this hypothesis. In Phase I the mean RE error

rate was 2.08 and the mean RC error rate was 1.78. These

results were in the expected direction. However, the

Phase II results were not in the expected direction. The

mean RE error rate as 1.05 was not higher than the mean









-79-

RC error rate which was 2.08. Therefore, since the

results of both phases were not in the expected direction,

Hypothesis 8 is rejected.


Summary of Experiment III Results

Experiment III examined the effects of Multiple

schedules of social reinforcement in which one of the

two components reinforced correct responses and the other

reinforced error responses. Although all eight hypotheses

were rejected, there was remarkable consistency among the

results. Whereas Phase I results for each hypothesis were

in the expected direction, Phase II results were not in

the expected direction. These interesting results will

be discussed in Chapter V.












CHAPTER V
SUMMARY AND DISCUSSION

Experiments I, II, and III attempted to determine

whether there was a predictable relationship between the

Multiple schedule of social reinforcement and the subject's

mean rate of responses. This chapter discusses these

results and the similarities and differences between them

and other investigations.

The results of the three experiments suggested that

while the experimental manipulations did not produce data

in the expected direction overall, there was remarkable

consistency among the results. This consistency has

important implications for future investigations of social

reinforcement schedules. The chapter concludes with a

discussion of those implications.

Briefly, the following was found: (1) Experiment I

demonstrated differential responding to Multiple schedule

components during Phase I. Generally the schedule which

generated higher mean rates in Phase I produced higher

mean rates in Phase II; (2) Experiment II's design

contrasted the differential effects of changing one tutor's

reinforcement frequency on rates of reading and the effect

of that change on the rate of response to the unchanged

tutor; (3) Experiment III demonstrated the irreversibility


-80-








of response from Phase I to Phase II when at least one

of the tutors consequated error responding.

Experiment I demonstrated differential responding

to Multiple schedules of social reinforcement. Although

the data did not replicate previous research, the consistency

of the data suggests that response rate was under the

control of the schedule.

The Multiple schedules used in the three subexperiments

each included a VR schedule and an FI schedule as the two

components. Flash Card and Oral Reading mean response

rates generally followed a consistent pattern in each

subexperiment. Of the Flash Card and Oral Reading per-

formances which changed, four unexpectedly generated higher

rates in the FI component than in the VR component. In

the fifth instance, the Flash Card performance of S4 in

Subexperiment 1-2, the VR component showed a higher mean

rate in both phases than did the FI component. In the last

case, the Oral Reading performance of S3 in Subexperiment I-1,

neither generated a consistently higher performance.

In general, then, Experiment I results showed that the

VR with its higher reinforcement density, generated lower

response rates than did the FI. These results are supported

by the Gewirtz (1967) findings in the area of satiation

and deprivation of social reinforcement. Gewirtz concluded

that the effectiveness of an approval word varied inversely

with the frequency of its occurrence prior to the reinforce-

ment test.





-82-


Basic animal operant research reviewed briefly in

Chapter I found that the VR schedule typically generated

higher rates of responding than did the FI schedule

(Ferster and Skinner, 1957). While the Experiment I

results are inconsistent with that study's findings, the

particular characteristics of the present investigation

presented may have a replication of the Ferster and Skinner

findings. In this study the timing device continued to

operate during the delivery and "ingestion" of the social

reinforcement. In the cited research, the delivery of

reinforcement was controlled mechanically. During rein-

forcement delivery, the timing device shut down; therefore,

the calculation of rate of response did not include the

time used to deliver and ingest reinforcement.

A description of S3's behavior provides an example

of how this procedural difference may have influenced the

results. S3 who participated in Subexperiment I-1,

reacted emotionally to each reinforcement which seemed to

disrupt her attention to the materials. The higher

frequency of reinforcement during the VR condition and the

disruption induced may have lessened the overall time in

which the subject had to respond. Thus, the calculation

of rate may have been affected.

Staats, Finley, Minke, and Wolf (1964) demonstrated

results with Multiple schedules on human reading responses

somewhat consistent with basic animal research. However,







-83-

unlike the present study, the reinforcers were tokens,

which allow greater control over response rate.

Lovitt et al. (1971) found results contrary to the present

findings as well. Using single ratio token schedules,

these investigators generated differential oral reading

performance to CRF/EXT and FR30/EXT. As with the Staats et al.

(1964) study, schedule differences and the use of tokens may

account for the contradictory results.

Racial factors may also explain these results. As

discussed in the Literature Review, blacks and whites seem to

be responsive to different sets of social reinforcers. In

this regard perhaps the verbal statements selected as reinforcers

served instead as negative reinforcers. The subjects may

have responded at a slower rate on the VR schedule in order to

avoid the occurrence of the verbal statement. It would be

important in future research to rectify the procedural problem

already alluded to in order to clarify this point.

Experiment II demonstrated the contrast effect of changing

reinforcement frequency of one of two previously equivalent

components. "The change in behavior is called a contrast

when the change in the rate of responding during the presentation

of one stimulus is in a direction away from the rate of

responding generated during the presentation of the other

stimulus" (Reynolds, 1961, p. 57). Phase I Multiple schedule

components were VI60"L. Tutor 2 generated a lower Flash

Card and Oral Reading Phase I mean correct rate and therefore

was chosen to change to the VR3H schedule in Phase II. As







-84-

expected the introduction of the.VR3H increased the correct

rates in the changed component and lowered the correct rates

in the unchanged VI component.

Waite and Osborne (1972) provided the first demonstration

of behavioral contrast in children. Multiple schedules

influencing a simple operant were VI/VI and VI/EXT. The

current study supports their findings by producing this

contrast effect with different Multiple schedules using a more

complex response reading.

In contrast to the present experiment, Waite and

Osborne (1972) found equivalent rates during the VI/VI schedule.

Phase I rates were not equivalent in the present study.

Because of this lack of equivalent rates, the tutor with the

lower mean rate was chosen to change schedules. The results

are not weakened by the lack of equivalent rates in the first

phase.

Experiments I and II as a unit demonstrated the influence

of social reinforcement on Flash Card and Oral Reading

responses. Social reinforcement schedules were shown to be

the controlling variable. Certain factors in this study may

have enhanced this outcome. First, the use of different

tutors to serve as component stimuli may have hastened the

discrimination between schedules. Second, the control over

the rate of progress stabilized the effect of the "learning-

to-learn" phenomena. This allowed the response to fluctuate

with the experimental manipulations. Third, point delivery







-85-

was standard. The present study suggests that social

reinforcement, a more naturally occurring reinforcer than

a point or token, has a significant effect on a relevant

educational behavior.

Contrary to the results of Experiments I and II which

showed schedule control, Experiment III did not demonstrate

such control. Schedule effects in Phase I of the subexperiments

were not maintained in Phase II when the tutors switched

schedules. Thus, the tutor effect appeared to exert more

influence over the results than did the schedules. This may

have been due to the nature of the response being consequated.

In this experiment, at least one of the two Multiple schedule

components consequated error responses.

It is unclear, however, whether the type of reinforcement,

the consequation of errors, or the sequence of treatments

was more important in influencing these results. Each

subexperiment had equivalent schedules in terms of the schedule

specification. That is, both schedules of Subexperiment III-1

were FI30" and those in III-2 were VI60". One component in

each consequated correct responses and one consequated error

responses. The consequation of error responses by one tutor

in Phase I may have induced an emotional reaction which could

have exerted more control over the responses in Phase II

than the schedule. For these reasons, caution must be taken

in drawing conclusions from Experiment III.







-86-


Experiment III's irreversible results contradict the

findings of several other studies. Sajwaj and'Knight (1971)

found, for example, that errors consistently increased if

more attention was given them than corrects. Hasazi and

Hasazi (1972) found that removing attention for digit-

reversals significantly reduced errors. These investigations

sought to eliminate errors and no reversal was attempted;

whereas the current one attempted to bring errors under

control of the schedule.

The results of two informational reinforcement studies

(Buchwald, 1959; Spence and Dunton, 1967) also differ with

Experiment III's lack of results. In comparing the effective-

ness of Rb, RW, and Wb, Rb was found to be inferior. Therefore,

one would assume that the Wb conditions in Experiment III

would yield consistently superior rates. That they did not

may, of course, reflect the design of the current study

and the sequence of treatments which was much different from

the group design and CRF application of reinforcement in the

informational studies.

An analysis of the results of Experiments I, II, and III

leads one to conclude that Multiple schedules of social

reinforcement were shown to demonstrate differential effects

on reading behavior particularly when reinforcement was

contingent upon correct responses. When schedules of reinforce-

ment were applied to error responses, the interpretation of

the results was unclear. Emotional responses to the tutors,

as the sequencing interpretation would suggest, may influence







-87-


the results. It must also be made clear that although

differential responding resulted from the use of Multiple

schedules, the results are confounded by the inclusion

of different types of social reinforcement within the Multiple

schedules. The study demonstrated differential responding

to social reinforcement in a single subject design. Future

research to clarify these results or to simply look directly

at schedules of social reinforcement would be useful.

The conclusion, then, that social reinforcement was

the controlling variable influencing subject responsiveness

suggests that researchers attempt the control of complex

human behavior with more naturally occurring reinforcers.

The present research does not deny the contention that social

reinforcement plus tokens may be more effective than social

reinforcement alone (Lahey and Drabman, 1974; Staats et al.,

1962; Chadwick and Day, 1971). Rather, this study simply

suggests that social reinforcement alone can influence a

child's reading.

The methodology in this study could be useful in future

research on the teaching of reading. In order to demonstrate

control by social reinforcement, many aspects of the tutor-

subject interaction had to be standardized. The standardization

provides a useful base from which the researcher or teacher

could give structured input and observe its effects. Instructed

to ignore inappropriate behaviors, such as errors, and to

attend to appropriate or on-task behaviors such as correct







-89-

as a research tool. Researchers questioned the viability

of using social reinforcement and turned to token reinforce-

ment as a more manageable research tool.

Furthermore, historically social reinforcement research

tended to control subservient behaviors (Thomas, Nielsen,

Kuypers, and Becker, 1968) or irrelevant human behaviors, such

as rate of placing marbles in holes (Panda, 1972). The results

of the present research, however, suggest that factors of

social interaction can be specified in order that social rein-

forcement can be discreetly delivered within so complexly

defined a setting as remedial reading. In other words, social

reinforcement effects can be investigated in relation to com-

plex human learning.

The utilization of the Performance Determined Instruction

model (Gray et al., 1969) stabilized the "learning-to-learn"

phenomenon. This allowed the behavior to fluctuate as a

result of the manipulation of experimental variables. The

stabilization of accuracy within the 90 to 95 percent

accuracy level ties in with J. McVicker Hunt's theory of the

match necessary between the curriculum and the child to

promote intellectual accommodation (Hunt, 1964). The

influence of emotional responses of boredom or frustration

from too high or too low an accuracy level was controlled by

the use of the PDI system.







-90-

Critical reviews of past social reinforcement research

suggested the need for functional designs (Van de Riet, 1963;

Kennedy and Willcutt, 1964). The reversal and contrast

designs of the present study controlled reinforcement history,

the influence of characteristics of the subject, and the

influence of characteristics of the tutors. Therefore, these

results support the use of functional designs in social

reinforcement research.

An important area of research is opened by the current

project. An analysis of the variables controlling error

responding seems warranted. It remains of interest whether

a reading error response is analogous to an inappropriate

social behavior in terms of the controlling variables. The

present research suggests the need for more extensive analysis

of the factors which influence error response rate. For

instance, one could systematically replicate Experiment III

using a Multiple CRF/EXT schedule.

Methodologically, this study sustained implications for

research, teaching, and the treatment of reading disorders.

Basic research can be attempted in a school setting. Further-

more, the results suggest that basic research on relevant

human behaviors can occur simultaneously with the provision

of remedial training.

As social behavior has been reliably shown to be

influenced by environmental consequences, the present study

indicates that it influences reading behavior. The contrast

between one teacher's style and another's may have a







-91-

significant effect on an individual child's performance.

Likewise, teacher contrast can effect a child's emotional

responses and preferences. Consider the junior high program

in which a student might have five equally reinforcing or

demanding teachers first semester. Second semester the

math teacher is changed to either an extremely negative or

highly reinforcing one. The possible concommitant effect on

responding to math and to the other courses seems analogous

to the present contrast findings.

The implications of the present research for the teaching

of remedial reading are most promising. A training package

can be developed based on the methodological decisions made

for the current research. Instead of remediating specific

skill deficiencies in hopes that there would be some

generalization to overall reading strength, this project took

a more radical view of the teaching of reading. The research

assumed that if the child was placed at his 90 to 95 percent

accuracy level in a programmed curriculum,the interaction

between the tutor and the child would be critical to the

child's strength as a reader. Not only was the interaction

assumed to be important, but its dimensions were strictly

specified. Random interactions were kept at a minimum. The

control of social interaction with regard to both correct

and error responses in remedial reading situations is

important.







-92-


Summary

Reading behavior has been shown to be systematically

effected by social reinforcement. It is unclear whether

type of reinforcement or schedule of reinforcement was the

controlling variable. Procedural limitations and the

question of whether the verbal statements were in fact

reinforcing may explain why the results were not in the

expected direction. Analysis of error responding appears to

be a fertile area for research. Basic research on complex

human learning tasks seems viable within a public school

setting. The small number of subjects limits the generalizability

of these results, however, the clarity of the influence of

social reinforcement seems to be a needed addition to the

literature on the social reinforcement of reading. The

methodology developed for this study to control tutor-

subject interaction, rate of progress through a curriculum,

and the delivery of social reinforcement has important

implicationsfor research, teaching, and the treatment of

reading disorders.




-93-


APPENDIX A
POINT SHEET


r':f













APPENDIX B
DAILY DATA SHEET


Date Phonics

Child_

Tutor

Story


Social Reinforcement
Positive Comments
Negative Comments
Complaints


Flash Error Reading__ Error Writing Error

Time Time Time

Correct Correct Correct

1. 4. 1. 4. 1. 4.

2. 5. 2. 5. 2. 5.

3. 6. 3. 6. 3. 6.



































