using videotape modeling to facilitate generalized purchasing skills

Journal of Behavioral Education, Vol. 5, No. L 1995, pp. 29-53

Using Videotape Modeling to Facilitate Generalized Purchasing Skills

Thomas G. Haring, Ph.D., 1 Catherine G. Breen, M.A. , 2'6 Jan Weiner, Ph.D., 3 Craig H. Kennedy, Ph.D., 4 and Florene Bednersh, Ph.D. 5

Six students with moderate and severe disabilities were taught generalized purchasing skills through the use of videotape modeling in one, two, or three stores and in vivo instruction within one store. Training was conducted within three orders of treatments: (a) in vivo instruction followed by videotape training, (b) videotape training followed by in vivo instruction, and (c) concurrent videotape and in vivo instruction. Generalization probes were conducted in stores that were the same as those modeled on videotape, untrained stores that were infrequently probed (and never modeled on tape or taught directly), and novel stores that were probed only once after training. Results indicated the production of generalized purchasing skills by the students who received concurrent training and by the students who received sequential training. Videotape and in vivo training in isolation did not lead to generalized shopping skills. The results are discussed in terms of investigating the possible role of verbalization during videotape training on generalized responding and the effects of multiple probe interventions on inadvertent learning of critical skills.

KEY WORDS: videotape modeling; in vivo instruction; moderate and severe disabilities; purchasing; generalization probes.

In programming for stimulus generalization, one of the most frequently used strategies is sufficient exemplar training where individuals are sequen-

1professor, Department of Education, University of California at Santa Barbara, CA, (deceased). 2Doctoral Student, Department of Education, University of California at Santa Barbara, CA. 3Assistant Professor, Department of Education, California State University, Fullerton, CA. 4Assistant Professor, College of Education, University of Hawaii, Honolulu, Hawaii. 5Director of Special Education, Santa Barbara County Schools, Santa Barbara, CA. 6Correspondence should be directed to Catherine G. Breen, 1304 Camino Rio Verde, Santa Barbara, CA 93111.

29

1053-0819/95/0300-0029507.50/0 �9 1995 Human Sciences Press, Inc.

30 Harin~ Breen, Weiner, Kennedy, and Bednersh

tially or concurrently exposed to environmental conditions that include a range of similar and dissimilar stimuli that cue a stable pattem of responding (Stokes & Baer, 1977). When working with students with severe disabilities, most successful programs include instruction with 2 to 4 exemplars: either trainers, stimuli, or environments (e.g., Breen, Haring, Pitts-Conway, & Gaylord-Ross, 1985). While research strongly supports both the effectiveness and the usability of sufficient exemplar training as a strategy to promote generalization (e.g., Bellamy, Homer, & Inman, 1979; Gaylord-Ross, Haring, Breen, & Pitts-Conway, 1984; Haring & Lovinger, 1989), practical considerations often constrain the ability of teachers to in- struct in multiple contexts, particularly when teaching community-refer- enced skills. Many classroom programs have a low staff-to-student ratio and have limited access to community environments. Training in multiple community environments is clearly a costly undertaking in terms of travel time, travel cost, and opportunity cost. Opportunity cost is the nonacqui- sition of other worthwhile skills due to time and resources being used in- efficiently to obtain an objective.

Programs using simulated instruction within the classroom have been investigated as a means to supplement systematic instruction within a community environment with mixed results. For example, Coon, Vogelsberg, and Williams (1981) taught a student with severe disabilities to ride public transportation using a sequential treatment of simulated classroom instruction followed by instruction in the community. Results of this study indicated that although the student acquired bus boarding and bus departing skills in the classroom, only minimal generalization of these skills to the natural environment occurred. Independent bus riding occurred only after direct community instruction began. The authors speculate that minimal generalization may have occurred because of the difference in the materials used between simulated and in vivo instruction. Other studies, however, have shown that students with severe disabilities do benefit from simulation training (e.g., Nietupski, Welch, & Wacker, 1983; Page, Iwata, & Neef, 1976), particularly when simulation training is adjunctive to training in the natural environment (McDonnell & Homer, 1985; McDonnell, Homer, & Williams, 1984). It is reasonable to speculate that a variable that might be controlling these mixed results is the degree of accuracy of the simulation itself in representing the target stimuli from the natural environment.

Conducting training with videotapes is one means of producing a highly realistic simulation. Videotapes can more closely replicate the conditions found within the criterion environment. Persons represented on videotape can model behaviors that individuals will need to perform com- petently in the natural environment. Videotape modeling has been shown to be an effective tool in promoting the acquisition and generalization of

Videotape Modeling 31

functional skills. In a study by Haring, Kennedy, Adams, and Pitts-Conway (1987), three students with autism acquired and generalized a chain of purchasing responses using systematic instruction combined with videotape models of familiar peers shopping. Charlop and Milstein (1989) used video modeling to teach acquisition and generalization of conversational skills among children with autism. Thus, videotape modeling is emerging as a viable alternative or supplement to direct in vivo instruction to produce generalization.

In the present study, we sought to extend the work of Haring et al. (1987) which tested the effectiveness of videotape modeling in four probed stores following in vivo shopping training in one store. The present study looks at the effectiveness of videotape modeling when presented before, after, and concurrent to in vivo shopping training on performance in trained, probed, and untrained stores. The purposes of this study were to: (a) Assess the acquisition and generalization of independent purchasing responses when using videotape modeling followed by in vivo instruction, in vivo instruction followed by videotape instruction, and concurrent use of the two; and (b) test whether responding produced from videotape simulation and in vivo training could be extended to untrained stores (i.e., stores that had never been modeled on videotape or taught directly).

