Group Title: relative effects of modeling on the acquisition of wait-time by preservice elementary teachers
Title: The relative effects of modeling on the acquisition of wait-time by preservice elementary teachers
Full Citation
Permanent Link:
 Material Information
Title: The relative effects of modeling on the acquisition of wait-time by preservice elementary teachers and concommitant changes in dialogue patterns and pupil performance
Physical Description: x, 122 leaves : ill. ; 28 cm.
Language: English
Creator: DeTure, Linda Riley
Publication Date: 1976
Copyright Date: 1976
Subject: Elementary school teachers -- Training of -- Florida   ( lcsh )
Educational psychology   ( lcsh )
Curriculum and Instruction thesis Ph. D
Dissertations, Academic -- Curriculum and Instruction -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis--University of Florida.
Bibliography: Bibliography: leaves 116-121.
Statement of Responsibility: by Linda R. DeTure.
General Note: Typescript.
General Note: Vita.
 Record Information
Bibliographic ID: UF00098116
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000176140
oclc - 03057820
notis - AAU2618


This item has the following downloads:

relativeeffectso00detu ( PDF )

Full Text






[ -____

I dedicate this work to my husband, Francis and to
my children, Michael and Nicky, for sharing with me the
rather trying experience of being a graduate student.


I wish to acknowledge and thank Dr. John J. Koran,

Jr., the chairman of my committee, and Dr. Mary Budd Rowe.

Dr. Koran's assistance from the point of conceptualization

to the completion of this paper has been invaluable and

his enthusiastic support has kept my spirits kindled.

Hopefully, Dr. Rowe's continued influence and guidance

will be reflected in this project. I would like to

thank the members of my supervisory committee, Dr. Gordon

Lawrence and Dr. Vynce Hines, for their assistance and

Dr. Mary Lou Koran for her particular contribution to the

conceptualization of the study. I especially want to thank

Dr. Marion Fisher and Ms. Linda Esker from the National

Institute of Health for their generous assistance with the

computer and the analysis of the project. Also, Dr. Sue

Kinzer has my lasting appreciation for being such a per-

fect model.



ACKNOWLEDGEMENTS ........ ......................... iii

LISTS OF TABLES................................... vi

ABSTRACT......................................... viii


I. INTRODUCTION.... ........................... 1

The Proble S7 ........................ 1
Rationale .... .............................. 1
Review of Literature................... 6
Modeling .............................. 6
Modeling and Teacher Education........ 11
Wait-time.......... ................... 15
Wait-time Training................... 21
Summary of the Rationale of the Hypotheses..... 24
Statement of the Hypotheses ............. 25

II. EXPERIMENTAL DESIGN ....................... 27

The Design ............................. .. 27
Internal Validity.... ................ 28
External Validity...................... 30
Treatment Procedures .................... 33
Sampling............................... 33
General Procedures .................... 34
Treatment Materials.................. .. 37
The Models ............................ 37
Tasks ................................. 37
Measurements......... ..... ......... ..... 38
Student Tests.......................... 41

III. RESULTS................................... 43

Treatment Main Effects.................. 44
Main Effects ........................... 44
Wait-time II........................... 45
Wait-time I............................ 49
Outcome Variables........................ 51
Mean Length of Teacher Talk............ 52
Mean Length of Student Response........ 54
Proportion of Teacher Talk............. 56
Student Performance Measure ........... 58

Microteaching Sessions ....... ........... 60
Microteaching Session 1........ ........ 60
Microteaching Session 2.......... ...... 61
Microteaching Session 3................. 63
Pupil Performance Tests................ 64


Treatment Variables...................... 69
Interpretation of Treatment.............. 72
Wait-time II and Modeling.............. 72
Wait-time II and Feedback .............. 75
Wait-time I............................. 76
Outcome Variables........................ 78
Dialogue Variables......... ............ 78
Pupil Performance Test................. 81
Summary of Outcome Variables........... 83
Implications for Future Research......... 85


A. SET INDUCTION MATERIALS...................... 88

B. TEACHING TASK.................................. 92

C. MATERIALS LIST.............. ................... 100

D. WRITTEN MEASURES............. ................ 103

E. SERVO-CHART PLOTS OF WAIT-TIME............... 110

F. VARIABLE DEFINITIONS......................... 115

REFERENCES....................................... 116

BIOGRAPHICAL SKETCH............. ................ 122


Table Page

1. Experimental Design for Phase One ... . . 27

2. Experimental Design for Phase Two . . .. 29

3. Treatment Procedures and Times ....... 35

4. Reliability -of Written Measures ..... . 42

5. Means and Standard Deviations for
Wait-time II . . . . . . . . 46

6. Analysis of Variance for Wait-time II . . . 46

7. Tukey's HSD for Differences Between
Means Wait-time II . . . . . ... 47

8. Percent of Subjects at Criterion, Wait-time II 48

9. Means and Standard Deviations for Wait-time I 50

10. Analysis of Variance for Wait-time I . . .. 50

11. Tukey's HSD Test for Differences Between
Means Wait-time I . . . . . . ... 51

12. Means and Standard Deviations for Mean Length of
Teacher Talk . . . . . . . ... 53

13. Analysis of Variance for Mean Length Teacher
Talk . . . . . . . . .. .. . 53

14. Tukey's HSD Test for Differences Between Mean
Length Teacher Talk . . . . . ... 54

15. Means and Standard Deviations for Mean
Length Student Talk . . . . . . . 55

16. Analysis of Variance for Mean Length Student
Talk . . . . . . . . ... .. . 55

17. Tukey's HSD Test for Differences Between Mean
Length Student Talk . . . . . ... 56

Table Page

18. Means and Standard Deviations for Propor-
tion of Teacher Talk . . . . . ... 57

19. Analysis of Variance for Proportion of
Teacher Talk . . . . . . . ... 57

20. Tukey's HSD Test for Differences Between
Means Proportion of Teacher Talk . . .. 58

21. Means and Standard Deviations for Tests . . 59

22. Analysis of Variance for Tests . . . .. 59

23. Means, Standard Deviations and Inter-
correlations Session 1 . . . . . .. 61

24. Means, Standard Deviations and Inter-
correlations Session 2 . . . . . .. 62

25. Means, Standard Deviations and Inter-
correlations Session 3 . . . . . .. 63

26. Multiple Regression Summary Data . . . .. 64

27. Means, Standard Deviations and Inter-
correlations Test Scores . . . . .. 66


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



Linda R. DeTure

December, 1976

Chairman: John J. Koran, Jr.
Major Department: Curriculum and Instruction

The purpose of this study was to devise an effective

method for training teachers to extend wait-time, defined

as the time a teacher pauses after asking a question, wait-

time I (WTI), and after a student response, wait-time II

(WTII). The two treatments, a video and an audio model,

were tested with and without a feedback component. The

outcome variables measured were mean length of teacher talk

(MLTT), mean length of student talk (MLST), proportion of

teacher talk (PTT) and a student performance measure examined

by a process skills test.

Fifty-two preservice elementary teachers were randomly

assigned to four groups: audio no feedback; audio feed-

back; video no feedback; and video feedback. The subjects

taught a series of three science inquiry lessons to four

fourth or fifth graders in a microteaching setting. Each

teaching session was audio recorded and entry level wait-

times were calculated. Before Session 2 teachers observed


either a video or audio model depicting the desired criterion

wait-time of three seconds. Subjects taught Session 2 using

extended wait-time. Before Session 3 teachers in the feed-

back group listened to and rated their tape from Session 2

to determine frequency of criterion wait-time. The no

feedback group read an inquiry related article. Both groups

were instructed to use extended wait-time for Session 3.

The 208 children were administered the process test at the

end of each session.

Mean lengths of wait-time I and II and the dialogue

variables were calculated from the audio tapes with a

servo-chart recorder. Treatment effects were analyzed with

a split plot factorial ANOVA design. Dialogue variables

and student tests were analyzed by ANOVA. Multiple regres-

sion analysis was performed for each teaching session with

wait-time II as the dependent variable.

The results revealed that the video group achieved

significantly longer wait-time II than the audio group with

both groups improving significantly from the entry level

of 0.06 seconds. The effect of feedback was not signifi-

cant at the 0.05 level. However, it was sufficient to

bring the video group to a mean wait-time of 3.6 seconds.

Wait-time I increased significantly from Session 1 to Session

3 but the increase did not reach the desired criterion

for any session.

Of the outcome variables both mean length of student

talk and the proportion of teacher talk were highly

correlated with wait-time for all three teaching sessions.

As the length of WTII increased MLST increased and PTT

decreased. The mean length of teacher talk was found to

correlate with WTII significantly only for Session 2 dur-

ing which MLTT decreased and WTII increased. It was sus-

pected that teachers initially attempted to increase

wait-time by controlling their length of utterance. MLTT

increased to entry level for Session 3 and did not corre-

late significantly with wait-time. Test scores did not

correlate significantly with wait-time for any teaching

session possibly because of the wide fluctuation of the

wait-time variable between teaching sessions. Perhaps

wait-time is necessary, but not sufficient, to produce

the expected changes.

This study suggests that a video model would be an

effective training method for increasing teacher's wait-

time. The audio model with modifications could be explored

as an alternative if expence requirements prohibit the

production and the use of the video model. The addition

of feedback seems warranted to maintain performance al-

though it does not appear to be essential for the ac-

quisition of extended wait-time.




The Designr

The purpose of this study is twofold. In the first

part two training methods are compared to determine their

relative effects on the acquisition of wait-time, as de-

fined by Rowe (1973), by preservice teachers. During this

phase the training methods are the independent variables

and the teacher's puasing behavior is the dependent vari-

able. Phase two attempts to validate the utility of the

acquired teacher behaviors by measuring student outcome

variables. In part two, the teacher's behavior becomes

the independent variable and pupil performance measures

are the dependent variables.


Research in teacher education has become experimentally

oriented only within the last ten years. Prior to 1967

Denemark and McDonald (1967) found empirical research in

teacher education to be scanty in most areas and essentially

nonexistent in others. One outcome of the recent research

is that teacher behavior has become more operationalized

partially because of the development of instruments for

objectively recording teacher behaviors. With objective

assessment of behaviors, teacher education is approaching

a greater degree of systemization.

Reflecting and perhaps corresponding to a more

methodical approach to teacher education, a movement toward

competency based education (CBE) has gained widespread

recognition and popularity within the past five years. A

1972 AACTE survey revealed that greater than 60 percent of

the 783 responding teacher education institutions either

characterized their programs as competency based or were

in the developmental stages of establishing programs

(Houston, 1974). Florida, for example, has already invested

millions of dollars in CBE projects. Works such as the

Florida State University, The Middle School CBTE Project at

the University of Florida and the Self-Paced Training

Module Bank at the University of Miami indicate the impact

of the movement in the state.

With large numbers of prospective teachers already

being trained in competency based programs, educators

(Koran, J. J. & Koran, M. L., 1974-1975) have expressed a

need for the development of inexpensive, adaptable instruc-

tional strategies to be used in the programs. Before

devising suitable instructional methods, educators need

to examine the adequacy of procedures for identifying and

validating teacher competencies. Comprehensive lists of

competencies written for all levels from local to statewide

(Dodl, 1973) illustrate that a sizable number can be

identified. The important question regarding the desir-

ability and validity of paritcular competencies has not

been sufficiently answered. Despite the professional

expertise and number of dollars concentrated on competency

based teacher education, the relationship between teacher

competencies, instructional methodology and student gains

remains highly speculative still. Research reviews of

teacher education reported few studies relating cognitive

teacher variables to student achievement. In a review of

thirty-five studies relating teacher behavior to student

gains Rosenshine (1970) found the studies relating teacher

verbal behaviors to student achievement to yeild generally in-

conclusive results. Correlations, though consistently posi-

tive, were not significant. Similarly Rowe and DeTure

(1975), reviewing science education research, encountered

little empirical evidence supporting the notion that changes

in teacher behavior are accompanied by changes in student

achievement. The majority of the studies in teacher educa-

tion were designed to measure aspects of change in teacher

behavior but not changes in pupil performance.

Nevertheless a moderate number of findings indicate

that certain teacher competencies do have positive influence

on pupil gains. Rosenshine and Furst (1971) identified

eleven variables that can be linked to student learning.

In another extensive line of research Rowe (1974a) reported

highly supportive evidence relating teacher pausing be-

havior to ten student verbal characteristics. Two studies

(Koran, J.J.,Jr., & Koran, M.L., 1973; Santiesteban,

1974) have been conducted which trained teachers in a

particular competency and subsequently validated the train-

ing procedures by measuring related pupil achievement gains.

