You are on page 1of 10

Journal oj Consulting and Clinical Psychology

1968, Vol. 32, No. 5, 565-574

REDUCTION OF SOCIAL ANXIETY THROUGH MODIFICATION OF


SELF-REINFORCEMENT:
AN INSTIGATION THERAPY TECHNIQUE1

LYNN P. REHM 2 ALBERT R. MARSTON


University of Wisconsin University of Southern California

Male college students who reported anxiety in social situations involving


females were seen in 1 of 3 therapy conditions. The experimental therapy
(SR) involved increasing Ss' rate and accuracy of positive self-reinforcement
using a hierarchy of situations which 5s sought out between sessions. One con-
trol treatment (NS) used nondirective techniques. The 2nd control (NT)
involved instructions to Ss to work on their own and report weekly. Pretesting
and posttesting difference scores yielded greatest improvement for the SR
group on (a) self-reports of anxiety and overt behavior; (6) verbal output
in a test simulating social interaction; and (c) generalization to scores on the
Manifest Anxiety Scale and Adjective Check List.

Skinner (19S3) suggests that one of the desired behavior, but emits it infrequently
ways in which individuals control their own because of lack of positive self-reinforcement.
behavior is by the administration of rewards The present experiment was an attempt to
to themselves without environmental restric- demonstrate the therapeutic utility of a tech-
tions and contingent upon certain behaviors. nique for handling a clinical problem involv-
A series of studies (Bandura & Kupers, 1964; ing faulty self-reinforcement. The procedure
Kanfer, Bradley, & Marston, 1962; Kanfer & combines elements from a number of sources,
Marston, 1963a, 1963b; Marston, 1964, but essentially uses manipulation of overt
196Sa; Marston & Kanfer, 1963) have in- self-reinforcement to effect a positive change
vestigated variables which influence this phe- in self-concept with associated reduction in
nomenon of self-reinforcement in experimental anxiety and increased approach to feared situ-
analogues. In discussing the relevance of self- ations. The experimental effort was explora-
reinforcement for psychotherapy, Marston tory in the sense that it was aimed at demon-
(196Sb) asserted that certain clinical prob- strating the viability of the technique as a
lems can best be conceptualized and treated whole, compared to other procedures, rather
in psychotherapy in self-reinforcement terms. than at assessing the separate contribution
Behavioral problems which are described as of the several elements independently. The
involving a low self-concept can be seen as a general strategy was to compare a group re-
malfunction of the client's self-reinforcement. ceiving the total experimental therapy with
In particular, a desired behavior may occur two control therapies (a) nonspecific counsel-
but is negatively evaluated, leading to dis- ing; (b) minimal urging toward self-help.
comfort; or the client is capable of the The study investigated college males who
reported a problem in meeting and dating
1
This paper is based on portions of a thesis sub- girls; that is, the problem involved feeling
mitted by the first author in partial fulfillment of uncomfortable in social situations with girls
the requirements for the degree of Master of Arts
at the University of Wisconsin, 1966. This investiga- and avoiding such situations. It was assumed
tion was supported, in part, by a United States that Ss were otherwise adequately function-
Public Health Service Research Fellowship Award ing males with at least a minimum repertoire
S-F1-MH-29, 410-02 (MTLH) from the National of social skills, but that they evaluated them-
Institutes of Health to the first author, and in part selves negatively when they did contact girls
by Research Grant MH-12235 awarded by National
Institutes of Mental Health to the second author. or they avoided heterosexual contact because
2
Now at Veterans Administration Center, Wood, of negative self-evaluation. The experimental
Wisconsin. therapy was a behavior therapy of the instiga-
565
566 LYNN P. REHM AND ALBERT R. MARSTON