METHOD

Students

Six students with severe disabilities, ages 10-16, participated in the in- vestigation. The students were selected based on teacher nominations and parent interest. None of the students demonstrated the skills required for independent selection and purchase of single items from stores, and showed repeated failure to generalize learned skills to nontrained contexts. Chip, Ron, and Andy were diagnosed as having autism. Their mean composite score on the Vineland Adaptive Behavior Scales was 51, with mean age equivalents for the Communication subscale equal to 3-5, the Daily Living Skills subscale equal to 4-7, and the Socialization subscale equal to 2-4. Carl, Mary, and Joe had severe mental retardation. Their mean composite score on the Vineland Adaptive Behavior Scales was 35, with mean age equivalents for the Communication subscale equal to 2-1, the Daily Living Skills subscale equal to 2-11, and the Socialization subscale equal to 1-9. Additionally, all students followed simple verbal 2 step directions, and gave one to three word replies to questions such as, "What is she doing?" when presented with pictures. Individual scores on the Vineland ABS and teacher

32 Haring, Breen, Weiner, Kennedy, and Bednersh

made tests are given in Table 1. Four of the students (Mary, Joe, Chip, and Carl) attended integrated public school programs for youth with severe disabilities. Two students, Andy and Ron, attended school at a private resi- dential facility. All of the students spent at least 25% of their school day in community-based instruction.

Settings, Items, and Target Behaviors

The students' teachers and parents identified stores for inclusion in the study on the basis of student needs, proximity to school, and proximity to the students' homes. Types of stores were selected to best sample the range of possible environments in which the purchasing sequence could be used, which typically would be frequented by teenagers, and which presented the range of stimulus conditions which are present within stores which require purchases to be made. No systematic training had previously occurred in any of the selected environments for any of the students. These store types included bookstores, convenience stores, drugstores, gift shops, grocery stores, hobby shops, and record stores. A total of seven specific stores were chosen from the list of store types for each student to be ac- cessed during baseline and intervention phases. Additionally, two to five specific stores were selected to be probed post-intervention only. The list of stores differed for each student based on proximity from home and school. Of the stores selected, one functioned as an in vivo training site where systematic instruction took place, three as videotape generalization

Table 1. Student Assessment Data

Participants (Order of Treatment) a

Carl Chip Ron Mary Joe Andy

(IV, VT) (VT) (VT, IV) (IV + VT)

Age 14 10 16 13 12 12 Disability MR Autism Autism MR MR Autism Vineland ABS 34 50 45 44 28 58

Vineland ABS: Age Equivalents Communication 1-9 3-10 3-4 3-9 1-11 3-2 Daily Living Skills 3-0 3-10 5-0 3-0 2-10 4-11 Socialization 1-2 2-0 1-11 2-1 1-11 3-0

Teacher Made Tests Across Community Environments Mean % Correct 25 40 50 30 20 45

alV = in vivo, and VT = videotape modeling.


training sites where no systematic instruction occurred but that were represented and trained indirectly (as needed) through videotape simulation, three as untrained generalization probe sites where no systematic or videotape instruction occurred, and two to five as novel probe sites in which no systematic or videotape instruction occurred and were probed only one time following the completion of intervention. Note that the number of videotape stores actually employed depended on when in training a student demonstrated generalized selection and purchasing skills across untrained stores (i.e., if fewer than three videotape training stores were needed, the unneeded sites were identical to sites where no instruction took place). Four of the six students received videotape modeling training at their homes. Ron and Andy received videotape modeling training at a university setting.

Three items were selected for training or probing in each store for each participant. The items selected were those things that are typically purchased by same age peers as single items, and are located in various parts of the store. During in vivo training, one item was required to be purchased until the student independently completed 100% of the steps in the task analysis purchasing that item, After the item was independently purchased, a second item was required to be purchased, After the second item was independently purchased, the student was asked to purchase a third item. In probed and novel stores, students were asked to purchase one of three items with the item requested for purchase randomly selected from a list of three items.

A task analysis was developed to be "generic" across a variety of community stores (i.e., the same sequence was sufficient to make purchases successfully in all stores used in the study). Only those responses which were critical for independent participation in community stores were selected for training and assessment. The 12 target responses that were taught during in vivo and videotape training and assessed during probes were: En- tering the store, locating the aisle or section, selecting the item, going to the checkout stand, moving in line, putting the item on the counter, getting out money, giving money to the cashier, getting change, putting away the change, taking the item, and leaving the store.

Materials

During in vivo training sessions and all community probes, students were provided with a sufficient amount of money to make a purchase of a single item. The money was kept in the student's wallet or purse. Videotape training was conducted with the use of VHS videotapes of nondisabled


same-age peers making purchases in the stores to be trained. Videotape sequences were made by the students' instructors with handheld videotape recorders. Sequences were taped by walking several feet behind the nondisabled peer as s/he performed the sequence of responses necessary for selecting and purchasing an item. The videotaped episodes (i.e., one complete selection and purchase of a single item) ranged in duration from 3 to 5 min and showed standard sequences of shopping responses indicated above. The episodes varied slightly from each other according to the variation in stimuli within each store. A total of nine videotape sequences (i.e., selecting and purchasing each of three single items in three stores) were selected for each student to be used as needed during videotape modeling training.

Procedures

Baseline (A)

Baseline probes were conducted within each of the seven targeted stores for each student to assess the student's completion of purchases prior to training; that is, performance measures were conducted in the single in vivo training store, in the three stores in which purchases were modeled on videotape, and in the three stores identified as untrained generalization settings.

A session was begun with the instructor giving a verbal direction out- side of the store to "buy a (name of item)". If a student did not correctly perform a step of the task analysis, the conditions necessary for the next step in the sequence were set up by the instructor to allow for continuation of the chain. For example, if a student did not choose an item within 5 sec, the instructor blocked the student's view of the items, took an item and placed it in the student's hand; if a student did not locate a checkout stand within 5 sec, the instructor gave the student the minimal amount of physical guidance necessary to get the student to the checkout stand again blocking the student's view of environment specific visual cues which might serve as location prompts. The instructor gave no additional assistance or rewards during baseline probes; however, the students were allowed to consume or use the purchased item one half-hour after the probe during a social-leisure breaktime in the community or at school. Baseline probes lasted approximately 10 min.