Variables evolving from this kind of research would be

more likely to merit inclusion in teacher training pro-


Koran and Koran (1973) utilized a two phase approach.

In this study observational.learning was used to train

teachers to ask analytical questions and subsequently stud-

ent analytic responses were measured. Pupils of trained

teachers made a significantly greater number of responses

in a larger number of categories than did students of non-

trained teachers.

In a similarly designed study Santiesteban (1974) dis-

covered that students of teachers trained to ask observation

and classification questions scored significantly higher on

an observation and classification skills test than those chil-

dren who worked with teachers who did not master the skill.

Once teacher competencies have been appropriately

identified and validated instructional methods for training

teachers must be designed. Peck and Tucker (1973) suggest

that a systems approach will subsequently improve the effec-

tiveness of the program. An example would be the system

designed by Rosenshine (1970) which follows a pattern of

six cyclical steps. These are as follows: specific instruc-

tions for the desired behaviors; training procedures aimed


at the objectives; measurement of results according to

behavioral objectives criteria; feedback of results to the

learner; re-entry into the training procedure; and measure-

ment of results following the second training. The

training procedure in this study follows a similar pattern.

To reiterate, the research problem consists of two

phases; a training phase in which two methods are compared

for relative effectiveness and a validation phase in which

the competency is related to pupil outcome gains. The two

training methods compared, video and audio modeling, have

social learning theory as the theoretical basis. The re-

search of Bandura (1963, 1965, 1971), has contributed em-

pirical support for the theoretical tenents of modeling, or

observational learning, as a theory of learning. Here the

relative effects of a video model and an audio model on the

acquisition of wait-time by preservice elementary school

teachers are compared.

In phase two of the study the effects of wait-time, as

defined by Rowe (1973), are examined. The three verbal

outcome variables being measured are related to dialogue

patterns of a two player model. These are mean length of

teacher talk, mean length of student talk and the proportion

of speak-space controlled by the teacher. The fourth out-

come variable measured is pupil performance based on a

test of the objectives of the task. Common objectives of

the three tasks include making careful observations and

drawing inferences based on evidence.


Review of Literature

The literature reviewed here provides empirical evi-

dence to support the hypotheses and assumptions of this

study. Relative to the training methods are the theoretical

underpinnings of observational learning as a learning theory

and the utility of modeling as an instructional design. Next

the research identifying and clarifying the dependent variable,

wait-time, is reviewed which is followed by a review of wait-

time training.


From casual observation it is evident that many elements

of human behavior are transmitted by exposure to social models.

Historically, a number of explanations for imitative behavior

have been generated (Tarde, 1903; Allport, 1924; Guthrie,

1952; Miller and Dollard, 1941). However, because of a general

inability of the theories to explain the diverse effects that

modeled behavior can produce, theoretical controversy frequently

arose. By examining social learning theory, Bandura and Walters

(1963) have shown that observational learning can account for

many of the theoretical discrepancies that plague imitative

theories which usually rely on exact or nearly exact replica-

tion of learned behaviors.

Modeling can have as many as three kinds of effects

depending upon the circumstances involved. Observers may

acquire a whole new pattern of behaviors by observing the

performance of others; or modeling influences may serve to

strengthen or weaken inhibition of previously learned

responses. Also disinhibitory effects may become evident

when the learner observes models engaged in prohibited

activities without adverse consequences.

Social learning theory assumes that models operate in

an informative manner with the observer acquiring a sym-

bolic representation of the event. As a result Bandura

(1971) describes the modeling process as consisting of four

interrelated subprocesses each having an input upon whether

or not learning takes place. One subcomponent of observa-

tional learning is the attentional process. Simple exposure

to a model in no way assures that the observer will attend

to the model in the desired manner. If the observer fails

to recognize and differentiate the distinctive features of

the modeled behaviors, matching behaviors will not be ac-

quired. Upon attending to the stimuli the observer must

retain the essential elements of the act. As behaviors are

frequently observed without being immediately acted out, the

input must be stored in symbolic form. Thus retention of

the behaviors is the second subprocess.

Bandura, Grusec, and Menlove (1966) had children ob-

serve complex sequences of behavior. During observation the

children either watched attentively, coded the response into

its verbal equivalence or counted rapidly. A posttest dis-

closed that children who coded the material reproduced

significantly more matching responses than either of the


other two groups. Children engaged in competing symboliza-

tion acquired the lowest number of matched responses.

The third component of modeling is the motoric repro-

duction of the act in which the symbolic representation of

the model serves to guide overt performance. Finally a

learner may acquire a modeled behavior without ever per-

forming it if there are unfavorable sanctions or if there

are positive reinforcers for nonperformance. Reinforce-

ment and motivation may activate a previously learned but

unperformed act (Bandura, 1965). Anticipated consequences

of a particular behavior may not only prevent overt expres-

sion of the behavior but may interfere with learning the

behavior because the learner exhibits selective control

over the events to which he attends. Conversely positive

reinforcement serves as a powerful incentive for the learner

to attend to the modeled behaviors.

Each of the four subprocesses, attention, retention,

motor reproduction and reinforcement, can be influenced.

Characteristics of the subjects, the stimulus and the model

or models may act independently or jointly upon the learning

process. The following research studies tend to substanti-

ate the importance of these.

Because the subject must be able to focus his atten-

tion on the relevant features of the stimulus as well as

being able to perform it, developmental processes as mea-

sured by age could be expected to interact with the modeled

situation. A common finding, relating differential

responses of children, is that age influences acquisition

of behaviors (Denny, 1972; Leifer et al., 1972; Liebert

et al., 1969). Upper level elementary children demonstrate

more matching responses than younger elementary children.

Sex differences have been reported showing that males

tend to respond to models of either sex more readily than

females (Masters & Morris, 1971). Portuges and Feshbach

(1972) report different results when examining sex and social

status. Economically advantaged females acquired signifi-

cantly more incidental behaviors than did white males with

a positively reinforced white teacher. Disadvantaged

children of both sexes displayed fewer matched behaviors

than advantaged children of both sexes. The authors specu-

lated that the use of a bearded male model may have in-

timidated the female subjects resulting in fewer matched

responses. It may be also that the disadvantaged performed

less well than the advantaged children because the children

did not view the the models as being similar to themselves.

The effectiveness of a model increases if the model is per-

ceived as being similar to the learner (Burnstein, Stotland,

& Zander, 1961; Rosekrans, 1967: Stotland, Zander, & Natson-

las, 1962; Stotland & Dunn, 1962).

Relating the behavioral characteristics to the

stimulus act Bandura et al. (1966) found that a complex be-

havior is more easily learned when the act is broken into

its component parts.


Of the three characteristics influencing observational

learning the model is the most significant feature. An

array of studies support the following findings. Bandura

(1963) discovered that the power of the model is increased

if the model was perceived as having a high degree of compe-

tence and status. The most influential models in a child's

life are those who offer the major source of support and/or

control. Parents, teachers and peers are the most frequent

models for children (Bandura & Houston, 1961; Mussen &

Parker, 1965; Hartup, 1969). Although children tend to

imitate models similar to themselves, evidence concerning

sex and race varies and conclusions cannot be drawn with


It was also noted that multiple models, engaging in

the desired behaviors, produced more imitative behavior in

subjects than single models (Bandura, 19G9). Bronfenbrenner

(1970) found that if subjects desired to become members of

a group, the potency of group members as models was greatly


Reinforcement to the model may control the conditions

under which actual performance takes place (Masters &

Morris, 1971; Zimmerman & Pike, 1972; Geshuri, 1972).

While sanctioning behavior may either contribute to per-

formance or inhibit it, reinforcement is not essential for

either acquisition or performance (Bandura & Walters, 1963;

Bandura et a]., 1966; Bandura, 1969). In summary, modeling

alone produces behavioral changes as does modeling with

reinforcement either real or vicarious. Although the issue

of reinforcement to the model is intriguing it is not

central to this study. As there is research evidence to

show that modeling is effective without reinforcement, in

this study neither the teacher subjects nor the models re-

ceived sanctions.

Investigations of symbolic modeling indicate that the

physical presence of a model is not necessary for observa-

tional learning to occur if the essential features of the

behaviors are depicted pictorially or verbally (Bandura &

Mischel, 1965; Bandura, Ross & Ross, 1963). However, Bandura

and Menlove (1968) demonstrate that different forms of modeling

are not always equally effective. In instances where behaviors

controlled by strong inhibitions are expected to be performed,

live models generate better results.

Modeling and Teacher Education

The work of Bandura and others has helped establish

the boundaries of the modeling phenomenon in children.

Application of modeling theory to teacher education has

been explored also. In 1967 McDonald and Allen suggested

the utility of live or symbolic models illustrating desired

teacher behaviors as an alternative to more traditional

verbal descriptive methods of training. In view of

Bandura's research McDonald hypothesized that the model-

teacher could be trained to display unambiguous examples

of desired behavior. To further insure control of the

teaching task a film-mediated model of the desired be-

haviors was decided upon as a promising strategy. The

objective was to reduce the length of training required

and increase the subjects awareness of relevant cues.

McDonald and Allen foresaw both feedback and practice as

inherent parts of the modeling training procedures.

In a series of experiments designed to teach question-

ing techniques McDonald and Allen proceeded to probe the

relationship between modeling and feedback for relative

effectiveness in altering behavior. From the experiments

several significant results emerged. A filmed model was

consistently superior to written or verbal instruction. For

teaching questioning skills the video model was character-

ized as having distinctive cueing properties, an advantage

missing in the symbolic model. The most effective feedback

system was one in which the subject viewed his own per-


Peck and Tucker (1973) when reviewing feedback studies,

found consistent evidence to confirm the utility of giving

objective feedbacklabout specific features of training to

the subjects. The effect of feedback as a part of training

is incorporated into this study.

Other research in teacher education provides evidence

that the video model is superior to other kinds of models.

The primary training task of these studies was questioning

behavior. Koran, J. J., Jr.; Koran, M. L.; and McDonald, F.

(1972) found positive video models emphasizing both the

stimulus and response conditions to be superior to negative

models for training Stanford interns to use techniques of

asking observation and classification questions. Koran,

Snow and McDonald (1971) designed a video model treatment and

a written model to help teach interns to improve their abil-

ity to ask analytical questions. As might be expected the

video model was superior to the written model which was super-

ior to no treatment. The study also examined aptitude treat-

ment interactions between the model and learner abilities.

On two other occasions Koran, J.J.,Jr. (1970, 1971)

found the video model superior to a written and a self-rating

model for other dependent variables and other types of

conditions. Subsequently Koran, J. J., Jr., and Koran, M.L.,

(1973) compared two types of written models to determine rel-

ative effects on analytical questioning patterns of preser-

vice teachers. One group received a protocol model consisting

of a set of explicit definitions and sample questions. A

second group received a transcript of a teacher-pupil

interaction. Analytical behaviors were highlighted within

the communication. The control received no treatment.

Each of sixty-nine teachers prepared a twenty minute micro-

teaching session for eighth graders. The sessions were

audiorecorded and both teachers and students were ad-

ministered a written test. Upon analysis of the tapes the

protocol model proved to be more effective than the trans-

cript model for producing both types and frequencies of

analytical questions. On the written measures both

treatment groups scored significantly better than the

control except that here the transcript model subjects per-

formed better than the protocol model group. The pupils

of trained teachers scored significantly higher than the

control groups in being able to distinguish between types

of questions and in responsiveness.

Santiesteban (1974) in a study of classification and

observation question techniques compared a video model with

an audio model. No differences between video and audio

model subjects were found for any of the taped interaction

categories. Teacher and students of both model groups

performed significantly better than the control group.

This study suggests that if the competency to be acquired

is verbal that the audio model may be as useful as the

video model for training verbal behaviors.

As educational costs spiral Koran, J. J., Jr. and

Koran, M. L. (1974-1975) point out that training teachers

with modeling techniques is a promising alternative to

other more expensive methods. Observational learning pro-

cedures which are based on theory and research in psychol-

ogy as well as in education are both effective and inexpensive.

For these reasons modeling appears to warrant investigation

as a training procedure for competencies other than

questioning behavior. Training teachers to extend their

wait-time by means of observational learning has not been

explored or tested empirically. Hence,this study examines

modeling as a means of teaching wait-time.


Research that led to the theoretical development of

the wait-time variable was done primarily by Rowe (1974a,

1974b, 1974c). The idea of a pausing variable emerged as

a result of her work with various science curricula,

particularly Science Curriculum Improvement Study (SCIS).