tion type (Kanfer & Phillips, 1966) in that to place in rank order as to the amount of discom-
major emphasis was on the training of S in fort they would experience in each situation.
systematic self-help procedures. The major The Ss were then asked to designate a point on
the hierarchy at which they felt their discomfort
elements of the treatment involved gradual would become serious enough that they would pres-
approach to feared situations via a hierarchy ently begin to avoid such situations. Two additional
of stimuli, objective restructuring by S of hierarchy points were then designated which divided
his behavioral goals, and encouragement of items above and below this point into two smaller
blocks, thus creating four hierarchy levels. The Ss
increased positive self-reinforcement. were asked to try to make these divisions at points
where there would be a significant increase in dis-
METHOD comfort, with the limitation that there be at least
Subjects four items at each level.
The 5s for the experiment were 24 college males For instance, for one S, the first hierarchy level
who responded to either a verbal announcement included the item, "Taking a seat next to a girl in
made in psychology classes or a mimeographed notice class," the second level included, "Being introduced
sent to men's residence halls. The announcement to a gal while with a group of your friends," the
described the type of problem to be considered and third level included, "Calling a girl to ask her for a
requested anyone who felt he had such a problem date," and the fourth level included, "Dancing with
and wanted help in working on it to contact the a girl on a date." The most intensive physically
Psychology Department Research and Training Clinic sexual items on the list involved kissing. All Ss were
for an interview. Thirty students made appointments told that the study would not be concerned with
for the experiment. Of these, two reported homo- problems of more intense sexual behavior.
sexual problems and were referred for a general clinic (d) The 5s were provided with information sheets
intake. A third volunteer was rejected because he which they were asked to read before the first
was under the care of a psychiatrist at the time. therapy session to give them an idea of the theo-
Three Ss dropped out of the experiment during the retical background of the experiment. The informa-
first week and were replaced. Two had been assigned tion sheet for the SR therapy group explained the
to the Self-Reinforcement therapy group and a third concepts of reinforcement and self-reinforcement and
to the Nonspecific therapy group. the use of hierarchies and teaching machine tech-
In addition, the criterion test battery was admin- niques. The information sheet for the NS therapy
istered to 12 "normal" Ss. These were male students group gave the rationale that the experiment was
from Introduction to Psychology classes who signed an attempt to conduct short-term therapy in a
up for an experiment and received points toward more efficient manner. Some general principles of
the course grade for serving. nondirective therapy were included. The informa-
tion sheet for the NT group gave the rationale
Design that the experiment was examining the effect which
volunteering for therapy and the consequent commit-
A 3 X 4 factorial design was employed in the ment to change would have on an individual's own
analysis of the experiment. Three therapy groups approach to the problem.
consisted of the experimental therapy group, labeled (e) A battery of tests was administered to all Ss
Self-Reinforcement (SR), a Nonspecific therapy at the screening session. The taped Situation Test to
control group (NS) and a No Therapy control group be described below was given during the session
(NT). Four therapists each saw two 5s in each of proper and the paper and pencil tests were given
the three therapy groups. Data for the Normal Group to each S to be taken in the waiting room.
were handled in separate analyses.
Test Battery
Screening
The Situation Test was made up of two alternate
The screening interview conducted with each 5 forms each consisting of 10 social situations pre-
consisted of the following: (a) A screening question- sented orally on tape. The items were presented by
naire, followed by a brief interview, was used to a male voice which described a situation involving
determine that the client did indeed feel he had a a girl (e.g., "As you are leaving a cafeteria, a girl
problem, that uncontrollable environmental restric- taps you on the back and says . . ."). A female
tions were not the cause, and that the client was voice then read a line of dialogue to which the Ss
otherwise functioning adequately, (b) A position were asked to respond aloud (e.g., "I think you left
within the factorial design was randomly assigned this book."). The Ss were instructed to respond as
to each S, with the restriction that the experimenter they would in a real life conversation and their
doing the screening did not assign 5s to himself as responses were recorded on a second tape recorder.
therapist, (c) The Ss assigned to SR constructed a The experimenter left the room during the test. The
desensitization type hierarchy from 30 standard items two forms of the Situation Test were equated for
describing heterosexual social interactions. The items mean ranked discomfort level. Twenty male under-
were typed on 4 X S inch cards which Ss were asked graduates from Introduction to Psychology classes
REDUCTION OF SOCUL ANXIETY 567
had ranked the original pool of 20 mimeographed TABLE 1
items as they felt they would be ordered by male
RELIABILITIES OF DEPENDENT MEASURES PROM
peers with the proposed problem.
SITUATION TEST (PEARSON rs)
Seven different scales were derived from the
experimental Ss' responses to the taped Situation
Test. The Ss rated their own subjective anxiety for Parallel b
Measures Interscorer'
each item on a 7-point scale from "none" to forms' Intrascorer
"extreme." Estimates of the parallel forms reliability
of this and the other scales derived from the Situa- S anxiety .87* .— —