APPENDIX C

STANDARD BEHAVIOR CHARTS




Full Text

PAGE 1

SOCIAL ACKNOWLEDGMENT OF READING; MULTIPLE SCHEDULE EFFECTS by Pamela Phelan A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UtJIVERSITY OF FLORIDA 1976

PAGE 2

ilfiiilil 3 1262 08552 7777

PAGE 3

TABLE OF CONTENTS Page Abstract . v Chapter I. INTRODUCTION 1 Purpose and Importance of Study ......... 2 Background of Study: Two Major Research Projects. ......... 3 Schedules of Reinforcement 3 The Staats Study. 5 General Statement of the Research Purpose .... 9 Definitions of Terms 9 Limitations 13 II. LITERATURE REVIEW 14 Major Developments in Social Reinforcement Research 14 Satiation and Deprivation 15 Subject Characteristics and Their Influence on Social Reinforcer Effectiveness 17 Experimenter Effects 22 Informative and Affective Roles 24 Relationship between the Response and Reinforcement Delivery 26 S\immary of Historical Developments 30 ii

PAGE 4

Page II. Literature Review (cont'd) Schedule Research on Reading .......... 31 Schedules of Token Reinforcement . 31 Schedules of Social Reinforcement 34 Procedural Requirements of a Study of Social Reinforcement Schedxile Effects on Reading. . 35 III. METHODS AND PROCEDURES 42 Research Hypotheses 43 Experiment I ......... . 43 Experiment I^-^ . . . 45 Experiment III 46 Subjects ..... 48 Tutors 48 Remedial Reading Program . 49 Setting. 49 Apparatus. 49 Materials 50 Data Collection 50 Data for Data Analysis 51 Procedures . 51 Flash Card Procedures 51 Oral Reading Procedures. 52 Adaptation of the PDI Model 52 Point Reinforcement Procedures . . , 54 Experimental Procedures. . 54 Social Reinforcement Procedures. . 55 Lll

PAGE 5

Page IV. RESULTS . . . . . . .. . . . . . . 59 Experiment I. ....... 59 Subexperiment I-l 62 Subexperiment 1-2 64 Subexperiment 1-3 65 Siommary of Experiment I Results . 66 Experiment II 66 Siamnary of Experiment II Results. 72 Experiment III 73 Subexperiment III-l 73 Subexperiment III-2 77 Summary of Experiment III Results ........ 79 V. SmU^RY AND DISCUSSION 80 APPENDIX A Point Sheet. . 93 APPENDIX B Daily Data Sheet 94 APPENDIX C Standard Behavior Charts . 95 Bibliography 108 Biographical Sketch 115, IV

PAGE 6

Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SOCIAL ACKNOWLEDGMENT OF READING: MULTIPLE SCHEDULE EFFECTS by Pamela Phelan August, 1976 Chairman: Donald Avila Major Department: Foundations of Education This study investigated the influence of social reinforcement on the reading behavior of children enrolled in a tutorial reading center. Four black third grade pupils from a. Gainesville, Florida elementary school served as subjects. The basic data were daily rates of Flash Card responses and Oral Reading responses. Programmed instruction materials were adapted for use in this study. The study was divided into three experiments , each comprised of two experimental phases. The basic question being asked was whether there is a predictable relationship between the Multiple schedule of social reinforcement and the subjects' mean rates of response.

PAGE 7

Social reinforcement was delivered by university students who functioned as tutors. Two tutors were assigned to each subject. The tutors taught subjects on alternate days in one-to-one sessions over a six-week period. Experiment I investigated Multiple schedules of social reinforcement of correct responses in which the two Multiple schedule components were a Variable Ratio schedule and a Fixed Interval schedule. Schedule components assigned to tutors_j.n Phase I were switched in Phase II. Experiment I included three subexperiments . The Variable Ratio schedule was expected to generate higher rates of responding tha;n the Fixed Interval schedule in both phases. Experiment II used a slightly different design called a contrast design. Multiple schedule components were equivalent in Phase I and in Phase II one of the components was changed. Correct responses were reinforced. Phase I components were a Variable Interval schedule and one of these was replaced by a Variable Ratio schedule in Phase II. A contrast effect was expected. Similar rates of responding were expected in Phase I. In Phase II the rates for the changed tutor V7ere expected to increase from Phase I to Phase II and the rates for the unchanged tutor were expected to decrease from Phase I to Phase II. VI

PAGE 8

Experiment III, similar to that of Experiment I, investigated the effects of Multiple schedules in which one of the two components reinforced error responses. The tutor that reinforced corrects in Phase I reinforced errors in Phase II, and the tutor that reinforced errors in Phase I reinforced corrects in Phase II. Experiment III was comprised of two subexperiments . Reinforcement was delivered on equivalent Fixed Interval schedules in the first subexperiment and on equivalent Variable Interval schedules in the second one. Higher rates of correct responses were expected to be generated by the tutors who reinforced corrects in both phases. Likewise, higher rates of error responses were expected by the tutors who reinforced errors in both phases. The results of the three experiments were not in the expected direction. Two limitations of this study may have influenced the results. First, precise measurement of rate of response was impossible since the time used for reinforcement delivery was included in the total time used to calculate rate. Second, the verbal statements used by the white tutors as reinforcers may not have been reinforcing to the black subjects. This study demonstrates clearly, however, that social reinforcement research can be conducted on an educationally relevant behavior in a public school setting. CHaTrman" vii

PAGE 9

CHAPTER I INTRODUCTION All societies develop structures to resolve and modify conflict among their members , Mechanisms for allaying a conflict may be formal or informal. Laws and their enforcement codify norms while folkways and social standards informally govern behavior through social rewards and punishment. Social reinforcement is a societal or cultural tool for shaping and maintaining acceptable behavior. The methodological application of social reinforcement techniques in the management of human behavior is a widespread practice. These management techniques have been used for training and supporting behavioral appropriateness in varied settings such as schools, mental health treatment centers, and homes. Since incorporation into daily operating procedures, social reinforcement techniques have become a viable tool for change. \\Tiile the need for pragmatic psychological procedures is recognized and, indeed, pursued by social agencies, there is little evidence of sustained application of social reinforcement techniques to cognitive task performance in education. This lack became increasingly clear as extensive literature research for the present study progressed. -1-

PAGE 10

•2Pu rpose and Impor tance of the Study The major goal of this investigation is to examine social reinforcement effects on cognitive task performance. More specifically, this study investigates for the first time the effects of the systematic use of social reinforcement on the reading behavior of elementary school children. A second aim is to develop research methodology and procedures for the application of laboratory technology to the study of social reinforcement effects in reading. The successful specification and implementation of these procedures will provide a research strategy for the continuous assessment of social reinforcement effects on reading behavior. Furthermore, implementation within the context of a remedial training program encourages the use of the natural environment for basic research programs. It is significant that the literature on social reinforcement effects on cognitive task performance contains only a few references of its use over more than a few sessions' duration. Many current educators and research psychologists agree that (1) most of the existing studies are superfluous to an ongoing assessment of social reinforcement effects, and (2) that most studies fail to relate to the classroom teacher in more than an indirect fashion. For these reasons, the need for basic research in a context directly relevatit to the educational process is obvious .

PAGE 11

-3Past research germane to reading, using social reinforcement as the experimental treatment, has been limited to short term studies of letter discrimination, sight word vocabulary responding, and/or unit reading responses. This study investigates the effects of systematic manipulations of social reinforcement during long-term reading development. Background of the Study: Two Major Research Projects A substantial niomber of research reports which have direct and indirect bearing on this study will be reviewed in Chapter II. There are, however, two studies which have such a f^andamental impact on the current investigation that they will be noted at this point. The first, representing a substantial program of research, explicates the term schedules of reinforcement. The second describes a study on schedules of token reinforcement of reading and has a direct bearing on the present investigation. Schedules of Reinforcement B.F. Skinner conducted the foundation research in the area of schedules of reinforcement. A lengthy work, Schedules of Reinforcement (Ferster and Skinner, 1957) summarized the whole of his investigations. From this work, the term schedule of reinforcement is defined as a systematic arrangement of response-reinforcement contingencies which vary the proportion of reinforced responses or the time intervals after which reinforcement is obtained.

PAGE 12

Maay response-reinforcement contingencies have been labeled, and derive froratwo categories of schedules: simple and complex. The six simple schedules are Continuous Reinforcement schedule (CRF) , Fixed Ratio (FR) , Variable Ratio (VR) , Fixed Interval (FI) , Variable Interval (VI) , and Extinction (EXT) . These are defined in the Glossary. Complex schedules combine two or more simple schedules. Some of the complex schedules are the Multiple schedule, the Mixed schedule, the Tandem schedule, and the Concurrent schedule. Although it may be instructive to describe the various schedules and their pattern effects, it is beyond the scope of this paper to do so. However, description of the Multiple schedule is germane to an understanding of schedules in general and to the current study in specific. The Multiple schedule was used in both the present study and the token reinforcement study which will be discussed in the next section. A Multiple schedule of reinforcement is one in which each of two or more randomly alternating schedules of reinforcement is signalled by the presence of a different stimulus. An example illustrating a Multiple CRF/EXT schedule is selected to clarify the definition. Consider that a subject is seaced in a room with two lights overhead, one red and one green; with one lever and one receptacle in front of him. Illumination of the red light signals the CRF schedule, while illumination of the green light signals

PAGE 13

-5-^ . the EXT schedule. For each lever press while the red light is illviminated the subject received a penny in the receptacle. Lever presses in the green condition produce nothing. Thus, it can be seen that the two stimuli, red and green lights, signal the occurrence of the two schedules, CRF and EXT. Ferster and Skinner's major finding demonstrated that patterns of responding emerged unique to each schedule. For example, the VR schedule typically generated higher response rates than the VI schedule. Research has replicated these patterns with different species and reinforcers. The schedule was considered in control of the subject's behavior when the pattern emerged in the data. The Staats Study Schedule research on educationally important behaviors is rare. The utilization of social reinforcemeiit in such investigations is virtually non-existent. A discussion of a token reinforcement of reading study is chosen to provide a frame of reference for understanding the present social reinforcement study. The purpose of the Staats study (Staats, Finley, Minke, and Wolf, 1964) was to demonstrate schedule control of human reading behavior. A description of the methodology and procedures is followed by a discussion of the results and their implications for the present investigation. Each of four four-year-old subjects was assigned to one of four Multiple schedules of token reinforcement.

PAGE 14

6These schedules were Multiple CRP/EXT, Multiple CRF/VR, Multiple CRF/VI, and Multiple VR/VI. An apparatus presented materials , delivered marble reinforcers, exchanged marbles for trinkets, and recorded data. The material presentation panel was mounted on top of a table in front of the subject, A doorbell-type pushbutton was also on the table in front of the subject. Four plexiglass windows, one centered above three others, were mounted on the panel. The printed material was viewed through the top window; a matching stimulus and two others through the bottom three windows. Beneath each of the lower three windows was a button which, when pressed, activated a marble dispensing device if the correct match was selected. Correct matching responses were reinforced by the delivery of a marble to a small box on the table. The subject could immediately exchange the marble for a small trinket or could place it in a tube for accumulation toward one of four more costly trinkets. A cumulative recorder recorded the subject's responses, the point of operation of the marble dispenser, and the point of delivery of a toy. The operation of the doorbell, windows, marble dispenser, exchange equipment, and cumulative recorder was controlled by electromechanical equipment. A series of trainiiig sessions prepared the subject for the task of appropriately responding to the written materials, Initially simple echoic responding was reinforced. The response required for reinforcement gradually approximated

PAGE 15

>7that of the experimental response which was vocalizations of written letter combinations. The final chain of responses required for reinforcement was the following: the subject pressed the doorbell which resulted in the appearance of the reading characters in the windows; the Experimenter said the name of the character in the top window which the subject echoed; the subject, then pressed the button below the correct matching window which activated the marble dispenser. If the subject:, vocalized the correct response prior to the Experimenter's prompt, the Experimenter activated the marble dispensing machine. The chain began again when the subject pressed the doorbell for the display of a new set of characters in the windows . Marble reinforcement occurred throughout the training and the experiment. During training every correct response was reinforced with the presentation of a marble. When the subject was ready to begin the experimental response chain, the Multiple schedule reinforcement contingencies v^ere placed in effect. Two procedures controlled the change of component and stimulus requirements of the Multiple schedules. During one component, an overhead light in addition to a light above the apparatus was illuminated. For the other component, just the overhead light was illuminated. The change of components of the Multiple schedule was determined

PAGE 16

by the following: the CRF was in effect for 15 reinforcers, the VR for 5 reinforcers, the VI; for 10 minutes, and the EXT for 5 minutes. Response rates were reported for each of the components of the Multiple schedules. Results were discussed in terms of their compatibility with basic schedule research. Reference to this occurred in the previous section of this chapter. (1) The Multiple CRF/EXT subject produced rates in the CRF condition which exceeded the rates produced in the EXT condition. These results were compatible with basic research. (2) The Multiple CRF/VR subject produced rates in the VR condition in excess of those in the CRF condition. These results were compatible with earlier research. (3) The Multiple CRF/VI subject's production of higher rates in the CRF condition than in the VI condition was compatible with basic research results. (4) Unexpectedly the Multiple VR/VI subject produced approximately equal rates in both conditions. It had been assimied that the VR condition would generate decidedly higher response rates than the VI. In summary, each case in which the CRF schedule had been the comparison schedule resulted in data compatible with basic research findings. These results demonstrated that basic research can be conducted on a significant human behavior. The laboratory analysis of token reinforcement effects has wide generalizability to both the study of reinforcement schedules and the development of reading. In this regard, a study with

PAGE 17

-9a similar research paradigm which also utilizes social reinforcement would seem to have even greater applicability to hxjman behavior and the educational process. General Statement of t h e Research Purpose This study will demonstrate the effects of Multiple schedules of social reinforcement upon the rate of verbal responses to written material by elementary school children. A study of this nature is without precedence. The application of laboratory technology to the development of reading in pupils '"enrolled in a remedial reading program increases its significance. In Chapter II, the historical antecedents relating to the development of the specific research hypotheses will be presented and discussed. The review of literature is intended to place these introductory remarks within the perspective of an historical overview. The specific research hypotheses will then be presented in Chapter III where they are discussed with respect to research design and methodology. Definitions of Teinns The following list of terms and their definitions is provided for the reader in order that they might fully understand the phraseology used in this work.

PAGE 18

-10Baseline: the ongoing rate of a behavior prior to the introduction of the treatment. Behavioral contrast or contrast effect: the effect on performance on an unchanged schedule of reinforcement to which the subject is exposed. Celeration: basic unit of measurement of behavior change. Change in frequency per unit of time; movements/ minute/week. Concurrent schedule: Schedules which are in effect simultaneously, or which can follow one another in rapid succession. Conditioned reinforcer: a reinforcer whose reinforcing effect depends upon a history of training. The actual reinforcer maintaining the rate of the behavior is the stimulus immediately following it. In the case of the pigeon, the reinforcement for pecking is the sound and light accompanying the operation of the food dispenser. These stimuli in turn set the occasion on which the pigeon may go to the feeder and eat. Continuous Reinforcement schedule (CRF) : reinforcement of every instance of a target behavior. Contrast design: a design which will potentially produce the contrast effect.

PAGE 19

-11Giomulative record: a graphic record which emphasizes the rate of performance or its frequency. In a cumulative record a recording pen moves along the abscissa with passage of time and along the ordinate each time a performance occurs. Thus, the rate of the performance is demonstrated by the slope of the curve. Cumulative recorder: gives a continuous visual picture of the subject's rate of response. Extinction (EXT): the operation of nonreinf orcement . Fixed-Interval schedule (FI) : reinforcement is produced by the first response emitted after a certain time has elapsed since the last reinforcement. Fixed-Ratio schedule (FR) : the target response is reinforced upon completion of a fixed number of responses counted from the preceding reinforcement. Geometric mean: the appropriate method for obtaining an a:verage on a ratio scale. See Pennypacker, Koenig, and Lindsley (1972) for a description of how to determine the geometric mean on the Standard Behavior Chart. Movement: recorded behavioral event. Usually specified in terms of movement cycle with a beginning, middle, and end. Multiple schedule: A schedule in which reinforcement is programmed by two or more simple schedules alternating, usually at random, each schedule being accompanied by an appropriate stimulus as long as the schedule is in force. The simple schedules of a Multiple schedule are often called components of the Multiple schedule.