In l, qvo Training (B)

The same procedures were followed as during baseline with the following exceptions. A system of least prompts was used during in vivo train-


ing. If a response in the task analysis was not initiated within 5 sec of the natural cue (e.g., completion of the previous step, presence of needed item, presence of checkout stand) the student was given a verbal prompt. If the student did not correctly respond within 5 sec of the verbal prompt, a gestural prompt was given. If the student did not then respond within 5 sec of the gestural prompt, partial physical guidance was given until the student completed the step in the task analysis. (Note that partial physical guidance was necessary during only 1% of the training sessions). Independent correct responses occasioned verbal praise from the instructor. Once a criterion of 80% independent performance was attained, the frequency of verbal praise was reduced to every other correct step. Correct responses to prompts were followed by minimal feedback (e.g., "O.K.").

In the in vivo store, a total of three items were trained. Training the student to purchase a first item was conducted until the student met the criterion of independent responding on 100% of the steps in the task analysis. Training of a second item was subsequently conducted to criterion, followed as needed with training on a third item. Note that only two steps in the task analysis: Locate Aisle Or Section, and Select Item, differed across training of single items within the same store. In vivo training continued until the student met mastery criterion on at least one session for item 3 and demonstrated stability in responding within all untrained stores. Stability was defined by 2 consecutive sessions showing a maintenance or diminishing in performance of the responses in the task analysis. The number of sessions varied for all students from two to four sessions per week.

Once the student met mastery criterion across all three items, maintenance conditions were begun. Maintenance sessions followed the same procedures as those used under baseline conditions with the following ex- ception: praise was given directly following the successful independent completion of the task analysis.

~deotape Training (C)

Table 2 includes a set of questions that was developed for videotape training corresponding to the responses made on the purchasing task analysis. A videotape training session consisted of directing a student to: (a) Watch one videotape sequence of purchasing responses without interrup- tion; and (b) watch the sequence again, this time answering a series of questions regarding the videotape. One videotape training session was conducted two to four times per week (one session only per day) and after every third videotape instructional session a generalization probe was conducted in the same store. During step (b) of videotape training, the tape

36 Hating, Breen, Weiner, Kennedy, and Bednersh

Table 2. Questions for Videotape Training

Videotape Segment Question

Student enters the store. Where is she going?

Student finds the correct aisle or section. What is she doing?

Student selects item. What is she getting?

Student goes to the checkout line. Where did she go?

Student moves in line. Where is she standing?

Student puts item on counter. Where did she put item?

Student gets out money. What is she getting?

Student gives money to cashier. What did she do with money?

Student takes change. What did she get?

Student puts money away. What did she do?

Student picks up item. What did she take?

Student leaves the store. Where did she go?

was "paused" directly following the modeling of each response in the task analysis and the corresponding question was asked. If a correct response was given within 5 sec, verbal praise was given and the tape was restarted. If no response or an incorrect response was given within 5 sec, the correct answer was modeled by the instructor, the student was required to repeat the correct response, and the tape was re-started. During generalization probes of videotape training, the student was taken to the same store and asked to select and purchase one item. The procedures replicated those used under baseline conditions where no prompts, feedback, or rewards were given, other than the instructor input needed to allow for the assessment of the next step in the task analysis. As with baseline conditions, the student was allowed to consume or interact with the purchased item during a natural time at school.

Videotape training was conducted sequentially with tapes from three videotape generalization stores. Videotape generalization training was conducted with each taped sequence until a training criterion of 100% correct on the questions for one session was reached and when stability in the probe stores had been achieved, or until the student reached 100% accuracy for purchasing the item in the actual store. Once the criterion for one store had been met, training began with the videotape sequences from the second store. Videotape instruction continued as needed (in the second and third stores, respectively) until the training criterion was met across


all videotape sequences, or until generalized responding within all community videotape and untrained generalizations probes was demonstrated.

Generalization Probes

Probes were conducted to assess the degree of generalized responding to untrained sites during the various phases of the study. Generalization probes were identical to baseline sessions. No prompts, feedback, or praise were given to the students during these probes for the duration of the study.

The following types of generalization probes were conducted: 1. Videotape probes were conducted within the community stores mod-

eled through videotape to assess generalization from videotape models to performance in modeled stores. Videotape probes were conducted directly following every third videotape instructional session.

2. Untrained probes were taken in stores that were not modeled through videotape or taught through in vivo instruction to probe for generalized responding across stores. Untrained probes were conducted an av- erage of one time per week and occurred prior to in vivo or videotape instruction on that day.

3. Novel probes were taken to further assess the extent of generative responding to stores and items that had never been probed. Novel probes were conducted after the student demonstrated competence within all of the preselected stores and untrained probe stores. Novel probes were conducted in 2-5 additional novel stores to ensure that the learned skills were fully generalizable and not an inadvertent by-product of probing in the same stores.

Interobserver Agreement

Interobserver agreement sessions were conducted every fifth instructional or probe session within community sites and during every fifth videotraining session throughout the duration of the study. The trainer and the observer were positioned to observe and record students' performance independently. The point-by-point reliability formula (Kazdin, 1982) was used to calculate the percentage of agreements on occurrence and nonoc- currence. Within community sites there was a mean agreement of 93% with a range of 82% to 100%. There was 100% agreement during all videotape training sessions.


Design

A multiple probe across settings design was employed. For Chip and Carl, after the demonstration of stable baseline responding (A), in vivo training in a grocery store (B) was staggered across students. After the training criterion was met in the in vivo store across three items, videotape training was conducted for Chip using models of three items from the Hobby Shop, followed by models of three items from the Record Store (C1, C2). Carl was provided videotape training using models of three items from one store only, Record Store (C1). Thus, Chip and Carl were exposed to the experimental phases in the following order: A-B-C1-C2 for Chip and A-B-C1 for Carl.