tion Test were obtained from the 12 normal 5s who Rated anxiety .82* .46 .77* .47*
took both forms of the test in a balanced sequence, Rated adequacy .67* .94* .85* .69*
within the same testing session. The parallel reli- Rated likability .72* .58* .93* .65*
ability for Ss' anxiety ratings was r=.87 (#<.01). Anxiety signs .OS — — —
The recorded responses were independently rated Log latency .74* — — •—
by three pairs of undergraduate girls on three dimen- Number of words .85* — — —
sions: anxiety, adequacy of response, and likability.
a
Rated anxiety and adequacy scores for each protocol Based on normal S group, N = 12.
i> Based on random selection of 10 5s' data,
consisted of the total points summed over the 10 ° Based on two ratings of all 5s' data, N = 72.
items each of which was rated on a 7-point scale. *P< .01.
Likability was rated on a single continuous SO-point
scale for the overall impression derived from the psychology indicated whether they thought each
10 responses on each protocol (Stoler, 1963). The adjective would be indicative of a positive or nega-
sets of responses for pretesting and posttesting for tive self-concept if used by a male college student
all experimental Ss plus responses to both forms of to describe himself. Complete agreement between
the test by the normal Ss, a. total of 72 protocols, these judges determined the valence of the adjectives.
were retaped in a random order on seven tapes. Only adjectives which obtained complete agreement
These tapes, which consisted of 10 or 11 sets of were scored (123 positive plus 138 negative out of
responses, were then rated by each rater in a separate a total of 300 adjectives). The obtained proportions
order. Interrater reliability was obtained by having were transformed by an arc-sine function to more
each rater rate one tape twice (first and last in their closely approximate a normal distribution.
sequence. The second rating was used in obtaining
mean scores for the two raters for each testing. Procedure for Therapy Sessions: Self-
Interrater reliabilities were obtained from correlations
between the pairs of raters. The reliability data for
Reinforcement Therapy Group
these three ratings are summarized in Table 1. In the first session the therapist handled questions
In addition, the recorded responses were scored and reiterated some of the principles involved. The
for average number of words per response, average therapist then explained to S that he was to work
log latency of response, and number of anxiety signs. systematically up the hierarchy by attempting to get
Anxiety signs were scored one point for each response into situations at given levels and then evaluate his
such as: a failure to respond, stuttering of a word, performance and reward himself with self-approval
repetition of words or phrases, halting within a for each situation. The Ss were instructed to make
sentence, unfinished sentences or mispronunciations. records each night of each situation encountered
The resulting total was divided by the average during the day on forms supplied by the therapist.
number of words per response and the resulting The forms required a brief description of the situa-
proportions were given an arc-sine transformation tion, including date, time, and what transpired, the
in order to approximate more closely a normal equivalent hierarchy item number, the hierarchy
distribution. level, and the number of points awarded. Amount
A Situation Questionnaire consisted of the 30 hier- of self-reinforcement was systematized by a point
archy items used in the SR therapy presented in a system. Zero points were to be given when the
random order. The Ss rated the amount of dis- situation resulted in S's overt avoidance or escape.
comfort they would feel in each situation on a One point was to be given for getting into the
7-point scale. The Manifest Anxiety Scale (MAS; situation with minimal performance of the situation's
Taylor, 19S3) and Fear Survey Schedule (FSS), requirements. Two points were to be given for any
Form III (a revised version of the Fear Survey performance beyond this minimum, and three points
Schedule developed by Wolpe and Lang, 1964) were were to be given for any situation in which S felt
administered as measures of generalized anxiety. The he did particularly well.
specific FSS item, "Being with a member of the The therapist and client recorded sample goals for
opposite sex," was also analyzed separately. 1-3-point rewards for each of the hierarchy items.
The Adjective Check List (ACL; Gough & Heil- Goals were established using the individual S's inter-
brun, 1965) was given and scored as a self-concept nalized standards for his own behavior, rather than
measure. The number of positive adjectives checked possible external reinforcements which the S might
was divided by a, number of positive plus negative desire from the situation or the amount of com-
items checked. Four male graduate students in clinical fort he might experience. For instance for the item,
568 LYNN P. REHM AND ALBERT R. MARSTON