PAGE 20

•12Rate: basic unit of behavioral measurement. Generally, number of movements per unit of time. Specifically, movements per minute; the measure found on the Standard Behavior Chart. Schedule control: When performance appropriate to a particular schedule is generated by a stimulus which is not a primary reinforcer, the stimulus may be said to exert schedule control. Schedule of reinforcement : a systematic arrangement of response-reinforcement contingencies which vary the proportion of— reinforced responses or the time intervals after which reinforcement is obtained. Standard Behavior Chart: a six-cycle logarithmic chart designed for the graphic display of ratio data such as rate. Tandem schedule: A schedule of reinforcement in which a single reinforcement is programmed by two or more schedules acting in succession without correlated stimuli. Target behavior: see movement. Typically, this would be the dependent variable. Token reinforcement: usually a metal or plastic disc which an organism can carry around and exchange for privileges, food, or other items. VariableInterval schedule (VI) : the same as FI except that the time from a reinforcement to the availability of the next reinforcement is variable.

PAGE 21

-13Variable-Ratio schedule (VR) : the same as FR except that the nvmber of responses required for each reinforcement is variable, not fixed. Limitations Two limitations of this study may have influenced the results in a negative way, and one limitation influenced the generality of the findings. First, precise measurement of rate of response was impossible since the reinforcement delivery time was included in the total time used to calculate rate. It was expected that when compared with an Fl schedule a VR schedule of reinforcement would generate higher rates of response. However, since the reinforcement density was greater during the VR than during the FI, there was less total time in which the subject could respond. This would affect the rate calculation. Second, it is not clear whether the verbal statements used by the white tutors as reinforcers were in fact reinforcing to the black subjects. The subjects may have lowered their rate of response in order to reduce the frequency of reinforcement. This may also explain the negative results. In this regard, the term reinforcer is used throughout the text, although this is not meant to imply that the events used as reinforcers were in fact reinforcing. Third, since there were only four subjects, there is no intent to generalize the results to any other subjects or populations.

PAGE 22

CHAPTER II LITERATURE REVIEW This chapter integrates historical perspectives of social reinforcement research with the procedures of this study. The major developments found through review of the literature are (1) investigations of social reinforcement satiation and deprivation effects on conditionability; (2) evaluation of subject characteristics and their influence on the effectiveness of social reinforcers; (3) examinations of experimenter effect; (4) delineation of informative and affective roles of social reinforcement; and (5) systematic manipulations of the relationship between the response and reinforcement delivery. After indicating these developments, the literature review examines schedule research on the social reinforcement of reading anddiscusses the procedural requirements of such research. Major Developments in Social Reinforcement Research The five major developments represent differing views of the nature of social reinforcement. The purpose here is to present studies which most clearly exem.p 1 if y those beliefs. In this regard, no attempt was made to summarize all experiments using social reinforcement. Instead, studies were selected which illustrate typical practices. -14-

PAGE 23

-15Satiation and Deprivation In the past, theorists speculated that satiation and deprivation had opposing effects on social reinforcer effectiveness. Availability of social stimuli was said to determine the probable influence of reinforcement. Presumably deprivation enhanced the value of social reinforcement while satiation reduced the value. If so, then, reinforcement subsequent to a period of deprivation would result in a significantly greater production of behavior than would reinforcement after a period of satiation. A common method of inquiry was to compare responsivity to reinforcement after isolation and non-isolation. Typically an "isolated" subject waited alone in a waiting room for a period of time prior to performing the experimental task while an adult remained with "non-isolated" subjects. The adult provided reinforcement dur3.ng the task performance. The performance of the "isolated" subjects would be expected to be better than that of the "non-isolated" subjects. Studies like Erickson's (1962) and Gewirtz and Baer's (1958a, b) provided supportive results . The unstimulating conditions of institutions for the retarded have also served as settings for social reinforcement research. Klaber, Butterfield, and Gould (1969) in a unique study compared the responsiveness of retarded persons from two such institutions; one Institution (A) which allowed social interaction, and Institution (B) which did not.

PAGE 24

•16Interestingly enough, subjects selected from Institution A had in the recent past been residents of Institution B. Children from both institutions were placed in a room with an adult and some toys. The data showed that Institution B subjects played for longer periods of time and maintained less distance from the adult than did Institution A subjects. Thus, the transferred subjects who had moved from deprived to less deprived cir exams tances appeared to need social interaction less than the subjects who remained in the low social interaction institution. One important interpretation of these results might be that the effects of social deprivation are reversible. The study which demonstrated most succinctly the opposing effects of satiation and deprivation was the Landau and Gewirtz (1967) study. A critical feature of the design made the clarity of their results unequaled. Satiation levels were experimentally controlled and their effects on performance measured. "Well done's" administered at four frequencies, 4, 12, 30, or 60, consequent to correct discriminations describes the procedure. Subjects received one of the four frequencies of reinforcement during the satiation period and subsequently during the task performance. The greater the frequency of reinforcement during the satiation period the less effective was the reinforcement on task performance. Furthermore, subjects who had received

PAGE 25

-17the lower frequencies of reinforcement, 4 or 12, were conditionable, while the ones who received the higher frequencies, 30 or 60, were not. These results demonstrated that the effectiveness of a reinforcer may vary inversely with the frequency of its prior delivery. S ubject Characteristics and Their Influence on Social Reinforcer Effectiveness This section discusses conclusions drawn from the body of research which investigated individual characteristics, Experimenters undoubtedly perceived that attributes native to the person were primary determinants of social reinforcer effectiveness. Research in this area attempted to establish which social reinforcer or set of reinforcers modified behavior most successfully, given an array of particular characteristics. Categories reviewed are (1) Personality, (2) Race, (3) Social Class, (4) Intelligence, and (5) Age. Personality The many dimensions of personality have been written about elsewhere (Hall and Lindzey, 1957). A few of these dimensions are dependence and independence, high and low self -concept , positive and negative self-image, introversion and extr avers ion, and internal and external locus of control. Placement along a continuum from one extreme to another was theorized to govern one's need for affection, reinforcement, and the like. It is seen that personality dimensions can in part determine the need system of an

PAGE 26

-18individual which may relate to a need for particular kinds of feedback from others./^ A crucial inference arose out of the social reinforcement research on personality. Personality, as a construct, was thought to determine the likelihood that a specific reinforcer affected behavior. That is, if the social reinforcer matched the needs of the personality, then it was likely to have a positive effect on behavior. If, however, the feedback created cognitive dissonance, improvement was less likely to occur. Thus, if the content, intensity and timing of the input was discrepant from the individual's needs at the moment it would be discredited or noninfluential (Baron, Bass, and Vietzl, 1971; Todd and Nakamura, 1970). Race Family patterns and reinforcement styles may go hand in hand. Whereas there might be commonality among races, it seems evident that racial heritage exerts a strong influence on the development of family patterns (Hess and Shipman, 1965). Baron, Heckenmueller, and Shultz (1971) examined black and white responsiveness to reinforcement styles by comparing the efficacy of two kinds of praise on performance: task oriented and person oriented. Task oriented feedback made reference to the quality of the task performed while person oriented commented on the "goodness" of the person performing.

PAGE 27

-19In their study, blacks and whites responded differentially to the two forms of praise. Blacks given person oriented approval performed significantly better on the assigned task than those given task oriented feedback. Just the reverse was true for the white subjects: task oriented was more successful than person oriented. Although this study did hot provide evidence to assvmie racial differences exist in family reinforcement patterns, it did support the notion that blacks and whites differ in their responsiveness to two forms of verbal feedback. Social class Social class has been found to influence health, intelligence, and personality. Thus, investigators have assumed there might be some relationship between social class and one's susceptibility to social reinforcement. Marshall (1969) explored the effect of delayed and immediate feedback on the task performance of one hundred and sixty high and low class subjects. Subjects from both social classes learned equally as well with immediate verbal reinforcement. Low class subjects performed better with immediate than with delayed feedback while the high class subjects perfonned equally as well with either. Reiner (1974) conducted a comparative study of different reinforcers, social, material, or knowledge of results, on middle and lower class children. Subjects performed problem solving and motor learning tasks in each reinforcement

PAGE 28

-20condition. A dynamic interaction existed between social class, type of reinforcement, type of task, and grade level. Reiner concluded that the most potent reinforcers for particular types of children on specific kinds of tasks can be determined. These results stimulate thought about implications for training programs, individualized educational planning, and exposure to different reinforcement conditions. Intelligence Level of intelligence has been a discriminating factor in education since Binet developed his scales. In fact, it is common practice today for placement into a variety of regular and special education classes to be determined primarily by one's measured intelligence on group or individually administered tests. With an objective criteria for placement established, it seems reasonable for educators to search for motivational strategies for groups of children so readily identifiable. Van de Riet (1963) attempted to find just such a strategy by researching the effectiveness of praise and blame on the performance of educationally retarded and nonretarded children. Blame was found to be significantly more ef f ect5.ve with retarded children while praise worked best with non-retarded children. Moreover, clinical observation suggested that praise produced anxiety in the retarded subjects. It might be concluded that blame created less

PAGE 29

-21cognitive dissonance and therefore became a more effective incentive for retarded subjects than praise. Prominent researchers in the area of social reinforcement and the intellectually retarded. Panda and L3nich (1972), established strict guidelines for the choice of appropriate feedback used with retarded pupils in educational settings. Criteria were that the social reinforcers (1) not be redundant, (2) be contingent upon behavior, and (3) be reliably structured. Noncontingent occurrence of feedback was thought to inhibit performance. Thus, while verbal support may have been a necessary condition for learning, its effectiveness was enhanced by the contingent presentation of diverse social reinforcers which provided information about the quality of performance. A£;e Much has been written about the intellectual, physical, emotional and moral development of children (Mussen, 1973). Developmental psychologists have extensively described the behavior patterns inimical to particular age stages in a child's life. A little known area of developmental research examined changes in perceptions of the meaning of social reinforcers from childhood to adulthood. Solomon and Ali (1972) surveyed individuals from childhood to adulthood on their use of either content or intonation as the primary determinant of the meaning of a social reinforcer. Their data yielded distinct age trends.

PAGE 30

-22Content seemed more important for younger subjects, whereas intonation was fundamental in the adult subjects* evaluation of the meaning of the reinforcing stimuli. Summary To simmarize the section on subject characteristics, the following tentative conclusions are offered. (1) For social reinforcement to be effective its content, intensity, and timing must be appropriate to the momentary needs of the individual. (2) Racial differences in responsiveness to some kinds of verbal feedback may exist. (3) Criteria for the selection of social reinforcers are that the reinforcer not be redxmdant but be contingent on behavior and reliably structured. (4) Developmental trends appear in perceptions of the meaning of social reinforcers. It is inferred that the relationship between the subject and the effectiveness of a social reinforcer is a complex one whose determinants seem to vary from moment to moment. Experimenter Effects The now famous Rosenthal studies examined the influence of the experimenter on the results of psychological and behavioral research (Rosenthal, 1966), Several conclusions can be stated regarding experimenter effect on subject behavior. These may have implications for a study in which an adult interacts with a child as the treatment variable. The conclusions are as follows: (1) Subjects extract information from the investigator's attributes in order to

PAGE 31

-23respond in their preconceived "right" manner; (2) The experimenter as a person and what has happened to him before or happens during the experiment affects the subject's responses. The subject's responses subsequently affect the experimenter's behavior which in turn can alter the subject's response; (3) Characteristics of the experimenter influence the subject's behavior in a direct way but sometimes interact in a subtle fashion with subject characteristics, task characteristics, or situational characteristics; and, (4) The principal investigator must rely on the process of replication in order to control for experimenter effects. In a unique application of Rosenthal's work, Leventhal and Fischer (1970) attempted to determine whether differences in performance could be attributable to characteristics of the experimenter per se or to his reinforcing statements. Ninetysix subjects were exposed to experimental conditions in which the experimenter was seated near and away from the subject and did not reinforce, or in which he was seated near the subject but did reinforce. Dependent variables were rate and preference of placing marbles in holes. The results showed that the subject's rate differed even before reinforcement was initiated. Therefore, it was concluded that reinforcement itself was not solely responsible for the condition differences.

PAGE 32

"24It would appear that preference changes reflected the subject's effort to extract information from the experimenter's behavior, while base rates and rate changes reflected complex alterations in the subject's emotional state. . . It is difficult not to express some surprise at the unimportance of reinforcement . (p . 92) The authors qualify their conclusions by stating that "perhaps reinforcement effects would be noticeable if alterations were made in the procedure and experimental design"(p. 92). The remainder of this review discusses studies in which reinforcement did make a difference. Informative and Affective Role s Social reinforcement has been said to serve two purposes to inform about response accuracy and to create a positive feeling (Panda, 1971). A positive feeling is created when a person is made to feel good in order to be confident of making an appropriate choice. Response accuracy occurs when a person is informed whether their choice was accurate in order to continue successful choices. Regardless of divergency of purpose, the objective of both purposes of reinforcement is to motivate the subject to perform with greater accuracy. Informative role Information theory states that when motivating children the cueing and signalling aspects of reinforcement are more important than any feelings aroused. Investigators have compared the effectiveness of transmitting correct.

PAGE 33

incorrect, or no information to subjects after subject responses (SiegeX and VanCara, 1971; Meyer and Offenback, 1962; Spence, 1966). Typically information is given either verbally or with non-social cues, like marbles, subsequent to each task response. Experimenters divided subjects into three groups labeled "Rb," Wb," and"RW." In the "Rb" group subjects were informed when they were right but not when they were wrong. In the "Wb" condition only the wrong answers were consequated. In the "RW" group isubjects were informed of both correct and incorrect responses. Probably the most important educational application of information theory was a study by Panda (1970) in which eighty educable mentally retarded boys assigned to one of two conditions were tested on a concept learning task. Subjects received either RW information or RW plus noncontingent verbal support after each block of four trials . RW subjects, performed with significantly higher accuracy than did those who received RW information plus verbal support. Unconditional support seemed to serve as a dis tractor. Therefore, it appears that a reinforcer intended for one purpose, to make the subject feel good, may be counterproductive for another purpose, to improve the subject's performance. However, because noncontingent verbal support was found to be distracting does not imply that the same interpretation applies to contingent verbal support.

PAGE 34

-26A ffective role Affective feedback was intended to create either a positive or negative feeling in the child. In turn the feeling was thought to motivate the child to perform at a faster rate or higher accuracy. Drive reduction theory provides a frame of reference for understanding the role of affective feedback. Within such a framework, the subject performs to achieve a positive feeling or to reduce the presence of a negative feeling. Thus, instead of providing information the intent was to support or criticize the person so that performance would improve. Studies, then, attempted to determine whether praise or blame worked best for a particular type of subject or task (Binet and Vaschide, 1897; Leith and Davis, 1969; Kelly and Stevens, 1964; Hurlock, 1924). Relationship Between the Response and Reinforcement Delivery Developments thus far reviewed have conceptualized the nature of reinforcement on a theoretical basis. Investigators attempted to explain why social reinforcers had a reinforcing effect on behavior. This section attends to the empirical conceptualization. Empiricism concerns itself primarily with the effect. There is considerable emphasis to determine what environmental event may have precipitated the effect, but only minimal interest in understanding why (Sidman, 1960).