To control for possible order effects and to assess the educational fea- sibility of using videotape instruction alone, Mary was exposed to the experimental treatment in reverse order: A-C1-C2-C3-B, with baseline followed by videotape training using models of three items from the Gro- cery Store, then the Record Store, then the Book Store, followed by in vivo training in the Hobby Store. Ron was exposed to a partial sequence of the same type: A-C1-C2, with baseline followed by videotape training using models of three items in the Book Store, then the Gift Store.

For the last two students (Andy and Joe), a c o n c u r r e n t training package was implemented. Instruction for Andy and Joe included videotape training in one store (Book Store and Record store, respectively) and in vivo training in another store (Hobby store and Drugstore, respectively). Each day, they received videotape instruction, intermittently followed by a probe in the same store, and in vivo instruction in another store. For Andy, once criterion was met for all three items from the first videotape store, videotape training from another store (Grocery store) was begun while con- tinuing to train in the in vivo store. For Joe, no additional training was conducted as he met criterion of independent performance in all stores. Andy and Joe were exposed to the experimental phases in the order: A-CI+B-C2+B and A-CI+B, respectively.

RESULTS

Percent Correct and Cumulative Independent Purchases

The acquisition and generalization data across stores are displayed for each of the six participants in Figures 1-6. The data are displayed per session as the percent of responses correctly demonstrated (left hand axis, closed circles) and the cumulative number of independent purchases made


(right hand axis, open squares). Because the goal of training was to teach independent item selection and purchasing across multiple stores, the data are analyzed to reflect both the ongoing learning of individual responses and the demonstration of independent performance. The data also reflect two types of instructional sessions, training sessions presented within a multiple baseline format are represented in the figures for all but Joe and Andy in the top series of panels, and generalization sessions are represented in the single bottom panel of each figure. For Joe and Andy, training and generalization data are presented in two figures, one showing performance in training stores across students and settings, and one showing generalization data, that is, performance in untrained stores.

The data for Carl and Chip are given in Figures 1 and 2, respectively. For Carl and Chip, no independent purchases were made during baseline conditions in any of the stores, with stable baseline performances demonstrated within each store (M = 35.5% and 31.5%, respectively). During the first intervention phase (IV training, see top panels), Carl made 11 and Chip made 9 independent purchases in the in vivo training stores, but no independent purchases in the 6 untrained stores, although increases were seen from baseline in the acquisition of individual responses for both students (Carl: M = 62.5%, Chip: M = 35.7). During the second intervention phase (VT 1, 2nd panels), Carl and Chip continued to make purchases in the in vivo store, and made 4 and 5 (respectively) independent purchases in the corresponding videotape store (M = 92.8%, 92.4%). In addition, Carl made independent purchases in each of the untrained stores (M = 88.8%), in 5 novel stores (M = 100%), and in each of the stores once training had terminated (M = 100%). During VT 1, Chip made independent purchases in only 3 of 5 untrained stores (M = 76.9% across 5 untrained stores); therefore, training was begun in a second VT environment (record store, 3rd panel). During VT 2 training, Chip made purchases in all of the trained and untrained stores (M = 100%), in 5 novel stores (M = 100%), and in a sample of stores probed following the termination of training (M = 100%).

The data for Ron and Mary are presented in Figures 3 and 4, respectively. For both participants no independent purchases were made during baseline conditions with stable performance across all stores prior to intervention (M = 54.1% and 14.8%, respectively). During the first intervention phase (VT 1, see top panels), Ron made 1 and Mary made no independent purchases in the training store, with both showing increases from baseline in the training store to 82.6% and 59.5%, respectively. In addition, Ron independently purchased an item in 1 of the untrained stores (M = 77.8%) while Mary made no independent purchases in any of the untrained stores (M = 39.4%). During VT 2 (second panels), Ron made


O ~ f o r -

Carl

3 ~ s

~. 60 M a l n t e n i n c e

~ 3 . u

0 L loo

. 3

w11Io

N ~ O =

~ 0

L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D i , V ideo tape T r a i n i n g M a i n .

0 m-----O i ~J n ~ x ' ~ l ~ X Z - - G X ~ ' - - ! �9 . - j �9 ,

2 0 4 0 6 0 8 0 1 0 0

S e s s i o n s

Carl

B . L . ; In V iva T r s l n l n g

�9 �9 e

I w i

�9 �9

D �9 i

i

�9 : l i r a m ~ o t] n o ~

, i

2 O 4 O

�9 �9 m m m m o o a ~ - 2 o �9 .

sOo 0~ " no,~ O ~ �9 1 5

�9 ~0 V ldeo tapa Qd'nove I T r a i n i n g

novel ~ I M I I n -

o n DO, n n r r m ~ . [ ] , . ~ . a~, mDOU

6 0 8 0 1 0 0 1 2 0

t - i n

2O m , <

l S Z

O" ~ s " !

0 0

1 2 0

D .

"O

Q .

,..I

"13 C

O

0~

S e s s i o n s

Fig. 1. Percent of task analysis steps with correct responding (closed circles) and cumulative number of independent purchases (open squares) in trained stores (top two panels) and untrained and novel stores (bottom panel) for Carl. The boxes that encompass open squares in the lower panel identify assessments conducted in novel stores.


C h i p

I0O,

6 0 .

. ~ IO"

2

100-

"1o o

cc 4 0 -

~ . +. . .

0 I I0-

- ~ +.+.

~ . ~ -~ o

+~,~ ,----%~ , ,

i L . . . . . . . . . . . . . . . "1

i i ~ ,r, it-..Q

i

: V I deo t i pe M a i n t e n a n c e I T ra in ing i i I I

I i w - , | t i i

i i i

i i i i

2 0 4 0 5 0 8 0

Sess ions

i

B.L. I I In V lvo T+' i ln lng

i+-+i " n ~ i

2 o

+

i VT 1 � 9 i i

l

t i i

�9 i i i [] i i o

a I]: a a a :

n a + , |

4 0 6 0

Sess ions

m �9

M l l n t e n i n c e

1 0 0

I I

I lO 1 0 0

" 2 0

" ! 5

" t O

- 5

~ 0 t,-"

El t,- m

'13,1 I '41' m l

-1 ! I

(0 - 1 0 Z

i-" - s ::::1

O" o (1)

" ' !