While seeking the solution to another problem, an aware-

ness of pausing behavior developed. Although new curricula

generally upgraded the quality of science instruction in

the elementary schools, the quality and quantity of in-

quiry behavior did not improve correspondingly. In an

attempt to the discover the causes researchers like Rowe

explored the related areas such as adequate teacher train-

ing in the curricula, program deficiencies and the adequacy

of teacher science background. Analyses of these vari-

ables did not offer a satisfactory solution to the problem.

Even when these variables were controlled the level of

inquiry did not improve. Upon examining several hundred

tapes of different curricula taught in suburban, urban and

rural classrooms, a common factor of discourse became evi-

dent. In almost all cases the pace of instruction was


In the few examples of slower paced instruction,

classroom discussions more closely approximated the desired

inquiry behaviors. Children engaged in speculation, offered

alternative explanations, argued over interpretations of

the data and had longer speech sequences (Rowe, 1973,

1974a). In these classrooms pauses or wait-time averaged

about three seconds.

Wait-time is defined as the amount of time a teacher

waits in silence (a) after asking a question and (b) after

receiving a response. Typically teachers pace instruction

at a very fast rate with wait-time being under one second.

If the child does not begin to respond within one second,

the teacher either repeats the question or calls on another

child to answer. Upon receiving a response the teacher

usually waits less than one second before interrupting with

a comment or another question. Rowe (1973) noticed that

children's responses came in bursts with intermittent

pauses. By interrupting, the teacher fails to allow time

for the pupil to think through responses. Linked to rapid

pace is a high percentage of evaluative responses. This

combination serves to shift the focus from the instructional

phenomenon to the teacher-pupil relationship.

Under extended wait-time conditions Rowe (1973) identi-

fied ten changes that appear to take place in pupil behavior

and three in teacher behavior. The student outcome vari-

ables are as follows:

1. Student utterances increase in length. With short

wait-time students infrequently offer explanations of any

complexity. Data indicate that long wait-time II may be

responsible for the longer responses.

2. Unsolicited, but appropriate, responses increase in

number. This too seems to be affected more by wait-time II

than wait-time I.


3. Failure to respond decreases. When wait-time I

is short, no response or negative responses may be as high

as 30 percent.

4. Confidence in answers increases. Under long

wait-time fewer inflected responses are noted.

5. Speculative thinking increases. The reward sched-

ule appears to influence this variable.

6. Teacher centered discussion decreases and student-

student dialogue increases. With short wait-time and high

sanctioning schedules children rarely listen to one another.

7. Evidence is more often followed by or preceded by

statements of inference. Increase in wait-time II has a

desirable effect on the variable.

8. The number of child-asked questions increases

along with an increased number of child-proposed experi-


9. Contributions by "slow" children increase. With

slower paced instruction, responses come from a wider fac-

tion of the classroom.

10. Disruptive behavior decreases.

The three effects on teacher behavior which have been

identified as follows:

1. Teachers respond with greater flexibility as indicated

by fewer discourse errors. With a slowed pace the flow

of conversation does not become tightly structured.

2. The number and kind of teacher questions change.

As student responses become longer the number of teacher


questions decrease. The number of reflection and clarifica-

tion questions increase. The pattern of discourse is not

that of an inquisition as is often the case with short


3. Teachers' expectations of children.tend to change.

Thus it is evident that under prolonged wait-time a

number of desired changes in classroom verbal interaction

occur. Research other than Rowe's on the wait-time vari-

able has been fairly limited partially due to the fact the

research results have been published only recently. Two

doctoral studies examined the effects of wait-time on the

verbal dimension of classroom interaction.

Lake (1973) hypothesized that prolonged pauses would

contribute to both the type and complexity of student verbal

responses. Using the Bellack Classification Scheme, Lake

examined the empirical and evaluative cognitive processes

of classroom interaction. Seventy-two fifth graders were

randomly assigned to nine groups receiving either short or

long wait-time schedules. During the discovery phase of

the SCIS lesson, "Making Paper Airplanes,'! each group was

audio recorded and the questions were categorized according

to Bellack's Scheme. Analysis of results showed that ex-

tended wait-time contributed to a significant increase in

the incidence of conversational sequences and alternative

explanation in student talk. Pupil responses also increased

in cognitive complexity.

Garigliano (1971) reported somewhat discrepant results.

Initially he trained SCIS teachers to engage in extended

wait-times with the expectation of exploring the effects

of long wait-time on pupil variables. Outcome variables

included length of student responses, content oriented

student solicitation, inflected responses, I-don't-know

responses and pupil-pupil interactions. Only one variable

reached statistical significance, length of student re-

sponse. Considering, however, that none of his trained

teachers reached a wait-time of three seconds, it is question-

able whether any valid conclusions regarding the effects of

wait-time can be drawn.

Rowe's research (1974a) indicates that wait-time of

less than 2.7 seconds did not apparently contribute to any

of the postulated effects. Above 2.7 seconds and up to

4.5 seconds the output variables improved in value. In

this study a criterion wait-time of three seconds was


Other findings concerning the conditions and factors

affecting wait-time patterns were documented primarily by

Rowe (1974a, 1974b, 1974c). The model used throughout

this study for examining teacher-student dialogue was one

she conceptualized. The teaching unit is considered a

two-player model with the teacher treated as one player

and the group of students as the other player. With this

model moves between players would be expected to be more

or less equal. In fact research notes that most verbal


patterns are teacher dominated. Rapidly paced instruction

produces an interaction pattern that Rowe dubs inquisition.

Slower paced instruction results in an inquiry pattern with

all players employing the available moves equally. The

patterns exhibited by four students were found to be typi-

cal of patterns displayed by an entire classroom. How-

ever, generalizing the results of experiments in micro-

teaching settings to classroom as a whole should be done

with caution.

Another variable of considerable importance to the

inquiry process also became apparent during the research

analysis. Reward patterns influenced interaction con-

siderably (Rowe, 1974c; Lawlor, 1972). Students falling in

differing categories not only received different reward

schedules but were given different amounts of wait-time.

Teachers more frequently responded with nonevaluative re-

marks and the remarks were more likely to be appropriate

whin interacting with high ranking students. On the other

hand lower ranked students received a greater number of

positive rewards but the rewards were not necessarily

appropriate. For these pupils a random reward pattern

resulted with both correct and incorrect responses being

rewarded. Teachers allowed the lower groups less time to

think out answers. Wait-time, particularly species II, was

shorter for the low groups. Apparently teacher expecta-

tions developed early with reward patterns corresponding

to expectations.

These ideas are borne out in two other studies. Camp-

bell and Rose (1974), in a post hoc analysis of junior

high teachers' behaviors, confirmed Rowe's data. In ten

classes of low ability children and ten classes of high

ability children the incidence of three second wait-time

was generally low especially wait-time II. When wait-time

did occur, it was almost twice as likely to be with high

ability groups as with low ability groups.

Because research reveals that reward schedules can

greatly affect interaction, the model maintained a neutral

or reward free atmosphere.

A doctoral study by Chalker (1972) divulged another

interesting aspect of wait-time. When comparing social

studies teachers' scores on a measure of dogmatism with the

time they were likely to pause, she found teachers who

scored high on a measure of dogmatism less likely tospause.

Thus it appears that personal characteristics may interact

with a teachers' likelihood of engaging in extended wait-

time. If this be true, differential training procedures

might be expected to achieve the greatest results.

Wait-time Training

The review of the literature on wait-time training

made evident the unsolved problem of finding a successful

and economically expedient training procedure.

Rowe (1973, 1974a)had a success rate of 70 to 80 per-

cent with the method she employed. Ninety-six teachers


engaged in a series of three teach-reteach cycles. The

teachers taught the same four students in each cycle and

received specific feedback. She found during the training

cycles that certain verbal patterns of teachers interfered

with the acquisition of species II wait-time. Rowe de-

scribed these as follows:

1. Mimicry in which teachers repeat portions of

what the students say.

2. Two constructions which signal rejection of an

idea; Yes ... but;... though....

3. The command to "think" without providing either

a pause or cues concerning what one is to think about.

4. Evaluative comments following a student's

statement; fine, good. OK, right.

5. Questions which begin with the format, "Why

did you do that?" (Rowe, 1973).

Eliminating these speech patterns helped teachers

to achieve criterion wait-time II better but confounded

the results of the outcome variables. By manipulating

the two variables, speech patterns and wait-time, Rowe

was able to ascertain the separate effects on the ten

pupil outcome variables.

In early experiments of wait-time measurement proved

to be a technological problem. A stopwatch was too cumber-

som to measure pauses accurately. As might be imagined

motor control of watches and student discourse patterns

did not correspond very precisely.-

Difficulties with measurement were somewhat reduced

by transferring the sound from the tapes onto a servo-

chart plotter. The needle of the chart plotter was made to

track horizontally during silent pauses. Although a tre-

mendous improvement over the stopwatch, the chart plotter

still has some technical difficulties. One problem is that

the background noises are picked up and recorded on the

output making it necessary for someone to monitor each tape.

Despite the chart plotter's frailties the validity of ex-

periments measuring wait-time would be questionable with

out the aid of an objective measuring device.

A few other examples were located in which wait-time

was the manipulated variable and in which teachers were

trained to achieve extended wait-times. In the Lake (1973)

study wait-time was the only manipulated variable. However,

he conducted the instructional sequences himself and did

not train teachers. Lake practiced achieving long wait-

time in a pilot study that he conducted to familiarize

himself with the lesson-materials. Although training teachers

was not a part of his research, he offered a number of

suggestions for training including the use of video tapes

with cues and feedback sessions.

Garigliano (1972) devised a procedure in which teachers

were audio-taped during training sessions. After trans-

cribing each tape Garigliano and the teacher listened to the

tape while following the transcript for verbal rewards and

discourse patterns. Teachers were asked to concentrate on


the length of wait-time while the researcher rough counted,

1001, 1002, 1003. . The following teaching session was

the experimental session in which both wait-time and out-

come variables were measured. The single treatment was not

sufficient for teachers to reach criterion wait-time.

Garigliano's three experimental groups did not differ sig-

nificantly. Wait-time for all teachers ranged from 0.04

seconds to 2.69 seconds. Regardless of training some teachers

permitted longer pauses in discourse than others.

Garigliano's training procedures were similar to ones

used by Moriber (1971) with four college level instructors.

Although the results were confounded by an evaluation fac-

tor, his subjects did not reach criterion wait-time either.

Thus it seems imperative to find a training method that

yields good results.

Summary of the Rationale for the Hypotheses

The rationale for the hypotheses posed may be sum-

marized as follows:

1. The movement toward competency based teacher

education points out the need for identifying appropriate

teacher competencies. The competencies which are selected

as desirable traits for teachers should have both a theo-

retical and empirical framework with some evidence that

the variable will have an effect on learner outcomes. Once

competencies are identified instructional strategies that

produce maximum teacher gains need to be developed. In

_ _


addition to helping the teachers reach a criterion level of

performance for the competency, the instructional procedures

need to be economically feasible.

2. Rowe's research shows that teacher pausing be-

havior has considerable impact on the inquiry behavior of

children engaged in science activities. Extended wait-time

has been shown to cause change in at least ten learner out-


3. Training procedures for instructing teachers to

reach criterion wait-time have not met the criterion of

statement one. Some methods have proved ineffective; other

more successful ones require a long series of teaching


4. Modeling as a training procedure is known to be

successful for teaching certain verbal behaviors. Al-

though modeling research in teacher education has dealt

primarily with teacher questioning behavior, it is expected

that modeling would be equally effective for training other

verbal behaviors like wait-time.

Statement of the Hypotheses

This work is an extension of research already done on

wait-time and modeling. The purpose is to combine the

lines of research in seeking solutions to the problems of

finding appropriate training procedures and further clarify-

ing the boundary conditions of the variable, wait-time.

With this basis the following research hypotheses were

tested:at the 0.05 level of significance.

1. The mean length of teacher's wait-time I and II

as measured by servo-chart analysis will significantly in-

crease upon viewing a video model.

2. The mean length of teacher wait-time I and II as

measured by servo-chart analysis will significantly increase

upon observing an audio model.

3. The mean length of wait-time I and II of teachers

viewing the video model will be significantly greater than

the mean length wait-time I and II of teachers who observe

the audio model.

4. Mean length of wait-time I and II of teachers who

receive feedback will be significantly greater than the

mean length wait-time I and II of teachers not receiving


5. The mean length of teacher talk as measured by

servo-chart analysis will decrease significantly for

teachers using extended wait-time.