PAGE 35

M~ The shift in conceptualization from theoretical to empirical has correlated with methodological modifications in research strategy. There was a movement away from pre/postlike experimental designs to designs which prioritized the control of a phenomena. Although many studies exist which utilized group statistical designs, this section discusses only those which used single subject designs. Two basic methods of manipulating social reinforcement in order to investigate the relationship between the response and reinforcement delivery will be discussed. These methods were the presentation and withdrawal technique and the more complex schedule of reinforcement paradigm. Presentation and withdrawal The presentation and withdrawal technique has assumed some prominence in both research and applied settings. It is described here as it relates to the modification of social and academic behavior in the classroom. First, the research technique will be elaborated upon. Then, a group of studies which clearly exemplifies the practice will be examined. A single subject design, which is a one-to-one correspondence between the emission of target behavior and reinforcement, in addition to the baseline technique, characterized this research strategy. An idiographic model allowed the investigator to look at the effect of treatment

PAGE 36

-28• variables on the individual subject. A study typically included more than one subject but results were discussed relative to each individual ' s responsivity. Social reinforcement was either delivered for each occurrence of the desired behavior and/or withheld if an landesired behavior occurred. A general notion of the technique used to demonstrate reliable control of a behavioral change will be set forth. First, data was collected on the frequency of the target behavior over a period of time establishing a baseline level of its occurrence against which treatment was applied and its effect recorded. Third, the original baseline condition was reintroduced meaning that treatment was withdrawn. If the final frequency of behavior approximates that of the original baseline, it is inferred that the experimental treatment precipitates the effect observed in the treatment phase. Within this framework, researchers conducted studies in the classroom. The modification of aberrant social behaviors has increased classroom harmony and served to make children more acceptable to each other and adults (Thomas, Nielsen, Kuypers, and Becker, 1968). Several studies set as their goal the elimination of disruptive behavior and the increased production of on-task performance (Wasik, Senn, Welch, and Cooper, 1969; Thomas, Becker, and Armstrong, 1968; Broden, Bruce, Michell, Carter, and Hall, 1970; Hall, Lund, and Jackson, 1968) . Idiosyncratic problem behaviors

PAGE 37

-29of children such as crying (Allen, Hart, Buell, Harris, and Wolf, 1964), isolate behavior (Harris, Wolf, and Baer, 1967), passive play (Hart, Allen, Buell, Harris, and Wolf , 1964), and non-cooperative play (Redd, 1969) have been successfully treated as well. Academic behavior modification has received far less research attention than the modification of social behavior. One notable exception, though by no means the only one, a study by Hasazi and Hasazi (1972) , investigated the digit reversal behavior of an elementary school child. Baseline observation found that the teacher responded to reversals and ignored correct digit placement. Not atypically, she marked reversals incorrect and gave "extra help" on all sums incorrectly written. For the treatment the teacher was instructed to socially reinforce correct digit placement and ignore reversals. The results showed a sharp decrease. It is concluded from the research on the presentation and/or withdrawal of social reinforcement that the technique represents a powerful tool for modifying behavior as well as a viable research paradigm for the demonstration of the control of social and academic behavior. Schedules of social reinforcement In contrast to simply presenting or withdrawing social reinforcement contingent upon the occurrence of a behavior, schedules represent an intricate way of varying the presentation of reinforcement in relation to the response.

PAGE 38

-30As previously defined, a schedule is a systematic arrangement of response-reinforcement contingencies which vary the proportion of reinforced responses or time intervals after which reinforcement can be obtained. The resulting effect of such manipulation has been called the patterning effect. Due to methodological problems indigenous to such research, there have been few studies (Lahey and Drabman, 1974; Staats et al. , 1962; Chadwick and Day, 1971). The next section provides an overview of the broader topic, schedule research on reading. This topic includes but is not restricted to a survey of token schedule research and a discussion of the procedural application of such studies to one which investigates social reinforcement schedule effects. Such review provides a framework for resolving methodological problems. Summary of Historical Developments The historical developments reviewed included satiation and deprivation effects, subject experimenter influence, characteristics, informative and affective roles of social reinforcement, and the relationship between the response and the delivery of reinforcement. Each of these developments was presented within the context of their differing views of the nature of reinforcement. Attention was paid to the methodological and procedural components of studies which investigated those views. The next section integrates

PAGE 39

schedule research of reading with the procedural requirements of one which uses social reinforcement. Schedule Research on Reading Studies which manipulated token and social reinforcement schedules will be discussed in view of the application of methodological characteristics pertinent to the present investigation. Schedules of Token Reinforcement Token reinforcement is generally used within a system called the token economy (Ayllon and Azrin, 1968). The token economy refers to an economic system of payment for appropriate behavior in which the tokens earned can be saved and later exchanged for valuables. Tokens, then, can be exchanged for objects of value much the same as currency is exchanged for goods in our society. Their worth is enhanced by their relevance to real currency, by the fact that the subject can manipulate and transport them, and by their readily \jnderstandable value. One particular characteristic of tokens make their use as a reinforcer advantageous in research (O'Leary and Drabman, 1971). The delivery of the token is a discrete event which has a definite effect on the environment. Therefore, unlike food and social reinforcement, it can be reliably dispensed and counted. While Staats and his associates have conducted by far the most schedule research on reading behavior (Staats,

PAGE 40

-32 Staats, Schutz and Wolf, 1962; Staats, Finley, Minke, and Wolf, 1964; Staats, Minke, Finley, Wolf, and Brooks, 1964) other researchers have made significant contributions as well, A few studies that have a direct bearing on the present investigation will be presented. Lovitt, Eaton, Kirkwood, and Pelander (1971) investigated the effects of four reinforcement contingencies on oral reading. Subjects responded to the contingencies while reading from an experimental reader. Results were compared with reading rates of the same subjects from control readers. The experimental contingencies were designated as follows. In Experiment I if the pupil's correct rate exceeded and error rate was less than prescribed limits he was reinforced on a 25:1 ratio . For every two errors in Experiment II one token was taken away. In Experiment III for every correctly read word the pupil received one token. In Experiment IV the pupil earned one token for every thirtieth correct response. No tokens were delivered when the subjects were reading from the control readers. Procedurally, the pupil read from each book five minutes per day. The order in which he read the books was randomly alternated. Errors were handled the same throughout all Experimental groups. That is, if the pupil mispronounced a word, he was immediately told the word. Tokens could be exchanged for minutes of free time at the end of the ten minutes of reading from the control and experimental readers.

PAGE 41

-33The results are suinmarized as follows. In Experiment I, where the contingency affected both correct and error responses, both types of responses were influenced. In Experiments II and III, the contingencies focused on errors, and errors were more influenced than were correct responses. In Experiment IV where the contingency was placed on correct responses, correct responses were influenced in the first half of the experiment but not in the latter half. In most cases, both correct and error rates in the control reader were not as positively influenced as in the experimental reader. Sibley (1966) investigated reading rate and accuracy as a function of Fixed Ratio schedules of token reinforcement. Seven subjects of normal intelligence who were designated as poor readers participated in the experiment. There were three phases in the experiment. In Phase I nonexchangeable points were delivered on a Fixed Ratio for correct reading responses. In Phase II non-exchangeable points and tokens were delivered simultaneously on a gradually increasing Fixed Ratio schedule. Tokens were exchangeable for back-up reinf orcers . The final phase was a return to Phase I conditions. Points were not found to be initially reinforcing, but became strikingly so subsequent to being paired with tokens. Four of the seven subjects showed increased reading rates as the FR was increased. Three showed the reverse trend.

PAGE 42

-34Some evidence was found to indicate that high ability subjects read faster on the higher. Fixed Ratios. Hoeltzel (1973) investigated the effects of single and multiple-ratio schedules of token reinforcement on reading rates. Like the Lovitt et al. (1971) study the Experimental and Control reader approach was used. Four subjects read five minutes per day from each book. There were no contingent phases with the Control reader but with the Experimental reader contingent and noncontingent phases were alternated weekly for seven weeks. Generally, the results indicated increased correct responding during contingent phases of the Experimental reader as well as an overall tendency toward increased error acceleration. These results generalized to the Control reader only when the contingent phase was in effect with the Experimental reader. It is seen from the review of token schedule studies that results generally support the notion regarding patterning effect of schedules. That is, schedules produce particular effects which can be observable across reinforcers, behaviors, and subjects. Schedules of Social Reinforcement While there have been few studies which investigated social reinforcement schedule effects, the Mays (1970) study used social reinforcement schedules to determine their influence on error production. This investigation looked at the effect of schedules of reinforcement on the

PAGE 43

elimination of errors in oral reading. Social reinforcement, "Fine" and "Good," was delivered on either a VI or VR schedule. The VI schedule v/as programmed by tape and the VR by penciled marks in the book being read. The response xinit was a sentence and reinforcement was delivered if the entire sentence was correct. Although correct sentences were reinforced, the dependent variables in the Mays study were rate of error production and general error strength. Therefore, there was no attempt to analyze the direct influence of the schedule of social reinforcement on the response being reinforced. Procedural Requirements of a Study of Social Reinforcement Schedule Effects on Reading The application of procedures from studies of token reinforcement schedule effects to a tutoring situation in which social reinforcement is used is complex. This section will discuss three pertinent procedural aspects which need to be ascertained before such application can begin. These three procedural needs are (1) the importance of the discrete delivery of the reinforcer, (2) clarification of the nature of the reading response, and (3) the need to methodologically control the learningto -learn phenomenon. The application of these remarks will be presented in Chapter III.

PAGE 44

Importance of the discrete delivery of the reinforcer Ayilon and Azrin (1968) stated that the following rule should govern the selection of a conditioned reinforcer. It should "provide a distinctive and tangible stimulus event to bridge any delay between the desired response and the deliveiry of the reinforcer" (p. 77). Their work. The Token Economy , outlines the advantages of tokens as conditioned reinforcers. These advantages are (1) delivery is quantitatively related to the amount of reinforcement, (2) portability, (3) possession in infinite amounts, (4) usefulness to operate machines, (5) durability and tangibility, (6) ease of standardization of physical characteristics, (7) indestructibility, (8) non-duplicability, (9) reinforcement of response at any time, and (10) reinforcement occurrence without interruption. Social reinforcement possesses some of these advantages but also has some critical disadvantages. The advantages are (1) it represents an immediate stimulus following a response, (2) does not disrupt performance, and (3) allows the response to be strengthened. The disadvantages include the following: (1) it is a momentary and, therefore, intangible event, (2) the qualitative dimensions do not allow for standardization, and (3) it cannot be used to operate machines . Nature of the reading response There are many definitions of reading. The definitions range from reading as comprehension to, reading as decoding

PAGE 45

-ayes inger and Ruddell, 1970). Some experts maintain that a child has not read a word until he has comprehended it. Others define reading as the act of saying or decoding a written word. The proponents of the latter definition partial out the decoding and comprehension components of reading as separate behaviors . Decoding and the comprehension of whati was decoded are considered to be two distinct behaviors. The present investigation embraces this concept of reading. It was thought to be critical to the intent of the present research to narrow the response datum to the frequency of words read in time. Sidman wrote in Tactics of Scientific Research (1960) that the response should be one "whose relevant characteristics are easily recordable, with minimal interference from the process of observation itself. [Additionally]. . . there should be minimal constraint upon the subject's rate of responding" (p. 395). The repertoire necessary in order that the child can decode is complex. Words can be counted as discrete responses in time, but words are not equal in complexity. Each word decoded represents a lengthy chain of responses. The differing complexity of the words as units and as they fit in the context of the sentence and paragraph make for possible variability in resulting data. The response can be recorded cumulatively in time but each is not necessarily an instantaneous unitary response (Staats, Finley, Minke,

PAGE 46

-38and Wolf , 1964). Therefore, while the limitation of the response dattom to counting words read simplifies data collection, the attendant experimental results may not be as clear were the response a simple bar press. Control of the learning-to-learn phenomenon The previous section implied that the nature of the reading response induces inherent variability and thus confusion in the data. An additional complication is the need to continually increase the complexity of the reading material as skill^evel improves. Control of this learningto-leam phenomena is important in order that variability in the data be due to the experimental manipulations and not to materials or skill level. Two methods have been devised to assist in controlling for this variability. The first has to do with the selection of materials and the second has to do with the presentation of the materials to the subject. These two methods will be elaborated in the remainder of this section. Material selection . Staats, Finley, Minke, and Wolf (1964) wrote that in order to study experimental effects on long-term reading, materials were needed which do not affect the learning curve. The ideal material would be homogeneous in content, such as programmed reading materials. For this study the Science Research Associates Laboratory la (SRA Lab la) reading materials were chosen (Parker and Scannell, 1961)

PAGE 47

-39. Performance Determined Instruction Model (PDI ) was developed by Gray, Baker, and St ancyk (1969) at the Monterey Institute for Speech and Hearing as an instructional procedure for obtaining changes in reading performance. The methodology combines aspects of binary logic, laws of learning, and programmed instruction. The PDI model was designed as a method for decision making to discriminatie which procedures to employ with the individual subject. A host of procedures was developed by the Institute to help the child read at the desired reading level. One procedure, related to the acquisition of sight word vocabulary, is described here. The application of this procedure to this investigation will be presented in Chapter III. This precision instruction approach makes several assumptions regarding the nature of reading which relate to the development of the training procedures. First, reading is considered to be a learning task and, therefore, subject to the laws of learning. Second, decoding the word is antecedent to comprehension and therefore should be taught first. Third, a training system should provide frequent and immediate reward for successful performance. Fourth, characteristics such as visual perception, intelligence, visual motor coordination, visual and auditory sequencing and the like are considered not to affect differentially the prognosis for skill mastery unless the deficiency is of such a marked and gross degree that it results in reduced functional ability in nonreading activities.

PAGE 48

-40The procedure for the acquisition of sight vocabulary incorporates a computer logic system known as TOTE (TestOperation-Test-Exit) which was originally conceived by Miller, Galanter, and Pribram (1960). Basicly the steps within a TOTE system occur similarly to the following sequence. First, the behavior in question is measured against some standard. Second, if the behavior is discrepant to some specified degree from the standard an operation is performed to help the behavior reach the goal. Third, the behavior is tested again. If the behavior reached the standard, the program is terminated. If, however, the behavior did not reach the standard, the behavior is treated. The TOTE system provides a structure for decision making. More specifically the TOTE system was applied in the PDI model to improve sight word vocabulary performance. This application will be described in some further detail. A target performance of 92 percent accuracy plus or minus 2 percent was set as the standard against which to measure performance. Performance was measured against the standard and operations were chosen if the performance was either above or below the accuracy range. Questions were asked and based on the answer decisions were made regarding whether to terminate, to reduce task difficulty, or to increase task difficulty. The following questions or steps were devised to control the performance accuracy of sight word vocabulary performance. First, was the performance accuracy within

PAGE 49

-41the range prescribed for three successive sessions? If the answer was yes. Question 2 was asked. Was the reading age level up to the target age level? If yes , training was terminated. If the answer to Question 1 was no, Question 3 was asked. Was the performance above the range; that is, were three successive sessions above 94 percent? If the answer was yes, then, the subject was advanced to the next level of difficulty. If the answer to Question 3 was no; that is, three successive sessions were below 90 percent, then a decision was made to move to the level on which performance was last within the acceptable accuracy range. The next chapter will describe how the TOTE system and PDI model for sight vocabulary building was applied in this study. As indicated, it was considered necessary to develop procedures which would control the learning -tolearn phenomenon. Procedures maintaining accuracy level within a particular range were necessary to control variability in performance to such a degree that treatment effects might be more readily observable. It was important to describe the procedures here in detail as a frame of reference for understanding the unique procedures designed for the current study.