O ,,,,4k

Q . r

Q . (1)

I,,,41,

"O C: ,,.m I

O

U) <1) O)

' 2 0

5

I0

5

0

1 2 0

"2O

5

0

S

0

1 2 0

Fig. 2. Percent of task analysis steps with correct responding (closed circles) and cumulative number of independent purchases (open squares) in trained stores (top three panels) and untrained and novel stores (bottom panel) for Chip. The boxes that encompass open squares in the lower panel identify assessments conducted in novel stores.


Ron 100

P

| m . ~ , f -

Q) m 0,_ ~ Ioo

m

Ron

8 O

o ~ ~

~ 0

t

: V i d e o t a p e

) Training

* i

i , i l . i �9 M i i i

2O

M a i n t e n a n c e

;3 I I �9

- |

4 0 6 0

Sessions

| J

8 0 1 0 0

B.L. : i

�9 ! a �9 e o �9

ooO: �9 �9 o

o

i i 2O l e o

i

" " : . ' : -i. O e :

t

VT 2 ~ M a i n t e n a n c e i

13 ~ navel ."r--

GI: rim

4 0 6 0 8'0

Sessions

"20

" l S

"10

- 5

0

"20

"15

"10

- S

r

1 2 0

"2"J

"15

"10

- 5

0

1 2 0

O e-

3 r" m

m l

<

z r

3 O"

O

m

en r

"O

r r

r

O = r

m

m

Fig. 3. Percent of task analysis steps with correct responding (closed circles) and cumulative number of independent purchases (open squares) in trained stores (top two panels) and untrained and novel stores (bottom panel) for Ron. The boxes that encompass open squares in the lower panel identify assessments conducted in novel stores.

~o

=~

.~" ~'

~

Per

cent

C

orre

ct

~ =~

=~

(Gen

eral

izat

ion)

r'~

~

. U

ntr

ain

ed

end

N

ove

l P

rob

es

in

Vlv

o:

Ho

bb

y S

tore

Y

't"

3:

Bo

ok

Sto

rm

~ "

~ �9

7

''-

~ ..

....

....

....

....

..

~ ..

..

..

..

..

..

..

.

~ ..

..

"f

....

....

....

...

- ...

..

..

..

..

..

..

..

~,~.

~ :

~[

g

~+-.

'-

-+"

a"!

:.

' "

: -

+

~'~g

Per

cent

Cor

rect

(T

rain

ing)

VT

2:

R

eco

rd

Sto

re

VT

1:

G

roce

ry

Sto

re

. X

, ,

, t

+ .

=~

? ..

....

.. 1

..

..

..

..

i lii

iii ililf

sase

qoJn

d lu

apu

adap

ul

to J

aqm

nN a

^t|e

lnm

no

.<

o Q.


3 independent purchases in the second training store (M = 84.4%) and 5 purchases in the untrained stores (M = 92.0%), while Mary made no independent purchases from the trained or untrained stores (M = 76.2% in second training store, M = 61.0% in untrained stores). Due to moving away from the school district, Ron was unable to receive further training. Post- intervention probes for Ron in 3 novel stores yielded no independent purchases, although a mean performance of 83.3% correct. Mary's training continued using a third videotape store (third panel, book store). During VT3, Mary made 2 independent purchases in the book store (M = 87.7%), no independent purchases in either of the previous training stores (M = 89.3%), and no independent purchases in any of the untrained stores (M = 79.3%). A final phase of training in vivo was introduced (fourth panel, hobby store), yielding independent purchases in the in vivo training store (M = 83.9%), the 3 previous videotape training stores (M = 98.2%) and in 2 of 3 remaining untrained stores (M = 89.2%). Following training, independent purchases were made in 3 novel stores (M = 100%), and in probes of untrained and previously trained stores (M = 100%).

The data for Joe and Andy are displayed in Figures 5 and 6, respectively. To demonstrate functional control within this dyad, the data are displayed as a multiple baseline across participants as well as settings (for Andy). Figure 5 presents the training data for the dyad and Figure 6 dis- plays the generalization data for untrained and novel stores. As with the other participants no independent purchases occurred during baseline for either Joe or Andy, with stable baseline performances prior to intervention of M = 62.5% and 43.3%, respectively. During the first intervention phase (concurrent training of IV + VT 1, panels 1, 2, 3, 4), Joe made 6 independent purchases in the in vivo store (M = 87.9%), 6 purchases in the videotape training store (M = 93%), independent purchases in each of the untrained stores (M = 81.4%) and in 2 novel stores (M = 100%). Main- tenance probes in trained and untrained stores yielded independent purchases (M = 100%). In comparison, Andy made 6 independent purchases in the in vivo store (M = 87.3%), 3 independent purchases in the first videotape training store (M = 88.3%), but no independent purchases in the remaining untrained stores (M = 82.2%). Concurrent training continued for Andy with instruction in the original in vivo training store (drug store) and a second videotape training store (fifth panel, grocery store). During this phase, independent purchases were made in the second videotape training store (M = 95.6%), both of the 2 previous training stores (M = 95%), the remaining 4 untrained stores (M = 97.1%), and 3 novel stores (M = 100%). Maintenance probes in 6 trained and untrained stores also yielded 100% correct performance.

Videotape M o d e l i n g 45

Joe

�9 6 0 "

~ 6 0 .

2 0 '

O

..'.i ~ / " : "" i Ma in tenance

. - . :

e

i

,, e / i ,

20

15

10

"5

O

11:0 1

j60. �9 1~ 6 0 ,

8 gc60,

~ 2 0 '

V ideo tape T ra i n i ng

i

O e -

r

- ; . . . . . . . . ,,

V ideo tape T ra i n i ng

100 -

�9 80 0

!" 40 �84

~60

0

l eg ~

60 ,

40 .