6. The mean length of student response as measured by

servo-chart analysis will increase significantly with

teachers using extended wait-time.

7. The total proportion of time the teacher controls

the speak-space as measured by servo-chart analysis will

decrease significantly for teachers using extended wait-


8. Student performance scores on a test of process

skills will be significantly higher for teachers using

extended wait-time.



The Design

A modified pretest-posttest design for repeated mea-

sures was uses in this study. The design, as described

in Table I, was selected instead of a posttest only design

because it offered two advantages. Entry behavior for

preservice teachers would be measured making the assump-

tion of equivalence between groups unnecessary. Secondly

because the potential number of subjects for the study was

limited, the design for repeated measures afforded larger

cell sizes.

Table 1

Experimental Design for Phase One

Treat- Treat-
Measure TreatMeasure Measure
ment ment

Teachers Teachers

S R R X3 0
01 R X1 O2
R --- 03

R X 0
01 R X2 0 R3 03
R --- 03

X1 = video model, X2 = audio model, X3 = feedback,
01 = recording of teaching Session 1, 02 = recording of
Session 2, 03 = recording of Session 3.



In the study fifty-two preservice elementary teachers

conducted three fifteen minute microteaching sessions.

Pretest data were collected during the first session.

Teachers were then randomly assigned to two treatment groups.

Group I received an audio taped treatment and Sroup II re-

ceived a video taped treatment. The second set of observa-

tions were collected during the second teaching session.

This phase of the experiment appears diagramatically in

Table 1 also.

At the end of microteaching session two, the two groups

were again divided. Half of each treatment group was

randomly assigned to a feedback or a no feedback group.

A third teaching session was conducted and the third set of

observations were collected.

For phase two the outcome variables were calculated

from the taped recordings of each session and the written

measure was administered to each child at the end of each

microteaching session. The microteaching groups consisted

of a preservice teacher and four fourth or fifth graders.

The children were randomly assigned to teachers and four

groups were formed according to the teacher's treatment

group; audio-no-feedback, audio-feedback, video-no-feedback,

video-feedback. The experimental design at this phase is

shown in Table 2.

Internal Validity

Stanley and Campbell (1963) describe possible threats

to internal validity. Many of these are controlled in the


pretest-posttest design. As the design utilized in this

study was modified somewhat control of threats to validity

are described below.

Table 2

Experimantal Design

for Phase Two

Treat- Treat- Treat-
ment Measure ment Measure ment Measure

X1 01 02 R X2 03 04 X3 05 06

X1 01 02 R X2 03 04 X3 05 06

X1 01 02 R X2 03 04 X3 05 06

X1 01 02 R X2 03 04 X3 05 06

X1 = teaching Session 1, X2 = Session 2, X3 = Session
3, 01 = recording from Session 1, 02 = test 1, 03 = record-
ing from Session 2, 04 = test 2, 05 = recording from ses-
ion 3, 06 = test 3.

History and maturation were controlled by the short

duration between teaching sessions. Since wait-time is

calculated from tapes and not from written measures, test

sensitization was not expected to occur.

Instrumentation is a potential problem due to the

manner in which wait-time is calculated. As the verbal

behaviors were transferred to a chart-plotter, care was

taken during recording to prevent background noises and

interference. When noise is picked up on the tape, it is

transferred to the chart recorder vertically instead of

horizontally. Additionally each tape was monitored by the

experimenter and extraneous noises were marked on the


Statistical regression was not expected as both stu-

dents and teachers were randomly assigned and not blocked.

Since all three treatments did not take place on the same

day some mortality occurred. The chance of mortality is

equalized over groups and treatments through randomization

of assignments. Selection was also controlled by randomiza-


External Validity

Although internal threats to validity may be controlled

through design and randomization to be equalized between

groups, threats to external validity may pose particular

problems to experiments of this nature. Bracht and Glass

(1968) describe some of the more common sources of error.

Threats to external validity are divided into two broad

classes, population validity and ecological validity.

Error related to population validity centers around

selection of the sample population and the interaction of

personological variables. In this study there is little

reason to believe that the sample population, a group of

preservice teachers at the University of Florida, is char-

acteristically different from the target population, pre-

service teachers in general. Generalizing to a target

population of teachers at large is less valid because the

effect teaching experience on training wait-time has not

been explored.

Regarding interaction effects some are expected to

occur. The range of wait-time exhibited by teachers

varies considerably. While most teachers do not permit long

pauses of up to three seconds, there is wide variability

in the 0.0 to 3.0 second range. Garigliano (1971) found

a range of 0.4 seconds to 2.69 seconds in his sample. Per-

sonal aptitudes may conceivably account for much of the

variation. Because continued research is needed to examine

the interaction effects more closely, a post hoc analysis

of the data is planned.

A question associated with treatment is whether or not

the changes in behavior like extending wait-time are stable

over time. Will the teachers who acquire criterion wait-

time as a result of treatment continue to demonstrate the

characteristic without further treatment? Due to the short

duration of the experiment the question cannot be adequately

answered. As treatment was generally successful a longi-

tudinal study on wait-time stability would be an appropriate

follow up study.

In the design it would appear that multiple-treatment

interference would be a threat. Teacher subjects were sub-

jected to one treatment one day and a related treatment a

day later. The question raised in the study is whether or

not feedback is a necessary component of the treatment.

It is not considered a separate treatment. In order to


make a decision about the importance of feedback each treat-

ment group was divided into a feedback and no feedback

group. Statistical differences between groups were ex-

plored as well as differences in change between groups.

The Hawthorne effect could possibly influence results

but probably will not for two reasons. Alachua County

children frequently experience teaching situations with

preservice teachers because inservice teaching is an inte-

gral part of the teacher education at the University of

Florida. Thus for the pupil subjects it was not a parti-

cularly unusual situation. Also the children did not know

the nature of the experimental situation nor which variables

were being measured.

Novelty effects were controlled as much as possible

logistically. Microteaching sessions took place in the

school in a quiet room to keep down noise interference on

the tapes. The recording equipment, though placed as un-

obtrusively as possible, was visible. The children were

given the opportunity to experiment with the tapes and to

listen to themselves before the teaching sessions began in

order to reduce curiosity.

Although the design is a pre-posttest design, entry

wait-times were calculated from tapes and preservice

teachers were not aware that it was the variable being mea-

sured. Test and treatment interactions were therefore not


Because treatments occur with only one day intervals,

interaction of history and treatment should not bias the

results. Measurement of the dependent variable, wait-time,

and student outcomes were tallied from the tapes and servo-

chart plots by established criteria. The written measure

was administered immediately after each teaching session.

Within the limits of the experimental situation, every effort

was made to control threats to both internal and external validity.

Treatment Procedures


Fifty-two, college-level, junior and senior, preservice

teachers enrolled in the Childhood Education Program at the

University of Florida volunteered to be subjects in the

study. The subjects, all working in Science Methods, re-

ceived Science Learning Activity Credit for participating

in the study. The 208 microteaching, pupil subjects were

enrolled in three Alachua County elementary schools. A

breakdown of students by grade and school showed that:

one hundred-five fifth-graders attended Terwilliger School

in the Northwest section of town; fifty-six fourth-graders

attended Idlewild School in the Southwest section of town;

forty-seven fourth-graders attended Duval School in the

Northeast section of town. These schools represent a

cross-section of the city and have an overall 30 percent

minority population. The neighborhoods from which the pupil

subjects came included all economic ranges. All pupil subjects

were randomly assigned to micro-teaching groups.

General Procedures

Preservice teachers in the science methods class parti-

cipated in the project as an option to a regualr learning

activity. After a general introduction to experimental

procedures and time requirements they were asked to sign

up for a convenient sequence of microteaching sessions.

At the preteaching introductory session subjects were given

the set induction materials to explore and inquire about.

The purpose of this was to familiarize the subjects with

the basic concepts of the experimental tasks equally.

The three microteaching sessions were conducted dur-

ing two time blocks. The two lessons taught during the

first session were the "Cartesian Diver" and the "Bouncing

Raisins" tasks. The model treatment took place between the

two inquiry lessons. During the second block of time

subjects received feedback and thaght lesson three,

"Rolling Cylinders." The schedule for a complete experi-

mental sequence is presented in table 3.

Three microteaching groups were conducted simulta-

neously in one room. Teachers arrived about fifteen

minutes before the children to read over the lesson ma-

terials again. When the twelve children arrived from

their classroom, the task which focused on a discrepant event

was demonstrated by the researcher. Afterwards the chil-

dren had the opportunity to manipulate the experimental

materials. For example, in the "Rolling Cylinder" task

several children compared the relative weights of the

cylinders. Prior to the demonstration the researcher

Table 3

Treatment Procedures and Times

Day 1 Time Day 2 Time Day 3 Time


Set Induction Set Induction 20 Set Induction 10

a) Task 1 10 Demonstration 10 Fedback
Treatment 20
b) Task 2 10 Discussion 15 Demonstration 10

c) Task 3 10
Audio or Visual 20 Discussion 15

Demonstration 10

Discussion 15

Total time* 30 85 55


Demonstration 10 Demonstration 10

Discussion 15 Discussion 15

Test 1 15 Test 3 15

Demonstration 10

Discussion 15

Test 2 15

Total Time* 80 40

*Time is calculated in minutes.

instructed the children to observe carefully and to think

about what they observed but to wait until seated with a

teacher to discuss the tasks and ask questions.

With no experimental materials present the teacher and

pupils verbally explored the tasks for approximately fifteen

minutes. Following the discussion the teachers adjourned

to another room to observe a model and the children an-

swered the short fifteen item quiz. After observing the

model, teachers had the opportunity to teach a second in-

quiry lesson using long wait-time. The demonstration and

discussion followed the pattern of the first lesson with

the second quiz ending the day's session.

For the last teaching sequence, teachers arrived

thirty minutes before the children for the feedback treat-

ment. The feedback group listened to their tape from the

second teaching session. While they were listening to the

tape they were asked to determine whether or not they were

reaching three second wait-time. Time was estimated by

rough-counting the pauses, 1001, 1002, 1003. . The no

feedback group read an article related to teaching inquiry

by Rowe (1975). At the last inquiry session both groups

were instructed to try to exhibit extended wait-time be-

havior and set induction materials for the third task,

"Rolling Cylinders," were distributed. Demonstration, dis-

cussion and student testing followed.

The total experimental procedures in the schools were

completed in two weeks. The first week at Terwilliger, four


groups of twelve were conducted on a Monday-Wednesday

schedule and four on a Tuesday-Thursday schedule. The

second week four groups followed a Tuesday-Thursday sched-

ule and four followed a Wednesday-Friday schedule. The

latter week, Idlewild was tested in the morning and Duval

in the afternoon.

Treatment Materials

The Models

The first part of the video model consisted of four

minutes of a task demonstration, "Boiling by Cooling," which

was viewed by both groups. The latter part, of seventeen

minutes duration, consisted of a female teacher and four

fifth-grade students discussing the task. The models

clearly demonstrated three second pauses of both wait-time

species. The audio model was a taped recording of the

video model. Prior to observing either model, the teachers

were given a short, written description of wait-time that

included a definition of variables. The purpose of the

written description was to ensure that the subjects would

attend to and focus upon the desired behaviors.


A total of four tasks, depicting discrepant events,

were used in the study, three for the microteaching ses-

sions and one for the model. The following criteria were

established for selecting the tasks: a) the tasks should

focus on a single discrepant event; b) the topic should


have high initial interest level; c) the topic should be

appropriate for the grade level of the pupil subjects;

d) the topic should be sufficiently novel that preservice

elementary teachers would not be familiar with it; e) the

tasks should be sufficiently complex to provoke a fifteen

minute inquiry discussion; and f) the materials should be

inexpensive, easily available, and portable.

After careful review of science-methods learning

activities and various activity-oriented, elementary,

science curricula, four tasks were selected which met the

above criteria. These were "Boiling by Cooling,""The

Cartesian Diver," "Bouncing Raisins," and "Rolling Cylin-



Measuring the dependent variables required three

steps. The first step involved tape recording every micro-

teaching group at each teaching session. Recording equip-

ment included Wallensak audio-cassette recorders and Scotch

Brand extended-range, high-density, cassette tapes. Next

the audio signal from the tapes was transferred onto servo-

chart plots. The Heath-Kit chart recorder used for this

phase required technical modifications for audible sound.