"Calling up a girl for a date," a 2-point goal for one tered with the addition of a Posttherapy Question-
S was "extending the conversation to other topics." naire. The first 16 items of this questionnaire made
Goals such as "getting the date" or "feeling relaxed up a brief form of the Barrett-Lennard Relationship
on the phone" were discouraged by the therapist. Inventory (Barrett & Lennard, 1962, as revised in
In the first session, S was requested to focus on Spotts, 1965), Four items were selected from each
the lowest level of the hierarchy for the following of the inventory subscales. The rest of the question-
week. The 5s were encouraged to seek out these naire consisted of questions concerning 5s' attitude
situations in particular during the week, although toward the experiment, their progress, and their
they were also to record situations at other levels estimation of their own improvement. Statements
as they occurred. Hierarchy items were to be con- were presented and 5s were required to indicate
sidered as examples of classes of situations involving their degree of agreement or disagreement on a
similar interpersonal relationships. 6-point scale. They were also asked to indicate the
In subsequent sessions, 5s described each of the number of dates during treatment and whether this
situations which they had recorded during the week was a change in frequency.
and reported the points given to themselves. The
therapist reinforced the giving of 2 or 3 points for Follow Up
situations with verbal approval. For situations in A follow-up study was conducted in the period
which Ss gave themselves 0 or 1 point, the therapist 7-9 months after the end of the original study. All
asked them how they might have changed their 5s who could be located were contacted and asked
behavior to increase their self-evaluation. Therapists to come in to retake the test battery plus a follow-up
did not suggest new kinds of behavior. Increases questionnaire concerning their dating behavior and
in the number of points given were encouraged attitudes.
by the therapist. The therapist recorded the
number of points given, the equivalent item Therapists
number, and hierarchy level. The hierarchy level on
which 5s were instructed to focus increased in a The therapists were four third-year graduate
standard manner: Level 1 the first week, Level 2 students in clinical psychology, all of whom had had
the second week, Level 3 for the third and fourth at least one summer of clinical experience plus one
weeks, and Level 4 the fifth week. semester of therapy training in a practicum. Two
of these students (one of whom was the first author)
Procedure for Therapy Sessions'. Nonspecific also conducted the screening interviews. No 5 was
Therapy Group seen by the same person for screening and therapy.
Therapists met for two 1-hour sessions prior to the
For this group, therapists were instructed to em- experiment in which instructions were reviewed and
ploy a basically nondirective therapy technique which examples of situations, goals, point giving, etc., were
could include reflection and clarification of feeling, discussed. Therapists were asked to assume the atti-
low-level interpretations and questioning for infor- tude that all groups should show improvement and
mation. Interview content was unrestricted except to attempt to convey this attitude to all 5s.
that therapists were instructed not to suggest any
specific behavior which 5 should carry out and not RESULTS
to make any direct positive or negative evaluations
of the client's behavior. Therapy sessions were J Alternate Taped Situation Test Forms
hour long.
To avoid possible practice effects, two
Procedure for Therapy Sessions: No Therapy equivalent forms of the Taped Situation Test
Group were employed. The order of presentation of
Procedure for this group involved weekly sessions these two forms was reversed for half the 5s
where 5s were asked to report on how they con- in order to assess mean changes for all 5s. The
ceptualized their problem, how this conceptualization 5s within the same cell in the 3 X 4 factorial
had changed over the previous week, what they had design were treated as replications under the
done during the week to work on the problem, and assumption that the two forms of the test
what their plans for the following week were.
Therapists asked general questions and took notes were in fact equivalent. As a partial check
on the answers. They encouraged 5s to think through on this assumption, Therapy Group X Test
and work on the problem on their own but made no Form analyses of variance were performed for
specific suggestions and gave no evaluation of what all 24 5s on pretesting, posttesting, and dif-
5s were doing.
ference scores for all Taped Situation Test
Posttesting Session variables. The only significant test form effect
All 5s saw the experimenter who did the screening
was found for difference scores in average
for posttesting within 1 week after the final therapy number of words per response. The form se-
session. The same battery of tests was readminis- quence BA led to a greater average increase
REDUCTION OF SOCIAL ANXIETY 569
TABLE 2
COMPARISON OF ALL TREATED 5s AT PRETESTING WITH NORMAL 5s
I
Normal 5s Experimental 5s
\r _J n U l ~
Variable t

M SD M SD
Situation Questionnaire 71.8 16.50 121.3 21.62 6.79**
5s' anxiety ratings 23.8 6.78 37.3 7.98 4.87***
Raters' anxiety ratings 35.9 11.61 44.2 5.91 2.24*
Log latency .288 .123 .687 .140 7.98***
Anxiety signs .12 .132 .35 .383 2.58*
Adequacy ratings 46.9 7.58 45.8 4.10 <1.00
Number of words 10.1 4.10 9.2 3.87 <1.00
Taylor Manifest Anxiety Scale 15.8 7.24 23.3 8.10 2.64*
Fear Survey Schedule 199.0 14.70 228.1 63.54 2.08*
Gough Adjective Check List 2.21 .194 1.77 .400 4.31**

Note.—For normal 5s, N • 12; for experimental SB, N "23,


* t < .05.
** < ,01.

in number of words than sequence AB, Therapist X Therapy interaction effects were
F = 4.82, dj = 1/18, p < .05. Since this was found, only therapy group Fs are shown, All
the only significant finding out of 18 analy- differences were scored so that a positive
ses, and since even a tendency toward an ef- number indicates improvement. Thus, a de-
fect on the other variables would only increase crease in latency, a decrease in ratings of
the within cells error variance, Therapy anxiety and an increase in ratings of ade-
Group X Therapist analyses were considered quacy are all shown as positive numbers. For
appropriate. those variables which yielded a significant
positive correlation between pretesting and
Comparison of Pretreatment Experimental difference scores, the differences were divided
Subjects with Normal Group by the pretesting scores and the analyses of
To test whether the experimental Ss did variance were performed using the resulting
indeed form a deviant, more anxious sample proportions. This was done for the subjective
of the male college population, t tests were anxiety ratings made by the Ss (Table 3,
performed comparing the pretherapy scores of Line 1; r = .63, N = 23, p < .01), the anx-
the experimental Ss with the scores of iety ratings of the taped responses (Table 3,
the normal group. The results are presented Line 2; r — .43), and the Situation Question-
in Table 2. All variables, except the rater's naire scores (Table 3, Line 8; r — .34,
adequacy ratings and the average number of N = 23,p< .05),
words per response on the Situation Test, Analyses of the Situation Test variables
significantly differentiated the two groups. yielded significant results for two of the vari-
The two nonsignificant results were also in ables. Although the F value for Ss' anxiety
the predicted direction. ratings did not reach significance, subsequent
t tests supported the specific hypothesis that
Analysis of Improvement Measures8 the SR group would show significantly greater
Table 3 summarizes the results of the im- improvement than either of the control groups
provement scores between pretesting and post- (Table 3, Line 1). Similarly, the analyses of
testing. Since no significant therapist or variance of the average number of words
8 scale (Table 3, Line 7) did not yield a
One 5 dropped out of the SR group after two significant therapy group effect but subse-
therapy sessions. Analysis of variance used a least
squares solution (Ferguson, 1959, p. 259) for the quent t tests supported the hypothesis that
unequal N problem. the SR group would show a significantly
570 LYNN P. REHM AND ALBERT R. MAESTON