~2o.

0

Ma ln t *nance

e

i

i t i

u

e e

i

- . , . ,

20 40 60

Sess i ons

Ma in tenance

0

80 100

20

1$

10

"S

0

2O

lS

10

"5

0

20

" lS

' 10

$

0

20

lS

10

"S

0

120

O r

3 t - i

< (I)

z e - 3 r i !

O

en tD "O (I)

r

"-I

"O l -

O :3"

t~

Fig. 5. Percent of task analysis steps with correct responding (closed circles) and cumulative number of independent purchases (open squares) in trained stores for Joe (top two panels) and Andy (bottom two panels).


U . _ N

J o e 10o

J= o n 8O

>- = o .

i

0

: l l � 9 ~ l l m

i gl : �9 �9 �9 novel �9 ! . �9 " " . . = ==

In Vlvo + Videotape Training ',i a a a Malntonan�9

~o

, # i i i i

~ . . . . ~

Andy i

em 100 .O I V § V T 1 O r, 80

o 60 Z

40

20

C 0

. . . ; . O O

oeedm emem i �9 Mslntena ncll

IV + V T 2 ~ novel #

D #

D

j[tl novel

Q.

a i [ ] n n D o [~ m D =

s �9

Sessions

- 2 O

O e -

- 1 5 3 e- r a

- 1 0 <

Z S r -

3 ~ r

0.-,

5"

"(3 - 2 0 r

O .

- 1 5 ~

r - " 1 r ~

0

1 2 0

Fig. 6. Percent of task analysis steps with correct responding (closed circles) and cumulative number of independent purchases (open squares) in untrained and novel stores for Joe (top panel) and Andy (bottom panel). The boxes that encompass open squares identify assessments conducted in novel stores.

In summary, Chip and Andy received training in one in vivo store and two videotape training stores prior to demonstrating independent performance across all trained, untrained, and novel stores, with Chip's training being conducted sequentially (that is, training in one context was followed by training in a second context), and Andy's training being conducted concurrently (i.e, training in two contexts conducted simultaneously). Carl and Joe, who also received sequential and concurrent instruction, respectively, were taught in one in vivo store and one videotape training store prior to generalization of independent performance. Finally, Mary, who also received sequential instruction, with videotape training preceding in rive instruction, demonstrated generalized independent performance following instruction in three videotape training stores and one in vivo store.

The data show for each of the participants an increase in the percent of task analytic responses correctly performed from one intervention to the next. The increase in responding in stores where training is not present


may have been due to the presence of multiple responses common to each of the stores, allowing for multiple exemplar training of individual responses, or to the presence of multiple probes and therefore multiple opportunities to practice in each of the untrained stores. These two issues will be addressed further in the subsequent analysis of participant responding on steps in the task analysis and as points of discussion.

Analysis of Task Analysis Steps

Table 3 presents the mean percent correct responding per step in the task analysis for each of the dyads during each experimental condition. The table is included to show the generalization from the training store(s) to untrained stores on each response in the task analysis. The data show that for Chip and Carl during baseline conditions two steps in the task analysis (Pick Up Item, and Leave Store) were performed correctly during more than 50% of the probe sessions. During the first intervention phase, increases in correct performance in untrained stores were seen for all responses except three: Find Section, Select Item, and Move in Line. During the second intervention phase (VT 1), the response Move in Line was performed correctly at a substantially higher frequency than was seen during in vivo instruction; however, the responses which determined finding and selecting items remained low in accuracy when compared to the performance overall. During the third intervention phase, VT 2, performance across all responses reached the criterion level of 100% correct responding within untrained stores.

The data for Ron and Mary indicate that during baseline two responses -Pu t Item on Counter and Pick Up Item--were demonstrated more than 50% of the time. During the first intervention phase (VT 1), 4 responses continued to be seen during fewer than 50% of the probes, 5 responses occurred during between 50% and 60% of the probes, and the remaining responses were seen at or less than 75% of the time. During the second intervention phase (VT 2), the rate of correct responding increased for each step in the task analysis; however, 5 responses were demonstrated less than 75% of the time. During the third intervention phase (VT 3), 4 of 12 steps reached criterion level, while 3 steps in the task analysis continued to be inaccurate: Find Section, Move in Line, and Put Item on Counter. During the final intervention phase (In Vivo), 7 of 12 steps were emitted during 100% of the probe sessions, while one response, Move in Line, remained at an occurrence level of 60% of the probes.

The data for Andy and Joe indicate that during baseline one response in the task analysis (Take Change) was performed independently during a

Tab

le 3

. M

ean

Per

cent

Cor

rect

Un

trai

ned

Res

pons

es b

y S

tep

in t

he T

ask

Ana

lysi

s

Tas

k A

naly

sis

Ste

p

Car

l/C

hip

Ron

/Mar

y Jo

e/A

ndy

BL

a IV

b V

T1 c

V

'I2

BL

V

T1

VT

2 V

T3

IV

BL

IV

+V

T1

IV

+V

T2

1. E

nter

sto

re

23

81

2. F

ind

sect

ion

9 22

3.

Sel

ect

item

32

26

4.

Fin

d ch

ecko

ut

22

71

5. M

ove

in l

ine

9 23

6.

Put

ite

m o

n co

unte

r 27

70

7.

Get

out

mon

ey

4 52

8.

Giv

e m

oney

to

cash

ier

41

74

9. T

ake

chan

ge

45

85

10.

Put

mon

ey a

way

50

74

11

. Pi

ck u

p it

em

68

85

12.

Lea

ve s

tore

64

78

aBas

elin

e, b

ln v

ivo.

CV

ideo

tape

mod

elin

g.