This enabled the researcher to monitor each tape and to

identify wait-time species and speakers. Step three in-

volved the calculation of each dependent variables; wait-

time I, wait-time II, mean length of teacher talk, mean

length of student response, and proportion of teacher talk.


Transferring the sound onto the servo-chart proved to

be the most difficult and tedious part. All the extraneous

sounds, coughs, chair-squeaks, outside noises, et cetera,

were transferred onto the chart along with the desired

group discussion. These noises resulted in the needle

tracking vertically during silences when it should have

tracked horizontally. Additionally, on some tapes static

noise from the magnetic heads of the tape recorders was

transferred onto the servo-chart causing horizontal track-

ing to be irregular. In order to make distinctions be-

tween nuisance noise and discourse, the experimenter moni-

tored each of the 156 tapes. While monitoring, the tapes

were annotated to indicate teacher talk, student talk,

wait-time I and wait-time II.

Calculating wait-times was based on guidelines estab-

lished by Rowe (1973). Wait-time I is measured from the

time the teacher stops speaking until the time a student

responds. If the teacher asks a question, pauses, or calls

on another student and pauses again, the two pauses are

summed. Wait-time II consists of summing all pauses from

the time a student response begins until the teacher inter-

venes. Thus, if one student responds and pauses and a

second student responds and pauses, wait-time II is the

sum of the two pauses.

Using the two player model of dialogue, all four stu-

dents are considered to be a single player with the teacher

being the other player. Every teacher-student exchange is

_ _

considered to be a conversational set. If the teacher

intervenes with any response, i.e. "okay...good...all

right...uh huh," the conversational set is terminated.

For clarity, the term speak-space is used throughout the

discussion to connote the time occupied by a speaker during

a conversational set; that is, the dialogue which is stated

by a single player between pauses.

Mean length of each species of wait-time was calcu-

lated by tallying the total number of seconds for each wait-

time occurring during a four-minute discussion sequence,

and by deviding the total number of conversational sets

for a four-minute period. Two four-minute periods were

averaged to determine the mean for wait-time I and II. The

chart-plotter made possible accurate measurement of wait-time

to within 0.25 seconds. The speed of the chart drive was

set to move at the rate of 0.2 inches per second, with the

chart paper calibrated to 0.01 inch segments. Measuring

segments consisted of counting the calibrated distances the

pen tracked horizontally.

Mean length of teacher talk and student talk were cal-

culated by summing the seconds the teacher and students spoke

respectively and dividing by the number of teacher-student

exchanges. The ratio of student-teacher talk was measured as

a percent of the time the teacher occupied each speak-space.

Determining percent was done by dividing the total number of

seconds of teacher talk by the total number of seconds for

recorded period and then multiplying by 100.

Student Tests

At the conclusion of each micro teaching session,

pupil subjects were given a fifteen item, multiple choice

test. The students had the option of reading the test

silently and answering questions or of listening to the

questions being read aloud and then selecting answers.

The test questions were constructed to focus on three in-

quiry skills. One-third of the questions tested simple

observation skills. Another third required the children

to make an inference in order to answer the question cor-

rectly. Another third tested their knowledge of inquiry

vocabulary. The set induction materials given the

teachers included a list of process words they would find

useful for discussion and a list of behavioral objectives

for children which included observation and inference be-

haviors. The purpose of the set induction materials was

more for support, to assure teachers that their background

knowledge of the task was sufficient for a fifteen to

twenty minute discussion, than for a specific guide of

what to teach.

Test reliabilities which were calculated for the three

measures are reported in Table 4.

Reliability for the test measures was calculated using

the Kuder-Richardson formula 21 for reliability, which

gives reliability coefficients at least equal to, though

sometimes lower than, the Kuder-Richardson formula 20 or

the Spearman-Brown formula for split-halves (Ebel, 1972).

Table 4

Reliability of Written Measures

Measure N No. of Items Reliability

Test 1 208 15 .86

Test 2 208 15 .80

Test 3 204 15 .91



The main objectives of this study were: 1) to compare

the relative effects of two training procedures, video and

audio modeling, with and without feedback; and 2) to assess

the relative effects of high and low teacher wait-time on

teacher-student dialogue patterns and a student performance


Statistical assessment of the hypotheses will be pre-

sented in the following order: treatment effects for wait-

time I and II; the effects of treatment and wait-time on

the dialogue variables and performance measures. Treat-

ment effects were analyzed by analysis of variance.

Outcome variables were assessed by analysis of variance

for group effects. A modified Tukey's HSD Test was used

for comparison among means. Multiple regression analysis

was used to compare wait-time and the outcome variables

for each microteaching session to determine shared


Biomedical Computer Programs (BMDP, 1975) and the

Statistical Package for Social Sciences (SPSS, 1975) were

the statistical programs used to analyze the data.

Treatment Main Effects

The main effects were evaluated by comparing the

averaged wait-time means of each group for each teaching

session. The average wait-time means for subjects were

calculated from the audio taped recordings.

Because the knowledge of entry behavior was desired and

because treatment occurred at two levels, the design appro-

priate for analysis was a factorial design with block and

treatment confounded, the split-plot factorial-pr.q. (Kirk,

1968; Hays, 1963). In these analyses p levels of ai equal

two audio and visual; q levels of b. equal three, micro-

teaching Sessions 1, 2 and 3; and r levels of ck equal two,

feedback and no feedback. The design has two between block

treatments (A = modeling treatment and C = feedback), and

one within block treatment (B = microteaching session).

The data met the assumptions that npr subjects 'are

randomly assigned to modeling, feedback treatments with n

blocks within each pr level. The underlying structural

model for the design which test the hypotheses in the null,

or no difference, mode is presented below.

Main Effects

Analysis of wait-time II is described separately from

wait-time I, because wait-time II is controlled only by the

teacher and is therefore more likely to be influenced by

treatment than wait-time I. By definition wait-time II is


that period of time between the end of a student response

and the beginning of teacher talk. Wait-time I is the

period between the time the teacher completes a response

and the student intervenes. Thus, control of wait-time is

shared by the teacher and the student since both have equal

access to the speak-space after a teacher question. As

pupil subjects received no training, their portion of con-

trol of wait-time I would not be directly subject to treat-

ment effects. Consequently the data on wait-time II are

discussed before the wait-time I data.

Wait-time II

The entry level wait-time II mean for all subjects was

0.60 seconds with no significant differences between any

of the groups. Cell means, standard deviations and treat-

ment means for all wait-time II groups are listed in Table 5

and the analysis of variance summary table examining group

mean differences are reported in Table 6. Analysis of

variance for repeated measures revealed significant dif-

ferences between the microteaching sessions (F = 44.16,

p< 0.001). Tests for significance between means using

Tukey's HSD showed that the wait-time II for Session 2 and

Session 3 were significantly higher than Session 1. The

Session 3 wait-time II mean of 2.83 seconds was not sig-

nificantly higher than the Session 2 wait-time II mean of

2.32 seconds. The summary for Tukey's HSD Test is listed

in Table 7.

Table 5

Means and Standard Deviations for Wait-time II

Audio Audio Visual Visual Marginal
No Feed No Feed

Session 1
Mean 0.52 0.64 0.66 0.56 0.60
S.D. 0.55 0.54 0.37 0.33

Session 2
Mean 1.77 1.99 2.89 2.61 2.32
S.D. 1.44 1.02 1.90 1.49

Session 3
Mean 2.21 2.78 2.71 3.70 2.87
S.D. 1.31 1.75 1.13 2.32

Marginal 1.50 1.81 2.09 2.29

Count 12 13 13 14 52

Table 6

Analysis of Variance for Wait-time II

Source SS df MS F

Model (A) 9.83 1 9.83 3.96*
Feedback (C) 2.20 1 2.20 .89
AC .09 1 .09 .03
Error 104.24 42 2.48

Session (T) 126.99 2 63.49 44.16**
TA 4.55 2 2.28. 1.58
TC 4.77 2 2.38 1.66
TAC 1.26 2 .63 .44
Error 120.77 84 1.43

*p <.05


Table 7

Tukey's HSD Test for Difference Between

Means Wait-time II

Treatment Cell Mean Cell N

1) Microteaching
Session 1 .598 46

2) Microteaching
Session 2 2.32 46

3) Microteaching
Session 3 2.87 46

Contrasted pairs 1-2 1-3 2-3

Differences between
Means 1.72* 2.27* .55

Tukey's HSD: .644

*p < 0.05

For the between block treatments, a significant dif-

ference was found between the modeled treatment subjects.

The video model group achieved a significantly higher

wait-time mean than did the audio group (F = 3.96,

p< 0.05). The wait-time II mean for the feedback group

was not significantly higher than the mean of the no feed-

back group.

The percent of subjects reaching the desried cri-

terion wait-time of 3.o seconds for each treatment group

at Session 2 and Session 3 is reported in Table 8. No

subject reached criterion for Session 1. The subjects

not reaching the three second criterion are separated in-

to groups, those achieving wait-time between two and

three seconds and those who do not reach two seconds.

Although some subjects failed to reach the two second

level, the amount on increase was usually significant.

For example, althoughone subject reached a post treatment

wait-time II of only 1.93 seconds, her entry level wait-time

II was 0.38 seconds.

Table 8

Percent of Subjects at Criterion Wait-time II

Group A B C

Session 2

Video 30% 19% 51%

Audio 21% 21% 58%

Session 3

Feedback 62% 31% 08%

No Feedback 50% 31% 20%

Feedback 54% 23% 23%

No Feedback 33% 25% 42%

A = > 3.0 seconds, B = 2.0-2.99 seconds, C = < 2.0 seconds

Wait-time I

The average entry level mean for wait-time I was 0.99

seconds which was slightly higher than the entry wait-time

II mean. However wait-time I did not increase in magnitude

between sessions nearly as much as wait-time II. None of

the group means reached two seconds. The highest wait-

time I achieved by any subject was 2.95 seconds. Cell

means, treatment means and standard deviations are reported

in Table 9.

As with wait-time II, analysis of variance revealed

significant differences between microteaching sessions

although the magnitude of change was considerably smaller

(F = 11.10, p < .01). No differences between block treat-

ment groups were found to be significant but a significant

interaction occurred between the type of model and effect

of feedback (F = 4.21, p < 0.05). Wait-time I analysis of

variance summary is listed in Table 10. The breakdown of

significance between group means revealed a significant dif-

ference between microteaching Sessions 1 and Session 3, but

not between Session 1 and Session 2 nor between Session 2

and Session 3. Tukey's HSD Test also revealed that the mean

of the audio feedback group was significantly higher than

either the audio no feedback group or the visual feedback

group. The summary of Tukey's HSD Test is listed in Table

Table 9

Means and Standard Deviations for Wait-time I

Audio Audio Visual Visual Marginal
No Feed No Feed

Session 1
Mean 0.77 1.21 1.24 0.74 0.99
S.D. 0.48 0.85 0.64 0.33

Session 2
Mean 0.72 1.77 0.85 0.79 1.04
S.D. 0.49 1.68 0.59 0.51

Session 3
Mean 1.50 1.56 1.60 1.65 1.58
S.D. 0.84 0.49 0.76 0.84

Marginal 0.99 1.52 1.23 1.06 1.20

Count 12 13 13 14 52

S.D. = standard deviation

Table 10

Analysis of Variance for Wait-time I

SS df MS F

Model (A) 0.43 1 0.43 0.44
Feedback (C) 1.03 1 1.03 1.05
AC 4.13 1 4.13 4.21*
Error 41.24 42 0.98

Session (T) 9.82 2 4.91 11.10**
TA 1.72 2 0.86 1.95
TC 1.79 2 0.89 2.03
TAC 2.00 2 1.00 2.26
Error 37.13 84 0.44

*p < 0.05
**p < 0.01


Table 11

Tukey's HSD Test for Difference Between
Means Wait-time I

Treatment Cell Mean Cell N

1) Microteaching
Session 1 .99 46

2) Microteaching
Session 2 1.04 46

3) Microteaching
Session 3 :1.58 46

Contrasted pairs 1-2 1-3 2-3

Differences between
Means .05 .59* .54*

Tukey's HSD: .405

*p 0.05

Outcome Variables

The variables, mean length of teacher talk, mean length

of pupil response, proportion of teacher talk and the stu-

dent performance measure, were analyzed according to group

membership with analysis of variance. Because of the large

differences found to exist between microteaching sessions

for wait-time II, each teaching session was analyzed sepa-

rately by a stepwise multiple regression with wait-time as

the dependent continuous variable. In the analyses scores

on the outcome variables were regressed on wait-time scores

with the dependent variables entering the equation in order

of significance.