PAGE 50

CHAPTER III METHODS AND PROCEDURES The purpose of this chapter is to present those methods and procedures selected in order to investigate whether there is a predictable relationship between Multiple schedules of social reinforcement and subjects' mean rate of responses on Flash Card and Oral Reading tasks. Chapter I defined a Multiple schedule as one in which two or more simple schedules randomly alternated, and in which the occurrence of the simple schedules was accompanied by a stimulus. This study used tutors as the stimuli associated with the different simple schedules. Three experiments were designed to investigate the relationship between Multiple schedules and rate of response. In each experiment the Multiple schedule consisted of two simple schedules. Comparisons were made between the data derived from the subjects' rate of response in the presence, of the two simple schedules. Experimental procedures will be discussed in more detail in the final section of this chapter. The contents of Chapter III will be presented in the following sequence: (1) Research hypotheses, (2) Subjects, (3) Tutors, (4) Remedial Reading program, (5) Setting, (6) Apparatus, (7) Materials, (8) Data Collection, (9) Data for Data Analysis, and (10) Procedures. This latter section -42-

PAGE 51

-43includes those procedures common to all three experiments. These procedures are (1) Flash card procedures, (2) Oral reading procedures , (3) Adaptation of the Performance Determined Instruction model, and (4) Point reinforcement procedures. Finally, the chapter concludes with an overview of the experimental procedures unique to each experiment. Research Hypotheses A brief introduction to each experiment will be followed by a list of research hypotheses . The introduction will include the nature of the two simple schedules compared and a description of the experimental design used in each experiment. This information should provide a framework in which to understand the research hjrpotheses. Like the Staats (Staats, Finley, Minke, and Wolf, 1964) study reviewed in Chapter I, the research hypotheses will state which schedule is predicted to generate the higher rate of response. Experiment 1 Experiment I examined Multiple schedules of social reinforcement in which both simple schedules reinforced correct responses. Three subexperiments make up Experiment I, There were two experimental phases in each subexperiment with approximately 16 sessions in each phase. The simple schedules , a Variable-Ratio (VR) schedule and a FixedInterval (FI) schedule, assigned to tutors in Phase I were switched in Phase II.

PAGE 52

-44Research hypotheses This experiment tested six hypotheses relevant to Multiple schedule effects on Oral Reading and Flash Card responses. 1. Subexperiment I-l Flash Card Phase I and II mean VR correct rates will be higher than the mean FI rates. 2. Subexperiment I-l Oral Reading Phase I and II mean VR correct rates vTill be higher than the mean FI correct rates . 3. Subexperiment 1-2 Flash Card Phase I and II mean VR correct rates will be higher than the mean FI correct rates. 4. Subexperiment 1-2 Oral Reading Phase I and II mean VR correct rates will be higher than the mean FI correct rates, 5. Subexperiment s 1-3 Flash Card Phase I and II mean VR correct rates will be higher than the mean FI correct rates. 6. Subexperiment 1-3 Oral Reading Phase I and II mean VR correct rates will be higher than the mean FI correct rates.

PAGE 53

.45Experiment II , A contrast design is used in Experiment II to demonstrate the effects of Multiple schedules of social reinforcement. In this two phase experiment, the following contrast design is employed: (1) the two Phase I simple schedules are identical and (2) one of the simple schedules is changed in Phase II. This change is said to produce a contrast effect defined in Chapter I, Phase I simple schedules were both a Variable-Interval (VI) schedule. One of these schedules was changed in Phase II to a VR schedule. The tutor with the lower average rate in Phase I changed to a VR schedule in Phase II. Research hypotheses This experiment tested four hypotheses relevant to Multiple schedule effects on Flash Card and Oral Reading responses. 1. The median of the differences between correct rates of the two schedules in Phase I for Flash Card and Oral Reading is zero" (p = .05). This null hypothesis will be tested by the sign test (Siegel, 1956).

PAGE 54

-462. Experiment II Phase II mean VR rate will be higher than the mean VI rate for Flash Card and Oral Reading. 3. Experiment II Phase II mean VR rate will be higher than the mean VI rate for that tutor in Phase I for Flash Card and Oral Reading. 4. Experiment II Phase II mean VI rate for the iinchanged tutor will be lower than the mean VI rate for that tutor in Phase I for Flash Card and Oral Reading. Experiment III Experiment III investigated the effects of Multiple schedules of social reinforcement in which both simple schedules were the same but one reinforced correct responses and one reinforced error responses. Two subexperiments make up Experiment III. The simple schedules used in Subexperiment III-l were both an FI schedule. In Sub-^ experiment III-2 both schedules were a VI schedule. Each subexperiment was divided into two experimental phases. If assigned correct responses to reinforce in Phase I the tutor reinforced errors in Phase II and vice versa. Research hypotheses This experiment tested nine hypotheses relevant to Multiple schedule effects on Oral Reading and Flash Card responses.

PAGE 55

-471. Subexperiment III-l Flash Card Phase I and II mean correct rates for the tutor that reinforced corrects (RC) will be higher than the mean correct rates for the tutor that reinforced errors (RE) . 2. Subexperiment III-l Flash Card Phase I and II mean RE error rates will be higher than the mean RC error rate. 3. Subexperiment III-l Oral Reading Phase I and II mean RC correct rates will be higher than the mean RE correct rates , 4. Subexperiment III-l Oral Reading Phase I and II mean RE error rates will be higher than the mean RC error rates. 5. Subexperiment III-2 Flash Card Phase I and II mean correct rates in the RC component will be higher than the mean correct rates in the RE component . 6. Subexperiment III-2 Flash Card Phase I and II mean error rates in the RE component will be higher than the mean error rates in the RC component. 7. Subexperiment III-2 Oral Reading Phase I and II mean RC correct rates will be higher than the mean RE correct rates. 8. Subexperiment III-2 Oral Reading Phase I and II mean RE error rates will be higher than the mean RC error rates.

PAGE 56

-48Subjects Four eight-year-old third grade pupils from an elementary school in Gainesville, Florida served as subjects. All subjects were black, two were female and two were male. The four subjects who participated in the study were selected from a pool of 60 subjects. Two third grade teachers were asked to select pupils from their classes v7ho in their estimation were able to read at the mid-first grade level. The teachers picked seven pupils. This number X7as reduced to the final four necessary for the study. In order to be selected, the following criteria had to be met: (1) a grade level performance between 1.3 and 1.6 on the Lee-Clark Reading Test, and (2) 90 to 95 percent oral reading accuracy on a 1,5 grade level SRA Lab la story. Each of the final four met these criteria. Tutors Seven undergraduate University of Florida students served as tutors. Two were male and five were female. The students received university course credit for their participation in the study. Tutors took part in a ten hour training program prior to the beginning of the experiment proper. The program included a didactic overview of methods and procedures, supervised practice of the tutoring of other tutors and

PAGE 57

-49of the tutoring of nonexperimental pupils. Practice administering the procedures described in the latter portion of this chapter was supervised by the experimenter. Remedial Reading Program The subjects attended a remedial reading program directed by the experimenter. Daily hour long sessions were divided into 40 minutes of task performance and 20 minutes of play. The tasks were administered in the following sequence: flash card drill, oral reading, written comprehension exercises, and phonetic drill. Points earned during these tasks could be exchanged after forty minutes had elapsed. The flash card and oral reading performance provide the data for this study. Setting The setting was a room in Kirby-Smith Elementary which had been divided into seven tutoring stations. These were separated from each other by plywood partitions. In the remainder of the room were a play area and a materials center. Activities available in the play area included games , art materials , and comic books . ? Apparatus Apparatuses included timing devices and a point dispenser. Each tutoring station was equipped with either a standard stopwatch or an electromechanical timing device.

PAGE 58

-50These were used to clock task duration precisely and to aid in delivering points and social reinforcers at the appropriate times. The point dispenser was a self-inking rubber stamp which fit into the eraser end of a pencil. Points were placed on Point sheets specially designed for this study. An example of a Point sheet is shown in Appendix A. Materials Oral reading and flash card supplies were the materials used for this study. Reading selections from the Science Research Associates Reading Laboratory la (SRA Lab la) served as the oral reading materials. SRA Lab la contained seven levels of reading difficulty with 20 stories per level. The levels ranged from the 1.2 grade level to the 3.0 grade level. A flash card unit was made from words appearing in a particular level of stories. Each word in the reading level was represented once in its corresponding flash card unit. Data Collection The Daily Data Sheet (DDS) was designed to allow for the collection of data during the task performance. Space was provided for the tutors to note the time and also to note correct and error responses as they occurred during each task. Thus, the data collected could be easily

PAGE 59

-51corrected into the frequency of responses in time or rate of responding. An example of the DDS is shown in Appendix B. Data for Data Analysis The basic datum in this study was rate of responding. Rate was determined by dividing the frequency of correct or error by the duration of the task. These rates were recorded daily on the Standard Behavior Chart (SBC) and these raw data are shown in Appendix C. The results are expressed in terma_Qf the average rates of response per schedule per phase. Since these data are ratio data, the averages used are geometric means. Procedures Procedures common to the three experiments will be described in this section. These procedures are (1) Flash card procedures, (2) Oral reading procedures, (3) Adaptation of the PDI model, and (4) Point reinforcement procedures. Flash Card Procedures The tutor administered the flash card unit which corresponded with the oral reading level assigned for that day. Flash cards were administered for ten minutes. The following standard procedures were developed. If the word on the card was correctly read within three seconds, the tutor placed the card in the correct pile and presented

PAGE 60

-52the next card. If, however, the word was not read within three seconds or was read incorrectly, the tutor pronounced the word for the subj ect to repeat . The card was, then, placed in the error pile. When five of these error cards had accvmiulated, they were shuffled and represented. The above procedures were in effect until no more than two error cards remained. A new card was then presented. Oral Reading Procedures The subject read one reading selection orally in its entirety per session. Error responses were treated in a manner similar to those which occurred during the flash card performance. That is, if the subject did not read the word within three seconds or read the word incorrectly, the tutor pronounced the word for the subject to repeat. Each error was noted on the DDS as it occurred. Total number of correct responses was calculated by subtracting the total ntmiber of errors from the total number of words in the story. Adaptation of the PDI Model The rationale for the utilization of the PDI model was discussed in Chapter II. This section explicates those procedures which were adapted from the PDI model for the oral reading task of this study.

PAGE 61

-53Each 20 story level in SRA Lab la was divided into four five-story units. Stories one through five, six through ten, eleven through fifteen, and fifteen through twenty made up the five units. Like the PDI model, and accuracy level of 90 to 95 percent v/as established. A series of questions were asked of the data in order to decide whether to increase or decrease the level of difficulty of stories selected. If the subject read the first five stories in a level at an average of 90 to 95 percent accuracy, he continued to the next unit. If, however, the subject read the first five stories below an average of 90 percent accuracy, he re-read those that he had read the least accurately until he had attained an average of 90 to 95 percent for five consecutive days. If the subject read the stories in a unit at an average accuracy above 95 percent, the next unit was skipped unless the unit to be skipped was the first unit of a level. These procedures did not apply to the flash card task. As long as the subject read stories at a particular level, he responded to the flash cards which corresponded to that level.

PAGE 62

-54Point Reinforcement Procedures Point procedures included both their delivery and exchange. All subjects received points on a Fixed Interval 90 second schedule. That is, during each task of the 40 minute work session subjects received a point every 90 seconds immediately after the first correct response. Times for point delivery were listed in the upper left hand corner of the DDS. Point exchange occurred at the end of the session. The points could be spent for minutes of time in the free play area, for pieces of candy, or for polaroid pictures. Exchange ratios were 1 point per 1 minute of free play, 5 points per piece of candy, and 10 points per polaroid picture. Subjects could elect to save points. They were allowed to save any number of points and were allowed to buy no more than 15 minutes of free play, 2 pieces of candy, and 1 polaroid picture per day. Experimental Procedures Three experiments were designed to demonstrate the effects of Multiple schedules of social reinforcement on rate of correct and/or error Flash Card and Oral Reading responses. The previous section described those procedures which were common to each experiment. This section describes the social reinforcement procedures unique to each experiment.

PAGE 63

-55Social Reinforcement Procedures Descriptions of the kinds of verbal statements used as social reinf orcers , the delivery procedures , and the monitoring of their delivery is presented here. Kinds of statements used with the schedules Four groups of verbal statements were selected on the basis of their face validity and grouped into categories labeled High (H) , Low (L) , Consoling (C) , and Disapproving (D) The first two groups were used to consequate correct responses and the Xatter to consequate errors . Samples of these statements are shown in Table 1. Table 2 shows the schedules and the kinds of statements used in each experiment. Explanations of the schedules were provided in the Definition of Terms section of Chapter 1. Delivery procedures Tutors delivered social reinforcement according to a schedule. The schedule to which they were to adhere on a particular day was written in the upper left hand corner of the DDS. If the tutor was to reinforce on a time based schedule, a list of the times when reinforcement was due was provided. If, however, the tutor was to reinforce on a response based schedule, the list contained the n;jmber of responses per reinforcement. Space was provided on the DDS for the tutors to note the occurrence of social reinf orcers.

PAGE 64

-56Table 1. Verbal Statements Used as Reinforcers High (H) Three in a row right Great What a good reader Yeah, you got it That's a girl (boy) Hurray I Low (L) Good Fine Okay That's right Consoling (C) That was close Almost Half right is better than nothing Boy are these tough Maybe next time They've got to get easier You'll get the next ones There's another rough spot Parts of this story are pretty rough Better luck next time Everyone makes mistakes Disapproving (D) Wrong Don't miss it again Not again I just told you that one Try harder

PAGE 65

c 0) u
PAGE 66

-58Monitdring of social reinforcem en t delivery The experimenter monitored social reinforcement delivery by observing the tutQrs and giving them feedback. Two of the tutors were selected at random and observed daily: one during Flash Card and one during reading. A determination was made of the expected number of reinforcers to be delivered on a given task. Actual nximber of reinforcements was, then, counted during the observation. The tutors received written and oral feedback regarding the difference between the actual frequency of reinforcers and the expected number. The feedback was delayed until the end of the entire work session in order that the tutor not know that he or she had already been observed. In summary, this chapter has provided a description of the methods and procedures used in the investigation. The remedial reading program as well as a description of the sample and data collection process were provided. A detailed description of the procedures common and unique to each experiment was presented. The next chapter will present the results of this investigation with respect to the questions asked by the hypotheses outlined in this chapter.