100

100

40

69

95

100

100

62

98

100

59

100

23

41

50

27

87

15

63

88

62

100

40

75

86

91

87

56

98

100

92

100

17

59

82

73

100

28

86

94

86

100

23

34

73

27

60

13

84

88

97

100

73

59

73

45

80

33

91

100

95

100

33

41

59

91

100

44

100

100

95

100

27

34

68

73

100

46

100

100

97

100

43

75

86

82

87

82

100

100

100

100

30

59

86

100

100

21

53

94

97

100

73

56

91

100

100

51

98

100

97

100

40

59

82

100

100

33

98

100

g~

r~

gr


greater percentage of probes than were the other responses, although 5 additional responses were correctly performed during approximately half of the probes in untrained stores (Enter Store, Select Item, Get Out Money, Give Money to Cashier, and Pick Up Item). During the first intervention phase (IV + VT 1), 10 of 12 responses increased in accuracy, but most were below criterion levels. Those responses that were lower during this phase were: Find Section and Put Money Away. During the second phase of intervention (IV + VT 2), all of the responses reached criterion levels, with 8 of 12 responses occurring during 100% of the probes in untrained stores.

DISCUSSION

The data replicate and extend the findings of Haring et al. (1987) by demonstrating the effectiveness of direct community instruction combined with videotape modeling in promoting generalization of shopping skills. The data demonstrate that in vivo followed by videotape instruction, videotape followed by in vivo instruction, and concurrent instruction can be successful in facilitating the acquisition and generalization of individual and chained responses.

The results indicate that a greater frequency of independent purchases was observed for all participants across multiple stores following in vivo training in one store and videotape training (sequentially or concurrently conducted) in 1 to 3 additional stores. In addition, all participants demonstrated improved performance in stores which were trained using videotape modeling. The results may have important implications for instructional programming for learners with severe disabilities. The data suggest the use of videotape modeling as a supplement to systematic instruction both in terms of teaching novel behaviors and promoting generalization of learned skills.

The results suggest that those task analysis responses which vary across sessions and across stores may require the greatest number of training trials and training exemplars to be acquired and maintained by the students (see Table 3). The responses Find section and Select item required multiple in vivo and videotape training sessions by all 6 students prior to acquisition of the responses. These responses may have proven to be the most resistant to acquisition and generalization because of the changing nature of the stimuli which serve to cue the response across training sessions and environments. That is, students were often faced with the difficult task of locating a variety of items across sessions and stores requiring them to demonstrate a complex search strategy. In contrast, many other responses


within the task analysis were static or more genetic across stores and were more readily acquired and generalized (e.g., Enter store, Put item on counter, Take change, Pick up item), requiting fewer training trials or exemplars. Using videotape simulation may be a means for training the more difficult responses, allowing for modeling of multiple item exemplars within multiple community sites.

While these data indicate support for videotape simulation as a supplement to direct community training, the results should be interpreted with caution. One weakness with the current analysis is that there are two participants per order of treatment, allowing only conservative conclusions regarding the comparative strength of each procedure. Because of the smaller number of participants per treatment, it is premature to make firm conclusions regarding the effectiveness of videotape training, in vivo training, or a combination of the two. What is clear from these data is that training must occur in multiple settings, with what intervention or what sequence of interventions is not as clear. Additionally, conclusions regarding the in- fluences of disabling condition or level of participation on performance in vivo and using videotape models are not clear. Repeated replications are needed to form generalizable conclusions regarding the efficiency or effectiveness of one treatment order in contrast to another, the effects of one treatment in isolation, or the differential effects of disability or ability level on performance using simulated models.

A second concern is the use of an experimental design which relied on repeated probes for generalization. Multiple probe designs are used to, (a) evaluate the effectiveness of training or intervention on the generalization of skills across contexts, and (b) to identify specifically those points during intervention which most substantially predict the emergence of generalized skill acquisition or behavioral control (Horne r& Baer, 1978). Therefore, conducting repeated probes within contexts appears to be a necessary component for assessing generalization.

However, by placing individuals within contexts that are "not to be trained", but are to be used only as opportunities to measure generalization of behaviors, we may be allowing students to acquire skills inadvertently through increasing familiarity or inadvertent instruction from others in the probe store (e.g., through modeling or direct feedback). Horner, Williams, and Knobbe (1985) found increased maintenance of behavior when there were opportunities for performance than when there were not. Increased performance may also occur for untrained behavior given the presence of opportunities to perform the behavior. The procedures utilized in the current method heighten this problem. To obtain a complete assessment of the student's responding at each step in the chain, we required that the student be given an opportunity through trainer manipulation to perform


each response in the task analysis. These procedural modifications may inadvertently provide the student with additional cues and prompts regarding the specific critical behaviors, thereby increasing response accuracy that cannot be attributed to generalization processes. To ensure that this problem of learning via probing resulted in generative skill acquisition beyond the training and probes in stores, we assessed, post criterion, the responding of each participant within novel stores in which no probes had previously occurred. For all students who had received in vivo and videotape instruction, probes in novel stores yielded independent purchases being made in each store, lending support for responding as a result of generalized application of skills. The contribution to learning by experiencing multiple probes is an issue affecting a great deal of operant generalization research which should be addressed in future research.

There has been little procedural research concerning the use of videotape simulation. In future research, it may be important to investigate the possible functional role of the verbalization of the student watching the videotape during training on controlling generalized responding in actual stores. Videotape training can be seen as an example of say/do correspondence training. That is, verbalization of responses within videotape training may be serving to mediate performance in criterion stores. As such, reinforcement within the in vivo store (which appears for most students to be a necessary component in the production of generalized responding) may operate as the controlling variable, because of the reinforcement of the say/do correspondence (e.g., Baer, Detrich, & Weninger, 1988; Ward & Stare, 1990). If so, the effectiveness of videotape modeling might be viewed best as an example of rule-governed behavior. Within the present research, the participants were not taught to overtly state rules to govern their own behavior. They did, however, label the actions that they saw. Interestingly, Wacker et al. (1988) have also demonstrated that labeling-then-doing ef- fectively contributed to generalized responding. If the rule governance of behavior controls responding, then future application of this procedure might make the verbal statements during videotape training more explicitly rule-like. That is, instead of asking questions in the form of identifying what the model is doing (e.g., prompting a labeling response), questions would be asked in the form of what the subject herself will do in that environment (prompting a more rule-like statement). Finally, if students do not benefit from videotape modeling, prior training with simpler forms of say/do correspondences (e.g., where responses are performed in less dis- tant settings and where responses are less complex) might extend the bene- fits of videotape modeling to students with fewer skills (see, for example, the analysis by Stokes, Osnes, & Guevremont, 1987).