Analysis of variance results for group differences

of each outcome variable are reported first followed by the

multiple regression analysis for each of the three micro-

teaching sessions. When reporting the statistical re-

sults for the outcome variables the following abbreviations

will sometimes be used: mean length of teacher talk (MLTT),

mean length of student talk (MLST), proportion of teacher

talk (PTT), and student performance test (SPT).

Mean Length of Teacher Talk

The data for mean length of teacher talk revealed that

teachers talked an average of 6.74 seconds every time they

occupied the speak space for the pretreatment session 1.

Cell means, treatment means and standard deviations for

MLTT are reported in Table 12. Analysis of variance showed

significant differences between microteaching sessions

but not between treatment groups. Tukey's HSD Test indicated

that the mean length of teacher talk decreased significantly

from 6.74 to 4.86 seconds between Session 1 and Session 2

(F = 18.81, p < 0.01). The decrease proved to be only a

temporary change because the mean of the third session in-

creased to 6.89 seconds which is essentially equal to the

Session 1 mean. An analysis of variance summary table for

mean length of teacher talk is listed in Table 13 and Tukey's

HSD Test in Table 14.

Table 12

Means and Standard Deviations for
Mean Length of Teacher Talk

Audio Audio Visual Visual Marginal
No Feed No Feed

Session 1
Mean 5.99 7.68 6.92 6.34 6.74
S.D. 1.24 2.42 2.46 2.69

Session 2
Mean 4.81 5.39 4.92 4.31 4.85
S.D. 1.52 2.35 1.21 1.19

Session 3
Mean 6.44 7.21 6.48 7.39 6.89
S.D. 2.28 2.13 1.57 2.31

Marginal 5.75 6.76 6.11 6.02 6.17

Count 12 13 13 14 52

S.D. = Standard Deviation

Table 13

Analysis of Variance for Mean
Length Teacher Talk

SS df MS F

Model (A) 1.31 1 1.31 .65
Feedback (C) 7.31 1 7.31 1.18
AC 10.53 1 10.53 1.70
Error 259.85 42 6.19

Session (T) 116.65 2 58.33 18.81*
TA 2.00 2 1.00 .32
TC 4.30 2 2.15 .63
TAC 8.28 2 4.14 1.34
Error 260.45 84 3.31

*p < 0.01

Table 14

Tukey's HSD Test for Differences Between
Mean Length Teacher Talk

Treatment Cell Mean Cell N

1) Microteaching
Session 1 6.74 46

2) Microteaching
Session 2 4.86 46

3) Microteaching
Session 3 6.89 46

Contrasted pairs 1-2 1-3 2-3

Differences between
Means 1.89* .14 2.04*

Tukey's HSD: .881

*p < 0.05

Mean Length of Student Response

As noted in Table 15 in which cell means, treatment

means and standard deviations are listed, the mean length

of student response almost doubles from 4.59 seconds to

9.79 seconds between Session 1 and Session 2 and remains

high for Session 3. These differences are found to be

very significant (F = 33.35, p < .001). Although the

differences between treatment groups were not found to be

significant (F = 1.83, p = .16) the MLST for the video

groups at Session 2 are more than two seconds longer than

the audio groups. These distinctions disappear or are

possibly confounded by the effect of feedback for Session 3.

Analysis of variance and Tukey's Test are reported in Tables

16 and 17 respectively.

Table 15

Means and Standard Deviations for
Mean Length Student Talk

Audio Audio Visual Visual Marginal
No Feed No Feed

Session 1
Mean 4.32 4.73 4.28 4.98 4.59
S.D. 1.78 2.75 1.11 2.67

Session 2
Mean 8.08 8.89 11.93 10.33 9.79
S.D. 5.00 4.38 8.53 4.77

Session 3
Mean 7.83 8.54 10.96 9.68 9.24
S.D. 3.43 5.44 6.02 5.99

Marginal 6.74 7.38 9.06 8.33 7.88

Count 12 13 13 14

S.D. = Standard Deviation

Table 16
Analysis of Variance for Mean
Length Student Talk

SS df MS F

Model (A) 91.75 1 91.75 2.02*
Feedback (C) .05 1 .05 .00
AC 16.06 1 16.06 .36
Error 1894.54 42 45.12
Session (T) 757.57 2 378.78 33.36**
TA 41.47 2 20.73 1.83*
TC 6.11 2 3.06 .27
TAC 12.17 2 6.09 .54
Error 953.91 84 11.35

* p = 0.16
**p < 0.001

Table 17

Tukey's HSD Test for Differences Between
Mean Length Student Talk

Treatment Cell Means Cell N

1) Microteaching
Session 1 4.59 46

2) Microteaching
Session 2 9.79 46

3) Microteaching
Session 3 9.29 46

Contrasted pairs 1-2 1-3 2-3

Differences between
Means 5.21* 5.08* 0.50

Tukey's HSD: 1.684

*p < 0.05

Proportion of Teacher Talk

As with mean length of teacher talk and mean length of

student response the proportion of teacher talk was found to

be significantly different for microteaching sessions but

not for treatments (F = 99.64, p < .001). For teaching

Session 1 teachers talked about 60 percent of the time which

decreased to about 35 percent for Session 2. Session 3 in-

creased about 10 percent from Session 2. Cell means, treat-

ment means and standard deviations are listed in Table 18.

The analysis of variance summary is reported in Table 19

and Tukey's HSD Test is in Table 20. For ease of reading

the proportions in these tables are expressed in percent.

Table 18

Means and Standard Deviations for
Proportion of Teacher Talk

Audio Audio Visual Visual Marginal
No Feed No Feed

Session 1
Mean 59.86 62.60 61.60 52.66 59.11
S.D. 12.52 13.51 10.50 10.99

Session 2
Mean 38.96 40.29 34.87 30.42 36.10
S.D. 12.57 11.22 13.14 8.57

Session 3
Mean 47.12 49.43 40.03 .44.67 45.38
S.D. 11.99 16.78 11.79 13.27

Marginal 48.65 50.77 45.49 42.59 49.86

Count 12 13 13 14 52

S.D. = Standard Deviation

Table 19

Analysis of Variance for Proportion
of Teacher Talk

Source SS df MS F

Model (A)
Feedback (C)

Session (T)







*p = 0.077
**p < 0.001

Table 20

Tukey's HSD Test for Differences Between
Means Proportion of Teacher Talk

Treatment Cell Mean Cell N

1) Microteaching
Session 1 59.11 46

2) Microteaching
Session 2 36.10 46

3) Microteaching
Session 3 45.38 46

Contrasted Pairs 1-2 1-3 2-3

Difference between
Means 23.01* 13.28* 9.28*

Tukey's HSD: 7.49

*p < 0.05

Student Performance Measure

Analysis of variance for the test scores revealed a

significant difference between teaching sessions and a sig-

nificant interaction between teaching session, model treat-

ment and feedback. Cell means, treatment means and standard

deviations are presented in Table 21. A difference was

also found between model treatment groups which was sig-

nificant at the 0.076 level of probability (F = 3.30). The

audio group scores were generally higher than the video

group. Analysis of variance summary table is reported in

Table 22. The Tukey's HSD Test statistic for differences

between cell means is 5.789. This means that for any two

comparisons between cell means, a difference greater than

5.789 is significant at the 0.05 level of probability.


Means and Standard Deviations for Tests

Audio Audio Video Video Marginal
No Feed No Feed

Session 1
Mean 37.94 41.79 37.14 37.64 38.61
S.D. 7.93 9.91 6.93 7.75

Session 2
Mean 30.97 29.79 25.35 27.93 28.45
S.D. 8.90 7.97 5.01 8.04

Session 3
Mean 33.72 37.16 33.15 28.16 32.94
S.D. 11.75 10.53 7.58 7.60

Marginal 34.20 36.24 31.88 31.24 33.33

Count 12 13 13 14



Source SS df MS F

Model (A) 521.11 1 521.11 3.30
Feedback (C) 18.80 1 18.80 .12
AC 69.26 1 69.26 .44
Error 7580.28 48 157.92

Session (T) 2659.31 2 1329.65 46.89**
TA 35.08 2 17.53 .62
TC 55.94 2 27.97 .99
TAC 242.81 2 121.40 4.28*
Error 2772.32 96 28.36

*p < 0.05
**p < 0.001


Microteaching Sessions

Upon finding significant differences between teaching

sessions and all the outcome variables as well as the paus-

ing variable, wait-time II, the researcher used a step-

wise multiple regression analysis to see whether signifi-

cant correlations and variances were shared by the variables.

This analysis also helped to determine whether the wait-

time variable was behaving at a threshold level or on a

continuous increment level. For each teaching session the

dialogue variables were regressed on the independent var-

iable, wait-time. Test scores were analyzed in a similar

manner but separately in order to examine parital score

results of the three categories as well as the total

test results.

Microteaching Session 1

The means, standard deviations and intercorrelations

for wait-time II and the dialogue variables are presented

in Table 23. Initially both mean length of student talk

and percent of teacher talk had sufficient F's to enter

the regression equation (MLST, F = 15.28; PTT, F = 8.14).

However, once MLST entered the equation, the PTT, F to enter

dropped appreciably due to the high negative correlation

between the two variables (-0.75). Mean length of teacher

talk did not correlate significantly with either wait-time

II or mean length of student talk. For the first teaching

session only the mean lenght of student response shared

a significant amount of variance with wait-time having a

multiple R of 0.49. The amount of variance shared by the

variables accounted for about 24 percent of the total vari-


Table 23

Means, Standard Deviations and
Intercorrelations Session 1


0.64 WTII 1.000

6.53 MLTT 0.007 1.000

4.76 MLST 0.491 -0.178 1.000

57.39 PTT -0.381 0.580 -0.750 1.000

WTII = wait-time II, MLTT = mean length teacher talk,
MLST = mean length student talk, PTT = proportion of teacher

Microteaching Session 2

The means, standard deviations and intercorrelations

for teaching Session 2 are presented in Table 24. Initial

F's-to-enter were sufficiently high for all dialogue vari-

ables to enter into the regression equation. As in Session

1 mean length of student response had the highest F value

of 67.59. Mean length of teacher talk and percent of

teacher talk had F's-to-enter of 6.14 and 9.57 respectively.

The entry order for the variables was MLST, PTT and MLTT.

With MLST entered into the equation alone a significant

amount of variance was accounted for producing a multiple

R of 0.76 (F = 67.59, 1 & 49 df, p <.05). When PTT

entered the equation, multiple R increased to 0.802 (F =

8.52, 1 & 49 df, p < .05). Mean length of teacher talk did

not increase the multiple R significantly.

Table 24

Means, Standard Deviations and
Intercorrelations Session 2


2.29 WTII 1.000

4.73 MLTT 0.334 1.000

9.81 MLST 0.761 0.109 1.000

35.39 PTT -0.404 0.471 -0.702 1.000

WTII = wait-time II, MLTT = mean length teacher talk
MLST = mean length student talk, PTT = proportion of teacher

As in Session 1 mean length of teacher talk showed low

correlation with mean length of student talk. Unlike Ses-

sion 1 the correlations between wait-time II and mean

length of teacher talk were significant. In microteaching

Session 2 the amount of variance shared by MLST and wait-

time II was 58 percent which increased to 64 percent with

the addition of PTT. Mean length of teacher talk did not

increase the amount of shared variance significantly upon

entering the regression equation.

Microteaching Session 3

The means, standard deviations and intercorrelations

for this session are listed in Table 25. Although both

mean length of student talk and percent of teacher talk

had sufficient F's-to-enter into the equation originally,

only MLST yielded a significant multiple R (F = 19.76,

1 & 17 df, p < .01). Neither percent of teacher talk nor

mean length of teacher talk contributed significantly to

the amount of wait-time II variance.

Table 25

Means, Standard Deviations and
Intercorrelations Session 3


2.91 WTII 1.000

6.75 MLTT 0.2562 1.000

9.12 MLST 0.544 0.175 1.000

45.03 PTT -0.328 0.353 -0.726 1.000

WTII = wait-time II, MLTT = mean length teacher talk
MLST = mean length student talk, PTT = proportion teacher

As in Session 1 and 2 the mean length of student talk

had a high negative correlation with the percent of teacher

talk (-.762) and a low correlation with the mean length

of teacher talk (.1752). The multiple regression sum-

mary tables for the three teaching sessions are presented

in Table 26.