PAGE 67

CHAPTER IV RESULTS In this chapter the results of the study will be presented. The experiments and their respective subexperiments are grouped in units. A brief introduction to each subexperiment is followed by a presentation of its results. Included in these results sections are statements of the acceptance or rejection of the hypotheses relevant to each subexperiment.. Results are presented in tabular form in this chapter and Standard Behavior Charts of the raw data are to be found in Appendix C. The basic question being asked is whether there is a predictable relationship between the Multiple schedule of social reinforcement and the subject's mean rate of responses. Table 3 provides an overview of the results which will be presented in detail in the remainder of this chapter. Experiment I Experiment I examined Multiple schedules of the social reinforcement of correct Flash Card and Oral Reading responses. Three subexperiments were each comprised of two experimental phases. The schedule components assigned to the tutors in Phase I were switched in Phase II. The means of the correct rates by component and phase are presented for Flash Card and Oral Reading tasks. Error rate data was collected; but, it is not reported here since this experiment manipulated correct responses, -59-

PAGE 68

•60CO QJ J3 u o CU M-1 O o •H 4J U 0) •f-1 (U pa u o 0) u c cd •u a. 0) o o < CO cd bO c c H cd cd u (U CKj ^ to (U 0) r-l 43 to cd bO cd ^4-Hx: ox: PM u ex 0) u o <; u -( u (U 1X3 W) H to x: M Cd -OS cd ^x: O 4J Cd (U E cd (U cd cd x: u bOO •H xi-d (U cd r-< cd •HO ^* (U to 4J Cd cdi-i Pd M > o «4-4 c cd (u vU 4-J B Cd M M M > CO C cd cd x: cu pu B l o U 4J l ,H O rHlH ^ M bC (U (U •H 4J to "d cd cd cd U^ cu (l'd cu U CO c cd cd cd O x: o cd (U a> s^ x: M 4J "a C cd cd^H 0) 4J e
PAGE 69

-61c •H 4J o CQ Gi CO 0) rC 4J O a >. m o c o •H 4-1 O 0) 0) P£j ^^ o 0) u CO 4-) Ou (U o o <: CO
PAGE 70

-62" Subexperiment I-l An eight-year-old female served as the subject (S3). Tutors 1 and 2 were two females. The Multiple schedule components were VR3H and FI 10 minute L (FIIO'L). Results Hypothesis 1 stated that the mean VR rate will be higher than the mean FI rate within both phases of the Flash Card performance. Table 4 shows the correct rates and the means of those rates by component and phase for the Flash Card performance of S3. In Phase I the mean VR rate of correct responses was 4.98. The mean FI rate was 7.11. In Phase II the mean VR rate of correct responses was 5.27, and the mean FI rate was 5.56. These data show that the mean VR rate during the Flash Card performance was lower than the mean FI rate within both phases. Therefore, Hypothesis 1 is rejected. Hypothesis 2 stated that the mean VR rate will be higher than the mean FI rate within both phases of the Oral Reading performance. Table 4 shows the Oral Reading correct rates and the means of those rates . In Phase I the mean VR rate was 35.27 v/hereas the mean FI rate was 30.5, In Phase II the mean VR rate was 29.47, and the mean FI rate was 33.92. These data show that (1) the mean VR rate was higher than the mean FI rate in Phase 1, and (2) the mean VR rate was lower than the mean FI rate in Phase II.

PAGE 71

-63Table 4. Mean Rates of Correct Responses in EXPERIMENT I I-l

PAGE 72

-64Therefore, Hypothesis 2 is rejected because the mean VR rate was higher than the mean FI rate in Phase I alone. Subexperiment 1-2 A nine-year-old male served as the subject (S4) . Tutors 1 and 2 were a female and a male respectively. The Multiple schedule components were VR3H and FI 60 seconds L (FI60"L). Results Hypothesis 3 stated that the mean VR rate will be higher than the mean FI rate within both phases of the Flash Card performance. Table 4 shows the correct rates and the means of those rates for the Flash Card performance of S4. In Phase I mean VR rate was 11.46, and the mean FI rate was 11.14. In Phase II the mean VR rate was 9.44, and the mean FI rate was 8.69. These data show that the mean VR rate was higher than the mean FI rate within both phases of the Flash Card performance. Therefore, Hypothesis 3 is accepted. Hypothesis 4 stated that the mean VR rate will be higher than the mean FI rate within both phases of the Oral Reading performance. Table 4 shows the correct rates and the means of those rates for the Oral Reading performance of S4. The mean VR rate in Phase I was 29.41, and the mean FI rate was 31.75. In Phase II the mean VR rate was 28.79, and the mean FI rate was 31.56. Thus, the mean VR

PAGE 73

-65rate was lower than the mean FI rate within Phase I and within Phase II. Thus , Hypothesis 4 is rejected. Subexperiment 1-3 S4 participated in Subexperiment 1-3 subsequent to his participation in Subexperiment 1-2. Tutors 1 and 2 were both females. Multiple schedule components were VR3H and FI 10 minute L (FIIO'L). Results Hypothesis 5 stated that the mean VR rate will be higher than the mean FI rate within both phases of the Flash Card performance. Table 4 shows the Flash Card correct rates and the means of those rates . In Phase I the mean VR rate was 8.16, and the mean FI rate was 8.50. In Phase II the mean VR rate was 8.29, and the mean FI rate was 10.02. Therefore, the mean VR rate was lov7er than the mean FI rate within both phases. Thus, Hypothesis 5 is rejected. Hypothesis 6 stated that the mean VR rate will be higher than the mean FI rate within both phases of the Oral Reading performance. Table 4 shows the Oral Reading correct rates and the means of those rates. The mean VR rate in Phase I was 32.96, and the mean FI rate was 36.91. In Phase II the mean VR rate was 31.13, where the mean FI rate was 40.53. These data show that the mean VR rate was lower than the mean FI rate within both phases. Therefore, Hypothesis 6 is rejected.

PAGE 74

-66Summary of Experiment I Results . Experiment I examined Multiple schedules of social reinforcement in which one of the two components was a VR schedule and the other was an FI schedule. The results of the three subexperiments were discussed relative to the six hypotheses presented in Chapter III. The data clearly demonstrate that (1) Hypothesis 3 was accepted; (2) Hypotheses 1, 2, 4, 5, and 6 were rejected; (3) whereas Hypotheses 1, 4, 5, and 6 were rejected, the FI component maintained a consistently higher mean rate than did the VR component within Phases I and II; and (4) Hypothesis 2 results were in the expected direction in Phase I, but they were not in the expected direction in Phase II. The implications of these results will be discussed in Chapter V. Experiment II Experiment II examined Multiple schedules of social reinforcement across two phases in which (1) the two Phase I schedule components were equivalent, and (2) one of the Phase II schedule components was changed. The data for the experiment are the Flash Card and Oral Reading performance of S3 . Correct rates and the means of those rates are reported. S3 participated in this experiment after her participation in Subexperiment I-l. Tutors 1 and 2 were a female and a male respectively. Both Phase I Multiple schedule

PAGE 75

-67components were VI 60 seconds L (VI60"L) . Tutor 2 changed to a VR3H schedule in Phase II. Results Hypothesis 1 stated that the correct rates of the two Phase I components will be similar for Flash Card and Oral Reading performances. The null hypothesis, the median of the differences is not zero, will be tested by the sign test. Flash Card and Oral Reading results are discussed separately. Table 5 shows the Flash Card Phase I data points and the direction of the difference between the matched pairs. For the data in Table 2, x equals the number of matched pairs who showed differences which equals ten. For N equals 10, an X less than or equal to 1 has a two-tailed probability of occurrence under the null hypothesis of p = .022. This value is in the region of rejection for p = .05; therefore, the alternative hjrpothesis, that there is a significant difference between the correct rates of the two Flash Card components is accepted. Table 6 shows the Phase I Oral Reading data points and the direction of the difference between the matched pairs. For the data in Table 3, x equals the number of fewer signs which equals 3, and N equals the number of matched pairs who showed differences which equals ten. For N equals 10, an x less than or equal to 3 has a twotailed probability of occurrence under the null hypothesis of

PAGE 76

-68p = .344. This value is in the region of rejection for p = .05; therefore, the null hypothesis is rejected and the alternative hypothesis, that there is a significant difference between the correct rates of the two Flash Card components is accepted. Hypothesis 2 stated that the Phase II mean VR rate will be higher than the Phase II mean VI rate for Flash Card and Oral Reading performance. Flash Card and Oral Reading data are discussed separately. The Flash Card data of S3 shown in Table 7 indicates that the Phase II mean VR rate was 7 . 82 and mean VI rate was 9.86. These data show that the mean VR rate was lower than the mean VI rate. Therefore, Hypothesis 2 is rejected for the Flash Card performance. Oral Reading results as displayed in Table 7 show that the Phase II mean VR rate was 30.13 and mean VI rate was 32.86. These data demonstrate that the mean VR rate was lower than the mean VI rate; therefore, Hjrpothesis 2 is rejected for the Oral Reading performance. Hypothesis 3 stated that the Phase II mean VR rate v/ill be higher than the mean VI rate for that tutor (Tutor 2) in Phase I. Flash Card and Oral Reading are discussed separately. According to the Flash Card data presented in Table 7 the Tutor 2 mean VR rate was 7.82 and mean VI rate was 5.09. This indicates that the mean VR rate was higher than the mean VI rate. Therefore, Hypothesis 3 is accepted for the Flash Card performance.

PAGE 77

-69Table 5. Data Points and the Direction of Differences Between the Matched Pairs for the Phase I Flash Card Performance in EXPERIMENT II

PAGE 78

^70Table 6 . Data Points and the Direction of Differences Between the Matched Pairs for the Phase I Oral Reading Performance in EXPERIMENT II

PAGE 79

-71Table 7. Mean Rates of Correct Responses in EXPERIMENT II

PAGE 80

-72The Oral Reading data shoT^ni in Table 7 reveals the following: the Tutor 2 mean VR rate was 30.13 and mean VI rate was 29.43. These data indicate that the mean VR rate was higher than the mean VI rate. Therefore, H3T)othesis 3 is accepted for the Oral Reading performance. Hypothesis 4 stated that the Phase I mean VI rate for the unchanged tutor (Tutor 1) will be higher than the mean VI rate for that tutor in Phase II. Flash Card and Oral Reading will be discussed separately. The Flash Card data (Table 7) for Tutor 1 revealed the following: in Phase I the mean VI rate was 11.72 and in Phase II the mean VI rate was 9.86. These data show a decrease in Tutor I's mean VI rate from Phase I to Phase II; therefore. Hypothesis 4 is accepted for the Flash Card performance. Table 7 shows the Oral Reading performance. Tutor 1 Phase I mean VI rate was 36.51 and the Phase II mean VI rate was 32.86. These data also represent a decrease in mean rate from Phase I to Phase II; therefore, Hypothesis 4 is accepted for the Oral Reading performance. Summary of Experiment II Results Experiment II examined Multiple schedules of social reinforcement and used a contrast design to demonstrate the effects. These results are summarized as follows: (1) Hypothesis 1 was rejected, (2) Hypothesis 2 was rejected, (3) H3rpothesis 3 was accepted, and (4) Hypothesis 4 was

PAGE 81

-73accepted. Although there was a significant difference between the correct rates in Phase I, the introduction of the VR component in Phase II increased the mean rate for the changed tutor and decreased the mean rate for the unchanged tutor. An interpretive analysis of these results will be presented in Chapter V. Experiment III Experiment III examined Multiple schedules of social reinforcement in which at least one of the two components manipulated error felsponses. Two sub experiments were each comprised of two experimental phases. Schedule components across phases of each subexperiment were identical. The components assigned to tutors in Phase I were switched in Phase II. Correct and error rates and the means of those rates are reported since this experiment manipulated both correct and error responses. Subexperiment III-l SI, an eight-year-old female, participated in Subexperiment III-l. Multiple schedule components were FI 30 seconds L (RC) and FI 30 seconds C (RE) . Tutors reinforced responses at the same frequency; one reinforced correct responses and the other reinforced error responses. Tutor 1, a female, reinforced correct responses and Tutor 2, a female, reinforced error responses in Phase I. The tutors switched components in Phase II; therefore.

PAGE 82

-74Tutor 1 reinforced error responses and Tutor 2 reinforced correct responses in Phase II. Results Hypothesis 1 stated that the Flash Card mean correct rates in the RC component will be higher than the mean correct rate of the RE component in Phase I and in Phase II of the Flash Card performance. Table 8 shows the correct and error rates and the means of those rates for the Flash Card performance of SI. The Phase I mean RC rate was 8.03 and the mean RE rate was 6.37, These data show that the mean RC rate was higher than the mean RE rate. The Phase II mean correct rate for the RC was 6.03 and for the RE was 5,32. These data show that the mean RC rate was higher than the mean RE rate. The results from Phase I and Phase II support the acceptance of Hypothesis 1. Hypothesis 2 stated that the Flash Card mean error rates in the RE component will be higher than the mean error rates of the RC component in Phase I and in Phase II. Table 8 provides the data for this Hypothesis. The Phase I mean RE rate was 2.80 and the mean RC rate was 1.35. These data show that the mean RE rate was higher than the mean RC rate. The Phase II mean RE rate was 2.04 and the mean RC rate was 3.02. These data show that the mean RE rate was lower than the mean RC rate in Phase II. Since only the Phase I results were in the expected direction, Hypothesis 2 is rejected.

PAGE 83

-75Table 8. Mean Rates of Correct and Error Responses in EXPERIMENT III Phase I Phase II RC RE RC RE Flash Card Correct 8.03 6.37 6.03 5.32 III-l Error 1.35 2.80 3.02 2.04 Reading Correct 41.30 39.96 34.63 42.55 Error 0.77 3.07 2.19 2.03 RC RE RC RE Flash Card Correct III-2 Flash Card

PAGE 84

-76Hypothesis 3 stated that the Oral Reading mean correct rates in the RC component will be higher than the mean correct rates of the RE component in Phase I and in Phase II. Table 8 shows the correct and error rates arid the means of those rates for the Oral Reading performance of SI. The Phase I mean RC rate was 41.3 and the mean RE rate was 39.96. These data show that the mean RC rate was higher than the mean RE rate in Phase I. The Phase II mean correct rate for the RC was 34.63 and for the RE was 42.55. These data show that the mean RC rate was lower in Phase II than was the mean RE rate. Since only the Phase I results were in the expected direction, H)rpothesis 3 is rejected. Hypothesis 4 stated that the Oral Reading mean error rates in the RE component will be higher than the mean error rates of the RC component in Phase I and in Phase II. Table 8 provides the data for this hypothesis. The Phase I mean RE rate was 3.07 and the mean RC rate was 0.77. These data show that the mean RE rate was higher than the mean RC rate in Phase I. The Phase II mean RE rate was 2.03 and the mean RE rate was 2.19. These data show that the mean RE rate was lower than the mean RC in Phase II. Since only the Phase I results were in the expected direction. Hypothesis 4 is rejected.