TABLE 3
SUMMARY OF COMPARISONS OF TREATMENT GROUPS ON ALL DEPENDENT MEASURES

Treatment mean change scores /-test values


Line
no. Variable
SR NS NT SR-NS SR-NT NS-NT

1 Situation Test : Subjective anxiety .33** .16 .13 1.87* 2.13* 1.17
2 Situation Test : Rated anxiety .11 .07 .19* .46 -.88 1.34
3 Situation Test : Rated adequacy 4.17 2.20 2.50 .69 .58 .11
4 Situation Test : Rated likability 19.1* 7.4 8.9 1.21 .52 .69
5 Situation Test : Anxiety signs .19 .06 -.11 .62 1.48 .85
6 Situation Test : Log latency .09 -.02 .08 1.74 .17 1.57
7 Situation Test: Number of words 4.03** 1.60 .05 1.79* 2.93** 1.14
8 Situation Questionnaire 42** .19* .09 2.42* 3.43** 1.01
9 Fear Survey Schedule 10.1 21.3 24.0 -.60 -.69 .14
10 Fear of opposite sex 1.43** 1.00* -.13 .86 3.12** 2.26*
11 Manifest Anxiety Scale 5.14* -1.50 .88 3.00** 1.93* 1.07
12 Adjective Check List .45** .21 .17 1.49 1.80* .31

Note.—SR = Self-Reinforcement; NS = Nonspecific; NT = No Therapy.


* p < .05; For treatment means columns, * refers to significance of / tests of mean change vs. zero change.
**p < .01.

greater increase in output in social situations conditions which did not differ significantly
than either of the other groups. As shown in from each other.
Table 3, the SR group also had the greatest The Adjective Check List was scored as
increase in rated adequacy (Line 3) and a self-concept measure under the hypothesis
likability (Line 4) and the greatest decrease that the SR group procedure should raise
in overt anxiety signs (Line 5) and latency the frequency of positive self-concept state-
of response (Line 6). None of these last four ments to a greater degree than the other
differences reached significance, however. two procedures. Only partial support was
The therapy groups varied significantly on found for this hypothesis. The SR group ob-
the Situations Questionnaire scores (Table 3, tained the greatest increase in positive self-
Line 8). The SR group reported a signifi- concept adjectives checked. However, the
cantly greater decrease in discomfort in the overall group effect was not significant and
30 situations than either the NS or NT groups of the t tests for this hypothesis, only the
as shown by the t test. difference between the SR and NT groups
On the Fear Survey Schedule no significant proved to be significant.
differences were found between the Therapy The Postherapy Questionnaire consisted of
groups (Table 3, Line 9). However, for the a collection of items covering several areas.
single item most relevant to the problem in- Two items concerned the number of dates
vestigated (Being with a member of the which the 5s had during the experiment as a
opposite sex), a significant groups effect was measure of overt behavioral change. The SR
found (Table 3, Line 10). Both the SR and group reported a greater frequency of dates
NS groups reported greater decreases in rated than either of the two groups which did not
fear than the NT group. The SR group re- differ from each other (Mann-Whitney U
ported a greater decrease than the NS group, test, p < .OS). The 5s also indicated whether
but this difference was not significant. their dating during the experiment was an
Differences in generalization of anxiety increase, about the same, or a decrease as
reduction were found on the Manifest Anxiety compared with their usual dating. A com-
Scale (Table 3, Line 11). A significant groups parison of the SR versus the combined NS
effect was found on general anxiety reduction and NT groups on their indication of an
and significant t test differences were found increase versus no change or a decrease in
between the SR group and the two control dating yielded a significant x2 — 3.84, dj — 1,
REDUCTION OF SOCIAL ANXIETY 571
p < .05, the SR group more often indicating of 0-80 were derived from each 5, the 5-
an increase in dating. point item scale being reversed for negatively
The remaining items required Ss to respond stated items. A two-way analysis of variance
on a 6-point scale, as to the strength of their for therapy and therapist effects yielded no
agreement or disagreement with a series of significant effects. Nor were any subsequent
statements. Nine of these statements con- t tests between therapy groups significant.
cerned feelings of improvement or change in Means for the three groups were 54.9 for
attitude during the experiment (Table 4). SR, 60.6 for NS, and 55.5 for NT. There was
One-way analyses of variance of agreement no evidence from this data that relationship
scores yielded significant therapy group effects factors were responsible for any greater im-
for items concerning more interaction with provement on the part of the SR group. In
girls (Line 1), more anticipated interactions fact, this group rated the therapists as least
with girls (Line 2), more independent self- positive in the relationships.
evaluation (Line 4), more self-confidence The remaining nine items of the Postther-
with girls (Line 5), and a general statement apy Questionnaire concerned attitudes toward
of feeling helped by the experiment (Line 9). the experimental procedures and purpose of
On all but the last item t tests demonstrated the experiment. No significant results were
significantly greater agreement by SR group found in one-way analyses of these data.
5s than by either of the other groups. The
SR and NT groups did not differ significantly SR Group Records
on the item concerning feelings of being The SR group procedure was aimed at
helped by the experiment. increasing (a) the number of interactions
About one half of the questionnaire items with the opposite sex, (b) the level of in-
concerned the quality of the relationship with volvement in these interactions as opera-
the therapist. Sixteen items were selected tionalized by the hierarchy, and (c) the
from a revised version of the Barrett-Lennard amount of positive self-reinforcement result-
Relationship Inventory. Four items were ing from these interactions. The average
drawn from each of the original four sub- number of interactions recorded by the Ss for
scales of positive regard, empathic under- the 5 weeks of recording were, in order, 13.2,
standing, congruence, and unconditionality of 9.7, 8.0, 8.9, and 6.7. The general downward
regard. Total scores with a possible range trend over weeks was the reverse of what was