Table 26

Multiple Regression Summary Data

Variable Multiple Increase in

Session 1

MLST 0.491 0.241 0.241 15.27*

Session 2

MLST 0.761 0.580 0.580 67.58**

PTT 0.802 0.643 0.063 8.52*

Session 3

MLST 0.544 0.295 0.295 19.75*

*p < 0.05
**p < 0.01

Pupil Performance Tests

In addition to dialogue variables the students com-

pleted a fifteen item multiple choice test at the end of

each session. The quiz contained three types of questions:

observation, inference and vocabulary. Scores on the

test and partial scores of each component were regressed

on the wait-time variable.

In teaching Session 1 no significant multiple R oc-

curred. The F's-to-enter the regression were small, the

highest being observation questions with an F value of 0.1666.

Thus for teaching session the scores of the test did not

correlate to any significant degree with the time teachers


Regression analysis of Session 2 revealed an interest-

ing statistic. The mean length of wait-time two correlated

negatively with inference type questions. The F-to-enter

for inference questions resulted in a significant multiple

R of 0.302 (F = 4.9, 1 & 49 df. p < 0.05). The scores on

observation and inference questionsboth had negative corre-

lations with the mean length of wait-time. The variance

shared by inference questions and wait-time although sig-

nificant was small only accounting for about 9 percent to

the total variance.

For teaching Session 3 significance at .05 level was

not reached for any variable. The intercorrelations for

observation and inference questions and wait-time are low

though no longer negative. Table 27 presents means, stan-

dard deviations and intercorrelations for test scores at

each of the three teaching sessions.

Table 27

Means, Standard Deviations and Inter-
correlations Test Scores


Session 1

0.64 WTII 1.000

14.51 OBS 0.058 1.000

9.51 INF -0.032 0.466 1.000

14.05 VOC -0.001 0.522 0.567 1.000

38.27 TOT -0.006 0.714 0.778 0.854 1.000

Session 2

2.29 WTII 1.000

8.90 OBS -0.244 1.000

8.09 INF -0.301 0.510 1.000

11.41 VOC 0.075 0.126 0.306 1.000

28.56 TOT -0.149 0.684 0.771 0.689 1.000

27 continued


S.D. Variable WTII OBS

Session 3

2.91 WTII 1.000

13.41 OBS 0.137 1.000

10.65 INF 0.062 0.727

8.97 VOC 0.154 0.542

33.14 TOT 0.132 0.908

WIII = wait-time II, OBS = observation
inference questions, VOC = vocabulary
test score.



0.554 1.000

0.794 0.741 1.000

questions, INF =
questions, TOT = total





There are two basic objectives of this study. One is

to devise a reasonably expedient method for training

teachers to increase their wait-time behavior during in-

quiry lessons with children. The other is to examine ac-

companying changes in discourse patterns and pupil perfor-


Four hypotheses were designed to test the first ob-

jective related to training. Another four hypotheses were

formulated to test the effectiveness of increased wait-time

on dialogue patterns. For the sake of continuity, the

results of the experiment related to treatments will be sum-

marized and discussed first. For these hypotheses wait-time

functioned as the dependent variable and treatment groups

as the independent variables.

The second set of hypotheses testing the outcome vari-

ables will then be summarized and discussed. When testing

related outcomes, wait-time served as the independent

variable and dialogue patterns as the dependent variables.

Treatment Variables

In this study treatment occurred at two levels: one

in which teachers observed a model displaying extended wait-

time I and II behaviors and one in which teachers received

direct feedback of their own performance of wait-time be-

haviors. In order to maximize control over the subjects'

entering knowledge of lesson materials, all subjects were

exposed to a set of induction materials for a specified

amount of time. Thus teachers were assumed to have similar

knowledge of the materials on which to base the inquiry

discussions. Subjects were not given any specific format

for asking questions because it was suspected that too much

direction for the discourse would interfere with and con-

found the chances of achieving criterion wait-time.

Wait-time can be thought of collectively as all pauses,

or periods of silence, in teacher-pupil dialogue or it can

be separated into two subspecies, wait-time I and wait-time

II. In the discussion, the effect of treatment on each

specie of wait-time is examined separately.

The first hypothesis tested the change in mean length

of wait-time behavior for teacherswho viewed the video

model. Support for this hypothesis depends upon a sig-

nificant difference between the length of wait-time prior

to treatment and the length of wait-time after exposure to

the model. Analysis of variance indicated that the mean

length of wait time II was significantly greater after

exposure to the video model. Conversely, wait-time I did

not increase significantly after exposure to the video


Hypothesis two tested the difference between the

teachers' mean length of wait-time before exposure to an

audio model and after exposure. The statistical analysis

of wait-time II yielded a very large F value supporting

the hypothesis that exposure to a model would significantly

increase wait-time. As with the video model, wait-time I

of teachers who listened to the audio model did not increase

significantly from entry level to post audio treatment.

Hypothesis three tested the relative effects of the

video and audio models. Significant mean differences be-

tween the two groups supported the hypothesis for wait-time

II. No statistical differences occurred between the two

treatment groups for wait-time I.

Hypothesis four tested the differences in wait-time

between the group who received specific performance feed-

back and the group not receiving feedback. Results of

analysis indicated no significant differences between groups

for either wait-time I or II. However, the effect of feed-

back was confounded by the model treatment variable. Wait-

time II of the video model group was significantly higher

at Session 2 which served as the base level for the feed-

back treatment. A second analysis was done to examine the

average differences in the amount of increase that oc-

curred for the feedback and no feedback groups. The average

increase for the no feedback group, 0.26 seconds, and the

feedback group, 0.92 seconds, differed at the 0.076 level

of probability.

Differences between the feedback and the no feedback

groups for wait-time I were not significant. However, the

overall average of wait-time I did increase significantly

from teaching Session 1 to Session 3. This would appear

to be a delayed, overall, treatment effect rather than a

feedback effect.

For summarization the four hypotheses related to the

main effects will be stated below:

1. The mean length of teacher's wait-time I and II

as measured by servo-chart analysis will significantly

increase upon viewing a video model.

2. The mean length of teacher's wait-time I and II

as measured by servo-chart analysis will significantly

increase upon observing an audio model.

3. The mean length of wait-time I and II of teachers

viewing the video model will be significantly greater

than the mean length of wait-time I and II of teachers

who observe the audio model.

4. The mean length of wait-time I and II of teachers

who receive feedback will be significantly greater than

the mean length of wait-time I and II of teachers not

receiving feedback.

For wait-time II the data analyses supported the

first three hypotheses. Therefore the three hypotheses

were not rejected for wait-time II. For wait-time I,

despite an overall significant increase from pretest to

posttest the three hypotheses were not directly supported

because the mean length of wait-time did not reach the

hypothesized threshold criterion of 2.7 seconds. Hypo-

thesis four was rejected because there was no significant

difference between groups at 0.05 level of significance.

However, three of the four groups (audio feedback; video

no feedback; video feedback) reached the mean wait-time II

level of greater than 2.7 seconds with the video feed-

back group averaging 3.6 seconds.

Interpretation of Treatment

Wait-time II and Modeling

As previously reported in the review of literature

section, modeling has been shown to be an effective train-

ing procedure. Bandura (1965) has reported that the mod-

eled behaviors are most effectively transmitted when the

responding behaviors occur infrequently or are weakly

established. From the initial wait-time mean of 0.688

seconds, with a mean standard deviation of 0.006 seconds,

it is evident that the desired response of 3.0 seconds

is indeed rare. Thus modeling would be expected to have

great utility as a training procedure.

Various modeling formats, video, audio, written, or

protocol, appear to have differential rates of success.

In research using video models and written models, the


video model usually proves to be superior. (Koran, J. J., Jr.,

1970, 1971, 1972; Koran, J. J., Jr., & Koran, M. L., 1974-

1975). Evidence supporting the superiority of the video or

audio model is limited. Santiesteban's research (1974) found

no significant differences between the effects of the two

kinds of models.

Since the task is primarily verbal, without a large

psychomotor component, one might expect the difference be-

tween the two types of models to be negligible. In fact,

Santiesteban argued that the visual model might distract

the subject by presenting a too rich mixture of relevant

and irrelevant cues. If the models were manipulating ob-

jects, as well as discussing the processes, the subject's

attention to the desired behaviors might be redirected from

the verbal behavior to the manipulated activities. For

these reasons, the video tape used to model the pausing

behaviors portrayed only the five discussants with no sup-

plementary visual aids or materials.

In Bandura's theory of the effectiveness of modeling,

the whole modeling process is thought to consist of four

subprocesses, one of which is the attentional process.

Examining this process offers an explanation for why the

video model was superior to the audio model in this study.

It could be maintained that the video model produced

longer, teacher wait-times because the subject was required

to focus his attention in two modes, visual and auditory,

for the video model, but only one for the audio model. On

the other hand, with the audio model the visual senses were

not focused and it is possible that the general, surrounding

environment would produce many visual distractions.

An additional explanation, which might account for

the differences in performance of subjects observing the

audio and video models, is the attributes of the model.

Bronfenbrenner (1970) suggests that these dimensions of the

model have the greatest potential for influencing the model-

ing process. In the video model, the subjects were able to

observe many irrelevant behaviors which could not be detected

in the audio model. Portuges and Feshbach (1972) found in-

cidental behaviors of the model performed nearly as frequently

as essential behaviors.

Certain irrelevant personal characteristics of the

master teacher may have enhanced the value of waiting for

the subjects. For example, she may have looked particularly

relaxed and comfortable in her role. In the video model,

subjects could see that pauses appeared to be a natural part

of discussion. The pupil models could be observed also.

The subjects could see that the children were actively in-

volved in the inquiry process. Silences could be seen as a

dynamic phase of the learning process. The model children

exhibited facial expressions and hand gestures indicating

that something was happening during silence. In other

words, the subjects could "see" that silence was an ac-

tive part of the inquiry process. This may have served

as positive reinforcement for the value of waiting. In

the audio model these visual cues could not be detected.

Wait-time II and Feedback

In general, the feedback group did not differ sig-

nificantly from the no feedback group nor was the overall

increase from Session 2 to Session 3 significant. How-

ever, the results of the average increase, using Session 2

wait-time II means as the control or entry scores for

assessing feedback effect, had an F value approaching sig-


Before discarding the notion that feedback is not

a necessary element of the treatment, further investigations

need to be done. By rejecting the hypothesis at the 0.05

level of significance, the researcher may discard a

potentially valuable component of the overall treat-

ment procedure. Because of the differences noted in

Table 8, additional research may determine that feed-

back is necessary for some types of learners. Examination

of interactions between aptitude and treatment could

yield significant information for determining appropriate

training methods.

Although the differences between groups for the last

teaching session were not technically significant, the

differences were of practical significance, especially

when considering Rowe's hypothesis of a threshold effect

for wait-time. The model exhibited pauses of three seconds

or longer and this criterion was the goal of treatment.

The audio feedback and video feedback groups reached a

mean of 2.7 seconds and the video feedback group surpassed

the 3.0 second level.


Rowe (197j) reported a mean wait-time threshold effect

of about 2.7 seconds, below which no discernible changes

in the outcome variables occurred. If a threshold effect

does exist as suspected, the addition of a few tenths of a

second becomes practically significant. The feedback com-

ponent may be essential to get the subject to increase her

wait-time that half second or so needed to reach or exceed


Wait-time I

Wait-time I is discussed separately from wait-time II

because it did not increase significantly as hypothesized

nor was a mean greater than 2.0 seconds reached for any

group at any session.

By definition wait-time I is the pause space between

the time the teacher asks a question and the time the stu-

dent responds or the teacher asks another question. Al-

though the teacher subjects received training in wait-time

I, as well as wait-time II, she could only control the

length of time she waited before asking or rephrasing another

question. The student could begin a response whenever he

wished. The children participating in the inquiry ses-

sions received no wait-time training nor, in fact, were

they cognizant of wait-time being used.

Analysis of results of the wait-time I data reveals

that wait-time I increased significantly from Session I to

Session 3. These results were not totally discrepant with

expectations. In routine instruction, where the average


wait-time is 0.9 seconds (Rowe, 1974a), students learn they

must respond quickly, often impulsively, if they want a

chance to speak. In micro-teaching sessions where four

children are competing with each other, as well as the

teacher, for that 0.9 seconds, it is surprising that any

wait-time I exists at all. As a matter of fact, the

tapes from the initial micro-teaching sessions were charac-

terized by several student voices frequently chiming in

simultaneously after a teacher question. As teachers in-

creased the length of pauses in Session 2, the children

were given more time to express their ideas. Perhaps they

began to realize it was not necessary to grab the first speak

space in order to have an opportunity to talk. This notion

of delayed effect would seem to be supported by the data in

the form of a significant increase of wait-time I in micro-

teaching Session 3.