PAGE 85

-77Subexperiment III-2 S2, an eight-year-old male, served as the subject in Subexperiment III-2. The components of the Multiple schedule were a VI 60 seconds L (RC) and a VI 60 seconds D (RE) . Tutors 1 and 2 , a female and a male switched schedules from Phase I to Phase II. Results Hypothesis 5 stated that the Flash Card Phase I and Phase II mean correct rates will be higher in the RC component than in the RE component. Table 8 shows the correct and error rates and the means of those rates for the Flash Card performance of S2. In Phase I the mean RC correct rate was 4.93 and the mean RE correct rate was 4.52. The RC mean correct was, therefore, higher than the RE mean correct in Phase I. In Phase II the mean RC rate was 2,46 and the mean RE rate was 3.10. These data show that the mean RC rate was lower than the mean RE rate in Phase II. Therefore, since only the Phase I results were in the expected direction. Hypothesis 5 is rejected. Hypothesis 6 stated that the Flash Card Phase I and Phase II mean error rates will be higher in the RE component than in the RC component. Table 8 provides the data for this hypothesis to be tested. In Phase I the RE mean error rate was 2.31 and the RC mean error rate was 1.53. These data show that the RE mean error rate was higher than the RC mean error rate in Phase I.

PAGE 86

-78In Phase II the RE mean error rate was 1.88 and the RC mean error rate was 2.01. These data show that the RE mean error rate was lower than the RC mean error rate in Phase II. Since only the Phase I results were in the expected direction, Hypothesis 6 is rejected. Hypothesis 7 stated that the Oral Reading Phase I and Phase II mean correct rates will be higher in the RC component than in the RE component. Table 8 shows the correct and error rates and the means of those rates for the Oral Reading performance of S2. The Phase I data reveals that the mean RC rate was 17.66 and the mean RE rate was 17.81. Thus, the mean RC rate was lo^jer than the mean RE rate which was not expected. In Phase II the mean RC rate was 15.51 and the mean RE rate was 16.26. These results are not in the expected direction. That is, the mean RC rate was not higher than the mean RE rate in Phase II. Therefore, Hypothesis 7 is rejected. Hypothesis 8 stated that the Oral Reading Phase I and Phase II mean error rates will be higher in the RE component than in the RC component. Table 8 provides the data for this hypothesis . In Phase I the mean RE error rate was 2.08 and the mean RC error rate was 1.78. These results were in the expected direction. However, the Phase II results were not in the expected direction. The mean RE error rate as 1.05 was not higher than the mean

PAGE 87

-79RC error rate which was 2.08. Therefore, since the results of both phases were not in the expected direction, Hypothesis 8 is rejected. Summary of Experiment III Results Experiment III examined the effects of Multiple schedules of social reinforcement in which one of the two components reinforced correct responses and the other reinforced error responses. Although all eight hypotheses were rejected, there was remarkable consistency among the results. Whereas Phase I results for each hypothesis were in the expected direction, Phase II results were not in the expected direction. These interesting results will be discussed in Chapter V.

PAGE 88

CHAPTER V SUMMARY AND DISCUSSION Experiments I, II, and III attempted to determine whether there was a predictable relationship between the Multiple schedule of social reinforcement and the subject's mean rate of responses. This chapter discusses these results and the similarities and differences between them and other investigations. The results of" the three experiments suggested that while the experimental manipulations did not produce data in the expected direction overall, there was remarkable consistency among the results. This consistency has important implications for future investigations of social reinforcement schedules. The chapter concludes with a discussion of those implications. Briefly, the following was found: (1) Experiment I demonstrated differential responding to Multiple schedule components during Phase I. Generally the schedule which generated higher mean rates in Phase I produced higher mean rates in Phase II; (2) Experiment II 's design contrasted the differential effects of changing one tutor's reinforcement frequency on rates of reading and the effect of that change on the rate of response to the unchanged tutor; (3) Experiment III demonstrated the irreversibility 80-

PAGE 89

-81of response from Phase I to Phas6 II when at least one of the tutors consequated error responding. Experiment I demonstrated differential responding to Multiple schedules of socia:! reinforcement. Although the data did not replicate previous research, the consistency of the data suggests that response rate was under the control of the schedule. The Multiple schedules used in the three subexperiments each included a VR schedule and an FI schedule as the two components. Flash Card and Oral Reading mean response rates generally followed a consistent pattern in each subexperiment. Of the Flash Card and Oral Reading performances which changed, four unexpectedly generated higher rates in the FI component than in the VR component. In the fifth instance, the Flash Card performance of S4 in Subexperiment 1-2, the VR component showed a higher mean rate in both phases than did the FI component. In the last case, the Oral Reading performance of S3 in Subexperiment I-l, neither generated a consistently higher performance. In general, then, Experiment I results showed that the VR with its higher reinforcement density, generated lower response rates than did the FI. These results are supported by the Gewirtz (1967) findings in the area of satiation and deprivation of social reinforcement. Gewirtz concluded that the effectiveness of an approval word varied inversely with the frequency of its occurrence prior to the reinforcement test.

PAGE 90

-82-Basic animal operant research reviewed briefly in Chapter I found that the VR schedule typically generated higher rates of responding than did the FI schedule (Ferster and Skinner, 1957). While the Experiment I results are inconsistent with that study's findings, the particular characteristics of the present investigation presented may have a replication of the Ferster and Skinner findings. In this study the timing device continued to operate during the delivery and "ingestion" of the social reinforcement. In the cited research, the delivery of reinforcement was controlled mechanically. During reinforcement delivery, the timing device shut down; therefore, the calculation of rate of response did not include the time used to deliver and ingest reinforcement. A description of S3 's behavior provides an example of how this procedural difference may have influenced the results. S3 who participated in Subexperiment I-l, reacted emotionally to each reinforcement which seemed to disrupt her attention to the materials . The higher frequency of reinforcement during the VR condition and the disruption induced may have lessened the overall time in which the subject had to respond. Thus, the calculation of rate may have been affected. Staats, Finley, Minke, and Wolf (1964) demonstrated results with Multiple schedules on hximan reading responses somewhat consistent with basic animal research. However,

PAGE 91

tonlike the present study, the reinforcers were tokens , which allow greater control over response rate. Lovitt et al. (1971) found results contrary to the present findings as well. Using single ratio token schedules, these investigators generated differential joral reading 1 performance to CRF/EXT and FR30/EXT. As with the Staats et al. (1964) study, schedule differences and the use of tokens may account for the contradictory results. Racial factors may also explain these results. As discussed in the Literature Review, blacks and whites seem to be responsive to different sets of social reinforcers. In this regard perhaps the verbal statements selected as reinforcers served instead as negative reinforcers. The subjects may have responded at a slower rate on the VR schedule in order to avoid the occurrence of the verbal statement. It would be important in future research to rectify the procedural problem already alluded to in order to clarify this point. Experiment II demonstrated the contrast effect of changing reinforcement frequency of one of two previously equivalent components. "The change in behavior is called a contrast when the change in the rate of responding during the presentation of one stimulus is in a direction away from the rate of responding generated during the presentation of the other stimulus" (Reynolds, 1961, p. 57). Phase I Multiple schedule components were VI60"L. Tutor 2 generated a lower Flash Card and Oral Reading Phase I mean correct rate and therefore was chosen to change to the VR3H schedule in Phase II. As

PAGE 92

-84ekpected the introduction of the VR3H increased the correct rates in the changed component and lowered the correct rates in the unchanged VI component. Waite and Osborne (1972) provided the first demonstration of behavioral contrast in children. Multiple schedules influencing a simple operant were VI/VI and VI/EXT. The current study supports their findings by producing this contrast effect with different Multiple schedules using a more complex response reading. In contrast to the present experiment, Waite and Osborne (1972) found equivalent rates during the VI/VI schedule, Phase I rates were not equivalent in the present study. Because of this lack of equivalent rates, the tutor with the lower mean rate was chosen to change schedules. The results are not weakened by the lack of equivalent rates in the first phase. Experiments I and II as a unit demonstrated the influence of social reinforcement on Flash Card and Oral Reading responses. Social reinforcement schedules were shown to be the controlling variable. Certain factors in this study may have enhanced this outcome. First, the use of different tutors to serve as component stimuli may have hastened the discrimination between schedules. Second, the control over the rate of progress stabilized the effect of the "learningtolearn" phenomena. This allowed the response to fluctuate with the experimental manipulations. Third, point delivery

PAGE 93

-85was standard. The present study suggests that social reinforcement, a more naturally occurring reinforcer than a point or token, has a significant effect on a relevant educational behavior. Contrary to the results of Experiments I and II which showed schedule control, Experiment III did not demonstrate such control. Schedule effects in Phase I of the sub experiments were not maintained in Phase II when the tutors switched schedules. Thus, the tutor effect appeared to exert more influence over the results than did the schedules. This may have been due to the nature of the response being consequated. In this experiment, at least one of the two Multiple schedule components consequated error responses. It is unclear, however, whether the type of reinforcement, the consequation of errors, or the sequence of treatments was more important in influencing these results. Each subexperiment had equivalent schedules in terms of the schedule specification. That is, both schedules of Subexperiment III-l were FI30" and those in III-2 were VI60". One component in each consequated correct responses and one consequated error responses. The consequation of error responses by one tutor in Phase I may have induced an emotional reaction which could have exerted more control over the responses in Phase II than the schedule. For these reasons, caution must be taken in drawing conclusions from Experiment III.

PAGE 94

-86Experiment Ill's irreversible results contradict the firxdings of several other studies. Sajwaj and" Knight (1971) found, for example, that errors consistently increased if more attention was given them than corrects. Hasazi and Hasazi (1972) found that removing attention for digitreversals significantly reduced errors. These investigations sought to eliminate errors and no reversal was attempted; whereas the current one attempted to bring errors under control of the schedule. The results of two informational reinforcement studies (Buchwald, 1959; Spence and Dunton, 1967) also differ with Experiment Ill's lack of results. In comparing the effectiveness of Rb, RW, and Wb, Rb was found to be inferior. Therefore, one would asstime that the Wb conditions in Experiment III would yield consistently superior rates . That they did not may, of course, reflect the design of the current study and the sequence of treatments which was much different from the group design and CRT application of reinforcement in the informational studies. An analysis of the results of Experiments I, II, and III leads one to conclude that Multiple schedules of social reinforcement were shown to demonstrate differential effects on reiading behavior particularly when reinforcement was contingent upon correct responses. When schedules of reinforcement were applied to error responses, the interpretation of the results was unclear. Emotional responses to the tutors, as the sequencing interpretation would suggest, may influence

PAGE 95

-87the results. It must. also be made clear that although differential responding resulted from the use of Multiple schedules, the results are confounded by the inclusion of different types of social reinforcement vd. thin the Multiple schedules. The study demonstrated differential responding to social reinforcement in a single subject design. Future research to clarify these results or to simply look directly at schedules of social reinforcement would be useful. The conclusion, then, that social reinforcement was the controlling ivaxJLable influencing subject responsiveness suggests that researchers attempt the control of complex human behavior with more naturally occurring reinforcers. The present research does not deny the contention that social reinforcement plus tokens may be more effective than social reinforcement alone (Lahey and Drabman, 1974; Staats et al., 1962; Chadwick and Day, 1971). Rather, this study simply suggests that social reinforcement alone can influence a child's reading. The methodology in this study could be useful in future research on the teaching of reading. In order to demonstrate control by social reinforcement, many aspects of the tutorsubject interaction had to be standardized. The standardization provides a useful base from which the researcher or teacher could give structured input and observe its effects. Instructed to ignore inappropriate behaviors , such as errors , and to attend to appropriate or ontask behaviors such as correct

PAGE 96

-88responses, researchers and teachers can reliably observe the influence of the experimental manipulation of social reinforcement. Standardization of material presentation and of the rate of progress through the curriculum contributes more control procedures by which researchers and teachers could observe the effect of varying curriculum factors. Standardization of the variables in such a complex situation as a remedial reading tutorial interaction is critical to understanding and controlling the influence of idiosyncratic subject behaviors which impede progress. If procedures are standardized, such that behavior is not fluctuating due to an inconsistent environment, the experimenter or teacher can design additional procedures unique to each subject to facilitate learning. This study provides the first demonstration of the reversibility of social reinforcement effects on reading behavior. Several studies had shown the differential control of reading behavior by various token manipulations (Staats et al. , 1964; Lovitt et al. , 1971) but none had attempted to control reading behavior with social reinforcement. Moreover, some researchers had assumed social reinforcement to be a fruitless variable to investigate with respect to reading behavior or academic behavior (Lahey and Drabman, 1974; Staats et al. , 1962; Chadwick and Day, 1971). Comparatively, social reinforcement may be inferior to token reinforcement but its availability and societal acceptance makes it imperative that it not be denied

PAGE 97

-89as a research tool. Researchers questionned the viability of using social reinforcement and turned to token reinforcement as a more manageable research tool. Furthermore, historically social reinforcement research tended to control subservient behaviors (Thomas, Nielsen, Kuypers, and Becker, 1968) or irrelevant human behaviors, such as rate of placing marbles in holes (Panda, 1972). The results of the present research, however, suggest that factors of social interaction can be specified in order that social reinforcement can be disxjreetly delivered within so complexly defined a setting as remedial reading. In other words, social reinforcement effects can be investigated in relation to complex human learning. The utilization of the Performance Determined Instruction model (Gray et al. , 1969) stabilized the "learning-to-leam" phenomenon. This allowed the behavior to fluctuate as a result of the manipulation of experimental variables. The stabilization of accuracy within the 90 to 95 percent accuracy level ties in with J. McVicker Hunt's theory of the match necessary between the curriculiim and the child to promote intellectual accommodation (Hunt, 1964). The influence of emotional responses of boredom or frustration from too high or too low an accuracy level was controlled by the use of the PDI system.

PAGE 98

'-„.•' '"50-'; Critical reviews of past social reinforcement research suggested the need for functional designs (Van de Riet, 1963; Kennedy and Willcutt, 1964). The reversal and contrast designs of the present study controlled reinforcement history, the influence of characteristics of the subject, and the influence of characteristics of the tutors. Therefore , these results support the use of functional designs in social reinforcement research. An important area of research is opened by the current project. An analysis of the variables controlling error responding seems warranted. It remains of interest whether a reading error response is analogous to an inappropriate social behavior in terms of the controlling variables. The present research suggests the need for more extensive analysis of the factors which influence error response rate. For instance, one could systematically replicate Experiment III using a Multiple CRF/EXT schedule. Methodologically, tiiis study sustained implications for research, teaching, and the treatment of reading disorders. Basic research can be attempted in a school setting. Furthermore, the results suggest that basic research on relevant human behaviors can occur simultaneously V7ith the provision of remedial training. As social behavior has been reliably shown to be influenced by environmental consequences, the present study indicates that it influences reading behavior. The contrast between one teacher's style and another's may have a

PAGE 99

-91significant effect on an individual child's performance. Likewise, teacher contrast can effect a child's emotional responses and preferences. Consider the junior high program in which a student might have five equally reinforcing or demanding teachers first semester. Second semester the math teacher is changed to either an extremely negative or highly reinforcing one. The possible concommitant effect on responding to math and to the other courses seems analogous to the present contrast findings. The implications of the present research for the teaching of remedial reading are most promising. A training package can be developed based on the methodological decisions made for the current research. Instead of remediating specific skill deficiencies in hopes that there would be some generalization to overall reading strength, this project took a more radical view of the teaching of reading. The research assumed that if the child was placed at his 90 to 95 percent accuracy level in a programmed curriculum, the interaction between the tutor and the child would be critical to the child's strength as a reader. Not only was the interaction assumed to be important, but its dimensions were strictly specified. Random interactions were kept at a minimum. The control of social interaction with regard to both correct and error responses in remedial reading situations is important .