TABLE 4
SUMMARY OF TREATMENT COMPARISONS ON POSTTHERAPY QUESTIONNAIRE ITEMS

Treat-
M ment /-test values
Line Item effects
no.
SR NS NT /?a SR-NS SR-NT NS-NT

1 20 Interacts more with girls 4.7 3.4 2.8 4.48* 2.08* 3.05** <1.00
2 21 Probably will interact more in future 4.4 1.9 2.9 8.19** 4.17** 2.54** 1.66
3 22 More self-confident with girls 4.0 1.9 2.6 4.03* 2.91* 1.88* -1.03
4 23 More insight into problem 3.6 2.3 3.1 1.56 1.78* <1.00 -1.18
5 24 Can evaluate self more independently 4.1 2.4 3.0 4.67* 3.14** 2.03* -1.11
6 25 Did better with girls than had expected 3.4 2.0 3.0 2.17 2.08* <1.00 -1.45
7 26 More aware of feeling about self 3.0 2.0 3.3 1.94 1.50 <1.00 -1.87
8 32 Experiment was helpful 4.0 2.4 3.5 4.29* 2.93** <1.00 -2.03
9 33 More insight into self 3.3 2.6 3.4 <1.00 <1.00 <1.00 <1.00

Note.—SR = Self-Reinforcement; NS - Nonspecific; NT = No Therapy.