One other significant difference was found in the

analysis of wait-time I. An interaction between the model

and feedback treatments occurred. The audio model feedback

group achieved significantly higher wait-time I than either

the audio no feedback group and the video feedback group.

Interpretation of differences between these groups remains

questionable at best, because the differences, although

significant, are small and because no one reached criterion

level. A possible explanation is offered in regard to

pupil performance. Test scores for children in the audio

feedback group were consistently higher than the other three

groups for all teaching sessions. Perhaps the difference

can be attributed to the characteristics of the children,

instead of the treatment, because the children share control

of wait-time I. The children in the audio feedback group

scored higher on a test measuring observation and infer-

ence type skills at the entry level as well as the post

test level. Perhaps these children possessed the skills to

a greater degree than the children in the other groups.

If true, it is possible that when given the opportunity

to pause, these childern would be more likely to be reflec-

tive in nature than the other children. Wait-time I might

have been longer because the children waited longer be-

fore responding particularly if the teachers tried to

extend wait-time I.

Outcome Variables

Dialogue Variables

Three hypotheses were tested to examine changes in

the discourse patterns that accompanied increased wait-

time. Hypothesis five stated that as wait-time increased

the mean length of teacher talk would decrease. Hypothesis

six stated that as wait-time increased, the mean length of

pupil response would increase. Hypothesis seven was form-

ulated to examine the relationship between teacher-pupil

moves. Using the two player model, the pupils were expected

to have at least equal speak-space as the teacher. The

hypothesis stated that the total proportion of teacher

talk would decrease as wait-time increased.

Data results showed the length of teacher talk de-

creased sharply between the initial session and the second

session and wait-time increased between sessions. The

decrease in mean length of teacher talk proved to be only

temporary. For teaching Session 3, MLTT increased to the

entry level while wait-time II continued to increase for

Session 3. Furthermore, multiple regression analysis

revealed a significant correlation between wait-time and

length of teacher talk for Session 2 only. The two vari-

ables did not share a significant amount of variance for

either Session 1 or Session 3.

Examination of wait-time II and its relationship to

mean length of student talk revealed high correlations be-

tween the two variables for all three teaching sessions.

Mean length of student talk increased significantly from

Session 1 to Session 2 and remained high for Session 3.

The correlation between the proportion of teacher talk

and wait-time was high for Sessions 2 and 3. During Ses-

sion 1, the proportion of teacher talk was very high, about

60 percent. The proportion decreased significantly, to

about 35 percent, for Session 2 and increased slightly for

Session 3. A high negative correlation existed for the

portion of teacher talk and the mean length of student talk

for all three sessions.

These results were interpreted collectively because

the contribution of each affects the total changes in

teacher-student dialogue. The dialogue pattern of Session 1

can be characterized in the following manner. The teacher

appears to exert a large amount of control over the dis-

cussion. This is indicated by the large total amount of

time that she occupies the speak space and by the low mean

length of student talk. The lesson was intended to pro-

mote inquiry behavior, which is characterized by children

exploring ideas, observing and making inferences. Yet

each child was given little time to either think or talk

without teacher intervention. The children averaged about

four seconds per speak space or approximately one second

per child. This type of dialogue pattern may indicate the

development of an extrinsic motivation model as Rowe (1974)

suggests. Ogunyami (1972) inferred that dialogue in which

children build on each others ideas cannot occur under

these conditions. It seems unlikely that children Would

be able to develop or explore any idea under such a fast


Micro teaching Session 2 differed from the first

session in that wait-time II almost quadrupled and teacher

talk decreased by one-half while student talk doubled. It

could be inferred that teachers attempted to extend wait-

time pauses by simply talking less. The tapes revealed

that the speak space was more evenly distributed among all

the model players including the teachers making inquiry

less teacher centered. Although not examined quantitatively,

the tapes from Session 2 contained more instances of idea-

changing by students. Verbal contributions by all of the

student players became more equal.

In teaching Session 3, the overall length of wait-

time II increased and the mean length of student talk re-

mained high. However, the mean length of teacher talk

increased almost to the level of teaching Session 1. The

overall proportion of teacher talk increased only about

10 percent. It may be that teachers discovered that they

can both talk and control pauses. The characteristics of

conversation changed again. The teachers talked longer

during any single conversational set but, once yielding to

the students, were less likely to re-intervene into the

flow of conversation. Chart plots from Session 3 show that

pupil dialogue extended over 30 seconds without teacher

interruption in several places. Frequently all four voices

of the children could be heard in the conversational

sequence before the teacher would respond with another ques-


Pupil Performance Test

The last hypothesis tested the idea that as teachers

increased their wait-time, inquiry processes as measured

by written test should improve. Regression analysis showed

that scores on the tests did not share a significant amount

of variance with wait-time II for any teaching session.

Interestingly enough analysis of variance revealed sta-

tistically significant differences between various cell

means, but the differences were scattered over all groups.

A trend seemed to be that pupil test scores were higher

for audio groups than for video groups; this was so across

all testing sessions. The entry scores for the audio

groups were slightly higher, and remained so, negating

the likelihood that scores were related to treatment

group. Pupil test scores tended to be lowest for the

group of teachers having the highest mean wait-time II.

The scores from Session 2 were significantly lower

than the scores of Session 1 or 3. It is possible that

the level of difficulty was greater for the second test.

Upon examination of the "Bouncing Raisin" test the

inference questions appeared to require more background

knowledge than the inference questions on the other two

tests. Taking into consideration the variation in test

scores across groups and the low degree of shared variance

for test scores and wait-time II, hypothesis 8 is rejected.

The lack of correlation between the two variables,

test scores and wait-time, might best be interpreted by

considering the relationship between the variables over

time. Process behaviors, like making careful observations;

and like making inferences that are supported by evidence,

are skills that develop with practice, implying the

passage of time. If teachers regularly incorporate

criterion wait-time into inquiry sessions, the development

of these skills would be fostered because the children

would have an opportunity to use them. In a teacher

training process that occurs over several teaching se-

quences, the teachers behavior is expected to change be-

tween sequences. In this research the training schedule

resulted in large and conspicuous changes in teacher be-

varior from one session to the next.

The general dialogue pattern changed drastically be-

tween Session 1 and Session 2. It could be maintained

that these changes, instead of fostering the development

of inquiry skills, served to confuse the children. For

example, in Session 1 the dialogue pattern required the

children to compete for speak-space and for the teacher's

attention. This initial contact with the teacher would

lead the children to expect a similar pattern for the

second teaching session. Instead the inquiry pattern

changed considerably for Session 2 and then changed again

during Session 3. Inquiry skills like observing and

making reasoned inferences would be expected to develop

with a long wait-time schedule. To produce improvement

in these skills the long wait-time schedule would heed to

be relatively constant. Then changes in student behavior

could be examined as a function of the teacher's behavior

rather than as a function of change in teacher behavior.

Summary of Outcome Variables

The effect of increased wait-time on four outcome

variables was tested. The data supported two of the hy-

potheses. One measured the increase in mean length of

student talk and the other measured the decrease in pro-

portion of teacher talk. Two hypotheses were rejected.

The mean length of teacher talk did not decrease


significantly and consistently as the amount of wait-time

increased. Test scores did not respond to changes in wait-


A question is raised regarding the boundary condi-

tion, threshold wait-time. Rowe (1974) suggested in her

research that a threshold level needed to be reached be-

fore significant changes in the outcome variables occurred.

In this study it appears that for dialogue variables, of

which wait-time is only one, that a threshold effect is

not a necessary condition for change in other dialogue vari-

ables. Altering one variable in the dialogue system ap-

pears to cause proportional changes in the other dialogue

variables in the system. Garigliano's data support this

idea. Although his subjects did not reach criterion wait-

time, a significant increase in the length of student

response accompanied the increased wait-time values. This

supports the notion that wait-time does not have to be held

constant or at a particular criterion for these variables.

It may be that the threshold effect is necessary for

learner outcomes of a different nature than were measured

here. The results of this study cannot offer empirical sup-

port for this notion because it appears that the process

outcome variables results were confounded by the fluctuating

wait-times of the teachers between teaching sessions.

A second question is raised about the unique contribu-

tion of wait-time I on the outcome variables. It appears

that because the pupil subjects' wait-time behavior was not

controlled in the experiment that teachers were not able

to reach the desired criterion for wait-time I. Observa-

tions of the cassette tapes indicated that teachers at-

tempted to extend wait-time I as well as wait-time II but

that the pupil subjects intervened before wait-time I

could be reached. It could be speculated thatif given the

opportunity teachers who reached wait-time II criterion

would also reach criterion for wait-time I. The data,

however, do not contain the necessary information to meet

this assumption leaving the question of unique contribu-

tion for wait-time I unresolved.

Implications for Future Research

The study attempted to assess the effects of four

methods for training teachers to engage in wait-time. The

video model proved to have the most potential as a suit-

able means of training. Before advocating its success,

several questions need to be resolved. The conditions in

which the question of amount and kind of feedback as a

component of training has not been fully answered.

One way to examine the effect of training would be to

explore the differences between subjects who reach cri-

terion and those who do not. Perhaps personal attributes

are interacting with training effects. If subjects who do

not reach criterion are found to have common characteris-

tics, the training model could possibly be altered to

compensate for differences. An aptitude and treatment

interaction approach for matching learner characteristics

to instructional method could tease out valuable relation-

ships. These might determine relative values of various

training procedures for particular groups. A post hoc

analysis of the data is planned to examine some of the

questions raised in this study which could be originating

from individual differences interacting with treatment


Research examining the effects of wait-time on student

outcome variables warrants further investigation. The

data from this study suggest that in order to be able

to examine process outcomes successfully the independent

variables need to be carefully controlled. This is diffi-

cult because very few behaviors have clearly established

and stable criterion levels. By manipulating wait-time in

a variety of ways, a plethora of possibilities opens up for

exploring outcome variables. Investigations of the effect

of extended wait-time on the process skills commonly used

in the process elementary science curricula open many avenues

of research.

This research study provoked many ideas for future

research. More questions were raised than were answered.

Each variable that was investigated could be a potential

source for future research. Several related research

investigations are anticipated which may provide potentially

useful information concerning unresolved questions of treat-

ment and wait-time.






One purpose of this project is to train teachers to
engage in prolonged wait-time. Wait-time refers to the
pausing patterns of teacher-pupil dialogue. The teachers
verbal interaction with children can greatly enhance or
inhibit exploratory or inquiry learning.

A child learns by inquiry when he is presented with
a curious event and is allowed to observe and explore the
phenomenon. Frequently contradictory information will re-
quire further examination of the phenomenon. The explora-
tory phase of learning is often followed by a discussion
period with the teacher. The manner in which the teacher
capitalizes on the children's observations by asking
appropriate questions will affect learning. It is also
important during the discussion phase of inquiry that the
children be allowed time to think about what they have been

Research, however, has shown that teachers generally
don't give children a long enough period of time to think
about what they have done or time to think through their
answers. Pauses, or the length of time teachers are pre-
pared to wait a) after asking a question (WAIT-TIME I) or
b) after receiving an answer (WAIT-TIME II) are usually
less than one second. When teachers increase this "wait-
time" to three or more seconds, children have time to think
about their ideas and their answers are more likely to be
insightful. Discussion frequently becomes less teacher
centered and more student centered, characterized by chil-
dren talking with each other more and sharing and building
on each others ideas.

During this session please carefully observe a tape
demonstrating wait-time I and wait-time II. You should
pay particular attention to the length of each pause. Not
all pauses will reach three seconds criterion and some
pauses will be more than three seconds. As you observe
the tape, please rate the pauses to determine the number
of each kind of wait-time. Each time the teacher asks a
question you will hear a beep if the wait-time I has been
three seconds. Each time a child responds you will hear
two beeps if wait-time II has been reached. Thus you will
know whether you have rated each pause correctly.
*This title was not present on the original materials given to sub-
jects. 88

Diagramatically wait-time I and wait-time II look like

Teacher Student Teacher
asks a WAIT-TIME answers WAIT-TIME or other
question I question II student

At the end of this tape you will lead another inquiry
session with four children. During this session please try
to lengthen your pauses to three seconds. To help you
achieve the three second pause you might try to rough count
for three seconds, 1001, 1002, 1003. Try to reach the
three second pause level for both Wait-time I and wait-time

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.7 - mvs