An alternative hypothesis is that the role of verbalization during videotape instruction serves to direct a student's attention to the model. If attention to the videotape model alone (without verbalizing rules) is sufficient to produce generalized responding, then the controlling variable is most probably best explained by modeling alone. Thus, future research that com- pares procedures that direct attention to the videoscreen (e.g., reinforcing eye contact, or reinforcing responses to verbal prompts to touch stimuli on the screen) should be compared to procedures where the subject verbalizes the actions performed by the model, or verbalizes what he will do in that setting.

Students with severe mental retardation and autism can learn complex forms of generalization through modeling (Goldstein & Mousetis, 1989). Promoting generalization through videotape modeling is a potential means for achieving independent living objectives and social skill objectives in an efficient manner. Haring (1989) has captured the importance of training for generalization well, "Generalization is the capstone of instruction. With- out it, we fail as teachers. Achieve it, and increased independence and in- tegration are the true results" (p. 177).

REFERENCES

Baer, R. A., Detrich, R., & Weninger, J. M. (1988). On the functional role of the verbalization in correspondence training procedures. Journal of Applied Behavior Analysis, 21, 345-356.

Bellamy, G. T., Horner, R. H., & Inman, D. P. (1979). Vocational Habilitation o f Severely Retarded Adults. Baltimore: University Park Press.

Breen, C., Haring, T. G., Pitts-Conway, V., & Gaylord-Ross, R. (1985). The training and generalization of social interaction during breaktime at two job sites in the natural environment. Journal of The Association for Persons with Severe Handicaps, 10, 41-50.

Charlop, M. J., & Milstein, J. P. (1989). Teaching autistic children conversational speech using video modeling. Journal of Applied Behavior Analysis, 22, 275-286.

Coon, M. E., Vogelsberg, R. T., & Williams, W. (1981). Effects of classroom transportation instruction on generalization to the natural environment. Journal of The Association for the Severely Handicapped, 6, 46-53.

Gaylord-Ross, R. J., Haring, T. G., Breen, C., & Pitts-Conway, V. (1984). The training and generalization of social interaction skills with autistic youth. Journal of Applied Behavior Analysis, 7, 229-247.

Goldstein, H., & Mousetis, L. (1989). Generalized language learning by children with severe mental retardation: Effects of peers' expressive modeling. Journal of Applied Behavior Analysis, 22, 245-260.

Haring, N. G. (Ed.) (1989). Generalization for students with severe handicaps: Strategies and solutions. Seattle: University of Washington Press.

Haring, T. G., Kennedy, C. H., Adams, M., & Pitts-Conway (1987). Teaching generalization of purchasing skills across community settings to autistic youth using videotape modeling. Journal of Applied Behavior Analysis, 20, 89-96.

Haring, T. G., & Lovinger, L. (1989). Promoting social interaction through teaching generalized play initiation responses to preschool children with autism. Journal of The Association for Persons with Severe Handicaps, 14, 58-67.


Homer, R. D. & Baer, D. M. (1978). Multiple-probe technique: A variation of the multiple baseline. Journal of Applied Behavior Analysis, 11, 189-196.

Homer, R.H., Williams, J.A., & Knobbe, C.A. (1985). The effect of "opportunity to perform" on the maintenance of skills learned by high school students with severe handicaps. Journal of the Association for Persons with Severe Handicaps, 10, 172-175.

Kazdin, A. E. (1982). Single-case research designs: Methods for clinical and applied settings. New York: Oxford University Press.

McDonnell, J. J., & Homer, R. H. (1985). Effects of in vivo versus simulation plus in vivo training on the acquisition and generalization of grocery item search strategy by high school students with severe handicaps. Analysis and Intervention in Developmental Disabilities, 5, 323-344.

McDonnell, J. J., Homer, R. H., & Williams, J. A. (1984). Comparisons of three strategies for teaching generalized grocery purchasing to high school student with severe handicaps. The Journal of The Association for Persons with Severe Handicaps, 9, 123-133.

Nietupski, J., Welch, J., & Wacker, D. (1983). Acquisition, maintenance, and transfer of grocery item purchasing skills by moderately and severely handicapped students. Education and Training of the Mentally Retarded, 18, 279-286.

Page, T. L., Iwata, B. A., & Neef, N. A. (1976). Teaching pedestrian skills to retarded persons: Generalization from classroom to the natural environment. Journal of Applied Behavior Analysis, 9, 433-444

Stokes, T. F., & Baer, D. M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10, 349-367.

Stokes, T. F., Osnes, P. G., & Guevremont, D. C. (1987). Saying and doing: A commentary on contingency-space analysis. Journal of Applied Behavior Analysis, 20, 161-164.

Wacker, D. P., Berg, W. K., McMahon, C., Templemen, M., McKinney, J., Swarts, V., Visser, M., & Marquardt, P. (1988). An evaluation of labeling-then-doing with moderately handicapped persons: Acquisition and generalization with complex tasks. Journal of Applied Behavior Analysis, 21, 331-344.

Ward, W. D., & Stare, S. W. (1990). The role of subject verbalization in generalized correspondence. Journal of Applied Behavior Analysis, 23, 129-136.

using videotape modeling to facilitate generalized purchasing skills

Documents