' if = 2/20.
*p < .05.
**p < .01.
572 LYNN P. REHM AND ALBKRT R. MARSTON

expected but may be accounted for by several df = 14, p < .05, and "I feel I can evaluate
factors. Initial enthusiasm and ease of acces- my behavior in social situations more ade-
sibility may account for the greater number quately now," t = 2.02, df = 14, p < .05. No
of situations in the early weeks. Later on, 5s significant differences were found on items
tended to seek out only the higher, more diffi- concerning dating, though the SR group re-
cult, and less frequent situations and in some ported a mean frequency of dates twice as
cases failed to report some situations low on great as the combined control group (twice/
the hierarchy which then seemed less im- month vs. once/month).
portant.
The averages of each 5's mean item level DISCUSSION
on the 30-point hierarchies for all 5s in the Using a population whose severity of pa-
SR group were 10.0, 14.S, 16.2, 17.6, and thology lay somewhere between the normal
17.3 for the 5 weeks of recording. An analysis experiment volunteer and the patient who
of variance of the individual 5 means yielded spontaneously seeks psychotherapy, this ex-
a significant weeks effect, F = 2.79, df = 4/ periment has demonstrated greater improve-
30, p < .05. A trend analysis of these data ment for the experimental therapy group
yielded a significant linear component over (SR) than for either of the control groups on
weeks, F - 9.06, df = 1/30, p < .01. The SR a variety of measures. In terms of Lang's
group 5s did make progress up the hierarchy (1964) analysis of fear into three modes of
during the experiment. expression, greater reduction in anxiety has
The averages of each 5's mean number of been shown in (a) several measures of verbal
points awarded to himself for the situation report of discomfort in heterosexual social
were 2.39, 2.63, 2.68, 2.81, and 2.77 for the 5 situations, and (b) an approximation to direct
weeks. An analysis of variance of the mean measurement of overt-motor change (i.e.,
number of points per situation yielded a reports of more dates with girls and increased
weeks effect which did not reach significance, verbal output in the Situation Test). The
F = 2.02, df = 4/30, p < .10. A trend analy- third, somatic, mode was not tapped in the
sis of these data yielded a significant linear present study. In addition, generalization of
component, F = 7.21, df = 1/30, p < .05. improvement was demonstrated by greater
The SR group did increase in amount of self- reduction in Taylor Manifest Anxiety Scale
reinforcement during the experiment as de- scores by the SR group and a greater increase
fined by the mean number of points per situ- in positive self-concept was suggested by
ation which they awarded themselves. Gough Adjective Check List scores. While a
Follow-Up Treatment Differences number of the significant differences disap-
peared in the analyses of the follow-up data,
Sixteen of the original 23 5s were located a sufficient sample of the dependent measures,
and retested during the follow-up period (6 particularly those involving self-report, con-
SR, 4 NS, and 6 NT). The NS and NT 5s tinued to show the superiority of the experi-
were grouped into one combined control group mental therapy to warrant serious considera-
for follow-up comparisons. Pretesting to fol- tions of this approach to behavior modifica-
low-up differences were compared for the re-
tion.
sulting two groups on the various dependent
Since the strategy of this research involved
variables. The SR group showed significantly
greater improvement at the follow-up period an initial combining of several elements in the
on their self-report measures: (a) Situation experimental therapy, the control groups did
Questionnaire, t = 2.57, p < .05, (6) 5's not allow for a clear-cut identification of the
anxiety ratings on the taped situation active factors in the SR therapy. The control
test, t = 1.99, df = 14, p < .05 and (c) Tay- groups did seem to demonstrate that improve-
lor Manifest Anxiety Scale scores, t = 2.02, ment was not due to factors such as commit-
df = 14, p < .05. On the follow-up question- ment to change or a positive relationship with
naire SR 5s more strongly endorsed the items the therapist. Therefore, at this stage the ele-
"I understand myself better now," t = 1.84, ments of the SR therapy need to be further
REDUCTION OF SOCIAL ANXIETY 573

analyzed, both for theoretical reasons and to each situation was stressed to S. Kanfer and
specify future control conditions. Marston (1963a) demonstrated increased
The SR therapy was based on the assump- self-reinforcement with instructions facilitat-
tion that overt behavior and experienced ing lower criteria. Covert self-evaluation is
anxiety can be modified by the manipulation assumed to be a function of many cues in an
of cognitive states, that is, covert self-reac- ongoing interaction. The experimenters in-
tions. This assumption is essentially the one structed and encouraged Ss to use only the
made in traditional insight psychotherapy. carrying out of various behavioral acts as
The difference, at least in intent, was the their goals and bases for self-evaluation. Goals
concentration on a specific category of cogni- such as not feeling nervous or getting a posi-
tive response (i.e., self-evaluation or self- tive reaction from another person were dis-
reinforcement) in a systematic manner, em- couraged. The stimuli to which S was to re-
ploying learning techniques to manipulate spond were delimited to more clearly observ-
overt verbalizations of these responses. able behavior. To some extent this can be
The therapy was designed to shape positive seen as altering the perceived stimulus con-
self-reinforcement in such a way as to increase text in which S's self-evaluations are made.
its probability of occurrence in the presence Relearning in this context might be expected
of the discriminative stimuli arising from the to later generalize to other contexts such as
performance of the desired behavioral re- perception of feedback from others.
sponse, that is, approach to heterosexual so- The mechanism of generalization of a
cial situations. In turn, the increased self- behavior change brought about in therapy
reinforcement was assumed to increase the continues to be a difficult problem. Most pa-
probability of the desired approach response tients are not monosymptomatic, though one
on future occasions. The therapist acted, in symptom may be focal. Do we need to treat
theory, almost as if the patient were two each symptom or each area of behavior defi-
organisms: one socially responding and one cit? One of the hopes of the SR approach is
self-evaluating. Alternatively, he acted as if that self-evaluation responses have a very flat
he were attempting to shape a sequence of generalization gradient. As has been indicated
two chained responses: that is, a social ap- earlier, many patients suffer from over-gen-
proach response and a self-evaluating response. eralized negative self-evaluation. It is possible
According to this analysis, if S's positive that positive self-evaluation will also gen-
self-reinforcing behavior can be enhanced, then eralize broadly, and, in combination with the
the positive change in social behavior should patient's training in the use of response hier-
be self-sustaining, independent of the thera- archies, serve as a mediating mechanism for
pist. The problem then is to elicit the original the patient to shape other response clusters.
positive self-evaluations. Several procedures Finally, a comment about anxiety. The
were aimed at this goal. Individual situations patients reported experiencing anxiety and
were differentiated and arranged into a hier- the experimenters used reported reduction in
archy. Marston (196Sb) suggested that an anxiety as a major measure of improvement.
individual's self-concept in a particular area Yet the patients were explicitly told to avoid
may be tied to a single self-descriptive label. using absence of anxiety as a criterion for SR,
By differentiating among these situations, 5 and the above analysis of the SR therapy
is aided in evaluating his behavior in each approach avoided use of the concept of anxi-
situation individually. The formation of a ety. Therefore, an additional, untested as-
hierarchy also is a means of differentiating sumption of the approach must be stated:
situations on a dimension of difficulty, and of negative self-evaluation is a primary cue for
encouraging S's behaving in a graded series of anxiety; reduction of the former will reduce
problem situations. the latter. A corollary of this would be that
Explicit behavioral goals were set up with positive self-evaluation is a cue for responses
each individual S in accordance with his own which are incompatible with anxiety. At this
criteria for varying degrees of success in each stage of development, it is not possible to go
situation. The setting of minimal goals for further than this oversimplified statement of
574 LYNN P. REHM AND ALBERT R. MARSTON