PAGE 100

-92Summary Reading behavior has been shown to be systematically effected by social reinforcement. It is unclear whether type of reinforcement or schedule of reinforcement was the controlling variable. Procedural limitations and the question of whether the verbal statements were in fact reinforcing may explain why the results were not in the expected direction. Analysis of error responding appears to be a fertile area for research. Basic research on complex hximan learning tasks seems viable within a public school setting. The small number of subjects limits the generalizability of these results, however, the clarity of the influence of social reinforcement seems to be a needed addition to the literature on the social reinforcement of reading. The methodology developed for this study to control tutorsubject interaction, rate of progress through a curriculum, and the delivery of social reinforcement has important implications for research, teaching, and the treatment of reading disorders.

PAGE 101

93APPENDIX A POINT SHEET

PAGE 102

-9-^ APPENDIX B DAILY DATA SHEET Schedule

PAGE 103

APPENDIX C STANDARD BEHAVIOR CHARTS

PAGE 104

-95SlflNIH HScJ SINSRaAOH

PAGE 105

-97-

PAGE 106

-98-

PAGE 107

-99HiriNIH a'3d[ SXNaWSy\OH

PAGE 108

-100aifiNiH a3(i siMaisiaAOH

PAGE 109

-101Pi >
PAGE 110

-102-

PAGE 111

-103-

PAGE 112

-104-

PAGE 113

-105-

PAGE 114

-106-

PAGE 115

-107-

PAGE 116

BIBLIOGRAPHY Allen, K.E., Hart, B.M. , Buell, J.S., Harris, F.R. , and Wolf, M.M. . Effects of social reinforcement on the isolate behavior of a nursery school child. Child Development , 1964, 35, 511-518. Ayllon, T. and Azrin, N. The token economy . New York: Appleton-Century-Crofts, 1968. Baer, D.M. , Wolf, M.D., and Risley, T.R. Some current dimensions af-^a-ppl-ied behavior analysis. Journal of Applied Behavior Analysis , 1968, 1, 91-97. Baron, R.M. , Bass, A.R. , and Vietzl, P.M. Type and frequency of praise as determinants of favorability of self-image: an experiment in a field setting. Journal of Personality , 1971, 39, 493-511. Baron, R.M,, Heckenmueller , J., and Shultz, S. Differences in conditionability as a fact of race of subject and prior availability of a social reinforcement. Journal of Personality . 1971, 39, 94-111. Binet, A. and Vaschide, E. Experiences de forces masculaire et de fond chez les jeunes garcons . Annee Psychology , 1897, 4, 15-63. ^ Broden, M. , Bruce, C. , Mitchell, M. , Carter, V., and Hall, RV. Effects of teacher attention on attending behavior of two boys at adjacent desks. Journal of Applied Behavior Analysis , 1970, 3, 199-203. ' ~~" Buchwald, A.M. Experimental alternations in the effectiveness of verbal reinforcement combinations . Journal of E xperimental Psychology . 1959, 57, 351-361. Campbell, D.T. and Stanley, J.C. Experimental and quasiexperimental designs for research on teaching. In N.L. Gage (Ed.), Handbook of Research on Teaching . Chicago: Rand McNally, 1963. -108-

PAGE 117

-109Chadwick, B.A. and Day, R.C. Systematic reinforcement: academic performance of underachieving students. Journal of Applied Behavior Analysis , 1971, 4, 311-319. Erickson, M.T. Effects of social deprivation and satiation on verbal conditioning in children. Journal _ of Comparative and Physiological Psychology , 1962, 55, 953-957. Ferster, C.B. and Skinner, B.F. Schedu l es_of reinforcement . New York: Appleton-Century-Drofts,' 1957. Gewirtz, J.D. Deprivation and satiation of social stimuli as determinants of their reinforcing efficacy. In J.P, Hill (Ed. ) , Minriesota s3miposia on child psychology . Vol. 1. Minneapolis: University of Minn. Press , 1967. Gewirtz, J. and Baer, D. The effect of brief social deprivation on behaviors for a social reinforcer. Journal of Abnormal Social Psychology , 1958a, 56, 49-55T Gev7irtz, J. and Baer, D. Deprivation and satiation of social reinforcements as drive conditions. Journal of Abnormal Social Psychology , 1958b, 57, 165-172. Gray, B.B., Baker, R. D. and Stancyk, S.E. Performance determined instruction for training in remedial reading. Journal of Applied Behavior Analysis , 1969, 2, 255-263. Hall, C.S. and Lindzey, G. Theories of Personality . New York: John Wiley and Sons, Inc., 1957. Hall, Pv.V. , Lund, D. , and Jackson, D. Effects of teacher attention on study behavior. Journal of Applied Behavior Analysis , 1968, 1, 1-T2"; Harris, F.R. , Wolf, M.M. , and Baer, D.M. Effects of adult social reinforcement on child behavior. In W.W. Harteys and N.L. Smothergill (Eds, ) , The Young Child Reviews of Research . Washington, D.C.: NAEYC, 1967. Hart, B.M. , Allen, K.E., Buell, J.S., Harris, F.R., and Wolf, M.M. Effects of social reinforcement on operant crying. J ournal of Experimental Child Psychology , 1964, 1, 145-153. "^~ Hasazi, J.E. and Hasazi, S.E, Effects of teacher attention on digit-reversal behavior in an elementary school child. Journal of Applied Behavior Analysis, 1972 , 5, 157-1F2: Hess, R.D, and Shipman, V.C. Early experience and the socialization of cognitive modes in children. Child Development , 1965, 36, 869-886.

PAGE 118

-noHill, K. The effects of social reinforcement and nonreinforcement following success and failure as a function of test and of subject, sex of subject, and sex of experimenter. Unpublished doctoral dissertation, University of Minnesota, Minneapolis, 1965. Hill K. and Stevenson, H, The effects of social reinforcement versus non-reinforcement and sex of E on the performance of adolescent girls. Journal of Personality , 1965, 33, 30-36. Hoeltzel, R.C. Reading rates and comprehension as affected by single and multiple-ratio schedules of reinforcement within a token economy as measured by precision ' ' teaching techniques . (Doctoral dissertation. University Microfilms) , 1973 , No . 74-9066. Hunt, J. McV. The-^ycho logical basis for using pre-school enrichment as an antidote for cultural deprivation. Merrill-Palmer Quarterly, 10, 3, 1964. Hurlock, E.B, The value of praise and reproof as incentives for children. Archives of Psychology , 1924, 71, 21-25. Kelly, R. and Stevens, M.N. Comparison of different patterns of social reinforcement in children's operant learning. Journal of Comparative and Physiological Psychology , 1964, 57, 295-i296. Kennedy, W.A. and Willcutt, H. C. Praise and blame as incentives. Psychological Bulletin , 1964, 62, 323-332. Klaber, M.D. , Butterfield, E.C., and Gould, J.J. Responsiveness to social reinforcement among institutionalized • retarded children. American Journal of Mental Deficiency , 1969, 73, 890-895. Lahey, B.B. and Drabman, R.S. Facilitation of the acquisition and retention of sight-word vocabulary through token reinforcement. Journal of Applied Behavior Analysis , 1974, 7, 307-312. Landau, R. and Gewirtz, J.L. Differential satiation for a social reinforcing stimulus as a determinant of its efficacy in conditioning. Journal of Experimental Child Psychology . 1967, 5, 391-405. ; Leith, G.O.M. and Davis, T.N. Influence of social reinforcement on achievement. Educational Research, 1969, 11, 132-137.

PAGE 119

-111Leventhal, H. and Fischer, K. What reinforces in a social reinforcement situation--words or expressions? Journal of Personality and Social Psychology . 1970. 14, 83-94. Lovitt, T., Eaton, M. , Kirkwood, M. , and Pelander, J. Effects of various reinforcement contingencies on oral reading rate. In E. A. Ramp and B.L. Hopkins (Eds.), A new direction for education: behavior analysis 1971 , Kansas City, Kansas: University o± Kansas, 1971. ~ Marshall, H.H. Learning as a function of task interest, reinforcement, and social class variables. Journal of Educational Psychology . 1969, 60, 133-137"! Mays, B.K. Effects of variable-interval and variable-ratio schiedules of verbal reinforcement on error elimination in oral reading an exploratory study . (Doctoral dissertation. University of Tennessee) Ann Arbor , Mich : University Microfilms, 1970, No. 71-17, 759. Meyer, W. , and Offenback, S. Effectiveness of reward and punishment as a function of task complexity. Journal of Comparativ e and Psysiological Psychology. 1962. 55, 532-534. ^ ^ Miller, G.A. , Galanter, E. , and Pribram, K.H. Plans and ^ structure of behavior . New York: Holt, Rinehart , and Winston, Inc. , 1960. Mussen, P.H. The psychological development of the child . (2nd ed.) Englewood Cliffs, New Jersey: Prentice-Hall. 1973. O'Leary, K.D. and Drabman, R. Token reinforcement programs in the classroom: a review. Psychological Bulletin. 1971, 75, 379-398. [ — Panda, K.C. Effects of social reinforcement, locus of control, and cognitive style on concept learning among retarded childreiT (Doctoral dissertation, Indiana University) Ann Arbor, Mich: University of Microfilms, 1970,/ No. 71-6891. Panda, K.C. Social reinforcement: Theoretical issues and research implications. Psychological Studies , 1971. Panda, K.C. and Ljmch, W.W. Effects of social reinforcement on the retarded child: a review and interpretation for classroom instruction. Education and T raining of the Mentally Retarded , 1972, 7, Il5-i2'3" ,

PAGE 120

-il2Parker, D.H. and Scanneil, G, Science Research Associates Laboratory la. Chicago: Science Research Associates, 1961, Penny-packer, H,S. , Koenig, C.H. , and Lindsley, O.R. Handbook of the Standard Behavior Chart (Preliminary edition) . Kansas City, Kansas: Precision Media, 1972. Redd, W.H. Effects of mixed reinforcement contingencies on adults' control of children's behavior. Journal of Applied Behavior Analysis , 1969, 2, 249-254. Reiner, H. Children's learning as a function of type of reinforcement type of task, grade level, social class, and sex. Paper presented at the meeting of the American Psychological Association, New Orleans, August 1974. Rejmolds, L. J. , Light, J.A. , and Mueller, F.L. The effects of reinforcing quality or quantity on academic performance. Paper presented at the meeting of the AERA, New York City, February 1971. Rosenthal, R. Ex perimenter effects in behavioral research . New York: Appleton-Century-Crof ts , 1966 . Sajwaj , T. and Knight, P. The detrimental effects of a correction procedure for errors in a tutoring program for a young retarded boy. In E.A. Ramp and B.L. Hopkins (Eds . ) , A new direction for education: behavior analysis 1971 , Kansas City, Kansas; University of Kansas, 1971. Sibley, S.A. Reading rate and accuracy of retarded readers as a function of fixed-ratio schedules of conditioned reinforcement . (Doctoral dissertation, Florida State University) Ann Arbor, Mich: University of Microfilms, 1966. No. 67-5528. Sidman, M. Tactics of Scientific Research . New York: Basic Books, 1960. Siegel, A.W. and VanCara, F. Effects of different types of reinforcement on a young child's incidental learning. Child Development , 1971, 42, 1596-1601. Siegel, S. Nonparametric statistics for the behavioral sciences . New York: McGraw-Hill, 1956. Singer, H. and Ruddell, R.B. (Eds.) Theoretical models and processes in reading . Newark , Deleware : International Reading Association, 1970.

PAGE 121

-113Solomon, D. and All, F. Age trends in the perception of verbal reinforcers. • Developm ental Psychology, 7, 1972, 238-243. Spence, J.T. Verbal-discrimination performance as a function of instructions and verbal-reinforcement combinations in normal and retarded children. Child Development, 1966, 37, 269-281. Spence, J.T. and Dunton, M.C. The influence of verbal and nonverbal reinforcement combinations in the discrimination learning of middleand lower-class nreschool children. Child Development . 1967, 38, 1177-1186. Staats, A.W., Finley, J.R. , Minke, K.A. , and Wolf, W. Reinforcement variables in the control of unit reading responses. Journal of the Experimental Analysis of Behavior , 1964, 7, 139-149. Staats, A.W. , Minke, K.A. , Finley, J.R., Wolf, M. and Brooks, L.O. A reinforcer system and experimental procedure for the laboratory study of reading acquisition. Child Development , 1964, 35, 209-231. Staats, A.W. , Staats, K., Schutz, R.E. , and Wolf , M. The conditioning of textual responses using "extrinsic" reinforcers . Journal of the Experimental Analysis of Behavior . 1962, 5, 33-40. Thomas. D.R. , Becker. W.C. , and Armstrong. M. Production and elimination of disruptive class behavior by systematically varying teacher behavior. Journal of Applied Behavior Analysis . 1968, 1, 35-45. Thomas, D.R. . Nielsen, L.J.. Kuypers , D.S., and Becker, W.C. Social reinforcement and remedial instruction in the elimination of a classroom behavior problem. Journal of Special Education , 1968, 2, 291-306. Todd, J. and Nakamura. C.Y. Interactive effects of informational and affective components of social and nonsocial reinforcers on independent and dependent children. Child Development, 1970. 41, 365-376. Van de Riet. H. The effects of praise and reproof on pairedassociate learning in educationally retarded children. (Doctoral dissertation, Florida State University) Ann Arbor, Mich: University Microfilms, 1963. No. 63-6367.

PAGE 122

-114Waite, W.W. and Osborne, J. G. Sustained behavioral contrast in children. Journal of the Experimental Analysis of Behavior . 1972, 18, 113-117. Wasik, H. , Senn, K. , Welch, R.H., and Cooper, A. Behavior modification with culturally deprived school children: two case studies. Journal of Appl ied Behavior Analysis. 1969, 2, 181-194. ^

PAGE 123

BIOGRAPHICAL SKETCH Pamela Phelan was bom in Quincy, Massachusetts and raised in Plantation, Florida. She received her Bachelor of Arts in Psychology (June, 1970) and Master of Education (August, 1971) degrees from the University of Florida. Presently, she lives in Silver Spring, Maryland and is employed as a school psychologist with Montgomery County Public Schools in Rockville, Maryland. -115-

PAGE 124

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality as a dissertation for the degree of Doctor of Philosophy! Professor of Education I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy! uohn Newell Professor of Education I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality as a dissertation for the degree of Doctor of Philosophy. William Ware Professor of Education

PAGE 125

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor^f Philosophy. Professor of Psychology This dissertation was submitted to the Graduate Faculty of the College of Education and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August, 1976 Dean, 'College of/Education Dean, Graduate School