the relationship. The positive results of this KANFER, F. H., & MARSTON, A. R. Conditioning of
study seem to confirm the assumption, but self-reinforcing responses: An analogue to self-
confidence training. Psychological Reports, 1963,
much work is needed to determine whether re- 13, 63-70. (a)
duction of anxiety could have occurred with- KANFER, F. H., & MARSTON, A. R. Determinants of
out any emphasis on or apparent change in self-reinforcement in human learning. Journal of
self-evaluation responses. Experimental Psychology, 1963, 66, 245-254. (b)
KANFER, F. H., & PHILLIPS, J. S. Behavior therapy:
This analysis of the SR therapy points to A panacea for all ills or a passjng fancy. Archives
several needed controls for future research. of General Psychiatry, 1966, IS, 114-128.
These can be stated in the form of the ques- LAND, P. J. Experimental studies in desensitization
tions: "Is effective therapy due to: (a) in- psychotherapy. In J. Wolpe, A. Salter, & L. J.
Reyna (Eds.), The conditioning therapies. New
struction and reinforcement of use of SR, (b) York: Holt, Rinehart & Winston, 1964.
use of hierarchical approach to anxiety-pro- MARSTON, A. R. Variables affecting incidence of self-
ducing situations, (c) the statement of ex- reinforcement. Psychological Reports, 1964, 14,
plicit behavioral goals in each situation, (d) 879-884.
the de-emphasis of external evaluation and/or MARSTON, A. R. Imitation, self-reinforcement, and
reinforcement of another person. Journal of Per-
anxiety reduction as immediate behavior sonality and Social Psychology, 1965, 2, 255-261.
change goals?" Correlational analyses in the (a)
present study provided inadequate separate MARSTON, A. R. Self-reinforcement: The relevance
evaluation of these elements, and future re- of a concept in analogue research to psychother-
search must experimentally perform the sepa- apy. Psychotherapy: Theory, research and prac-
tice, 1965, 2, 3-5. (b)
ration. MARSTON, A. R., & KANFER, F. H. Human reinforce-
ment: Experimenter and subject controlled. Jour-
REFERENCES
nal of Experimental Psychology, 1963, 66, 91-94.
BANDURA, A., & KUPERS, C. J. The transmission of SKINNER, B. F. Science and human behavior. New
patterns of self-reinforcement through modeling. York: Macmillan, 1953.
Journal of Abnormal and Social Psychology, 1964, SPOTTS, J. E. Some effects of exposure to a psycho-
69, 1-9. therapy rating task in teachers of emotionally dis-
BARRETT-LENNARD, G. T. Dimensions of therapist turbed adolescents. Unpublished doctoral disserta-
response as causal factors in therapeutic change. tion, University of Wisconsin, 1965.
Psychological Monographs, 1962, 76(43, Whole No. STOLER, N. Client likeability: A variable in the study
562). of psychotherapy. Journal of Consulting Psychol-
FERGUSON, G. A. Statistical analysis in psychology ogy, 1963, 27, 175-178.
and education. New York: McGraw-Hill, 19S9. TAYLOR, J. A. A personality scale of manifest anxi-
ety. Journal of Abnormal and Social Psychology,
GOUOH, H. G., & HEH.BRUN, A. B., JR. The Adjective
1953, 48, 285-290.
Check List manual. Palo Alto, California: Con-
WOLPE, J., & LANG, P. J. A fear survey schedule for
sulting Psychologists Press, 196S.
use in behavior therapy. Behavior Research and
KANFER, F. H., BRADLEY, M. M., & MARSTON, A. R. Therapy, 1964, 2, 27-30.
Self-reinforcement as a function of degree of learn-
ing. Psychological Reports, 1962, 10, 885-886. (Received November 10, 1967)

You might also like