Parent-child interaction in the laboratory: Effects of role ... - Springer Link

20 downloads 0 Views 706KB Size Report
M. Knudson, Barclay Martin, and Leonard G. Rorer for their comments on drafts of ... hypothesis, subjects' VRM use should show clear effects of role differences.
Journal of Abnormal ChildPsychology, Vol. 9, No. 2, 1981,pp 229-241

Parent-Child Interaction in the Laboratory: Effects of Role, Task, and Child Behavior Pathology on Verbal Response Mode Use ~ W i l l i a m B. Stiles 2

Miami University, Oxford, Ohio M . Lauren W h i t e

University of North Carolinaat ChapelHill

Fourth-, fifth-, or sixth-grade children identified as aggressive, withdrawn, or nondeviant by a consensus o f teacher and peer ratings interacted with one of their parents in a modified revealed difference task. Transcripts of the interactions were coded according to a taxonomy of verbal response modes (VRMs). Comparison of VRM profiles showed small and equivocal differences between trait groups. However, differences between roles (parent and child) and differences between two phases o f the task ("reach agreement" on a problematic situation and "'tell you how you feel" in that situation) were large and highly significant. Results support the hypothesis that intersubjective coding categories, such as VRMs, are sensitive to role, task, and relationship variables but relatively insensitive to enduring individual differences in personality or behavior pathology.

Manuscript received in final form July 22, 1980. JCollection of the data on which this study was based was supported by grant no. MH12474 from the National Institute of Mental Health to Barclay Martin. We thank Barclay Martin for making these data available to us; we thank Faye Sultan, Kathy Cody, Brenda Huddle, Jan Markham, Andy Russell, and Dawn Tucker for coding transcripts; and we thank Roger M. Knudson, Barclay Martin, and Leonard G. Rorer for their comments on drafts of this article. 2Address all correspondence to William B. Stiles, Department of Psychology, Miami University, Oxford, Ohio 45056. 229 0091-0627/81/0600-0229503.00/0 9 1981 Plenum Publishing Corporation

230

Stiles and White

The three-channel hypothesis, developed from a review of the psychotherapy process literature (Russell & Stiles, 1979), distinguishes three kinds of information transmitted in spoken communication and proposes that each corresponds primarily, though not exclusively, to a distinctive class of coding categories. According to this hypothesis, the content channel carries information about central concerns, attitudes, personality, and psychodynamics; content categories, such as "furniture," "maternal references," or "oral strivings" concern the denotative, connotative, or symbolic meanings of speech units. The extralinguistic channel carries information about the speaker's mood or transitory emotional state; extralinguistic categories, such as "interruptions," "laughter," or "hesitation," concern speech-related behaviors defined independently of semantic or syntactic structures. The intersubjective channel, which is of central interest in the present article, carries information about the relationship between interactants, the task they are engaged in, and their roles in that task. Intersubjective categories, such as "question," "self-disclosure," and "advice" specify ways in which speaker and recipient are (briefly) related to each other; the existence of the recipient is implicit in the category. For example, questions ask the recipient for information; self-disclosures reveal personal information to the recipient; advice tells the recipient what to do. The present study applied a general-purpose intersubjective coding system, a taxonomy of verbal response modes (VRMs) fully described elsewhere (Stiles, 1978b), to laboratory interactions (originally gathered by Martin & Hetherington, 1971) between parents and children rated by teachers and peers as aggressive, withdrawn, or nondeviant. The interaction procedure included two phases, intended to sample problem-solving and affect expression styles, respectively. According to the three-channel hypothesis, subjects' VRM use should show clear effects of role differences (parent versus child) and task differences (problem-solving versus affect expression) but might be relatively insensitive to enduring individual differences in personality or psychopathology (aggressive versus withdrawn versus nondeviant trait groups). This formulation could help explain the disappointing progress in identifying pathological family interaction patterns by coding laboratory interactions of disturbed and normal families (see reviews by Frank, 1965; Hetherington & Martin, 1979; Jacob, 1975; Riskin & Faunce, 1972). Individual studies have found intersubjective coding categories that "significantly" discriminate groups, but the number of significance tests has been large and the pattern of results across studies has been inconsistent, raising the possibility that some of the reported discriminations reflect type I errors--rejection of the null hypothesis when it is true.

231

Parent - Child Interaction in the Lab METHOD

The Martin and Hetherington (1971) Study Children from 121 fourth-, fifth-, and sixth-grade classrooms in 18 Madison, Wisconsin, public schools were identified as aggressive, withdrawn, or nondeviant based on two sets of teacher ratings on a modified form of Peterson's (1961) Problem Checklist, taken several months apart, and one set of ratings by same-sex classmates using a modified form of the Peer Nomination Inventory (Wiggins & Winder, 1961). Relatively pure types were selected by requiring aggressive children to be rated low on withdrawal, withdrawn children to be rated low on aggression, and nondeviant children to be rated low on both. Trait groups (i.e., aggressive, withdrawn, nondeviant) were matched for child age and IQ, and were additionally compared for birth order, number of children in the family, and each parent's age, education, and occupational level. Groups were well matched on all of these variables except that families of aggressive boys (but not girls) were somewhat lower, and families of withdrawn boys somewhat higher, for father's occupational level and mother's education. Families who met selection criteria were sent a form letter offering $10 for their participation in each of two sessions. Approximately 6007o of the families in each subgroup agreed to participate. Details of subject characteristics and selection procedures are described by Martin and Hetherington (1971). In the first session, both parents were interviewed to gather detailed reports about parent-child interaction during the preceding 2-3 years. Ratings of the amount of aggression shown by the child toward parents, siblings, and peers, based on these interviews, supported the generality of the criterion traits; that is, trait groups varied significantly in aggression, with the aggressive group highest, withdrawn group lowest, and nondeviant group intermediate. Results for ratings of child-rearing practices were less clear-cut but did suggest plausible clusters of variables that distinguished trait groups from each other (Martin & Hetherington, 1971). In the second session, from which data for the present study were taken, semistructured interactions between each parent and the child were tape-recorded and later transcribed. In half of the families fathers participated first, and in the other half mothers participated first. After an initial warm-up exercise, each parent-child pair was given the following instructions while the other parent waited in another room: Phase 1 (Reach AgreemenO. "All parents and children have disagreements at times. No child is perfect or always does exactly what his/her father or mother wants. Nor do parents always act as their children might

232

Stiles and White

like them to. I would like for the two of you to talk to each other and decide upon some situation where you have occasional disagreements. Perhaps you prefer not to use the word disagreement, but any kind of situation that upsets one or both of you will do." An attempt was made to select a topic that involved some continuing source of conflict and was not too trivial. Typical "situations" were disagreements about bedtime, TV watching, and chores. Dyads were then told, " I would like for you to imagine that something like that has just happened. Talk to each other and try to reach agreement about what, if anything, should be done, both right now and in the future." Phase 2 (Tell How You Feel). When they had completed the "reach agreement" phase, dyads were told: "All right, now I would like for you to talk to each other and tell each other what your feelings, your emotions, are when you are in this situation, or what your feelings are about this situation. Just tell each other what your feelings and emotions are."

Both phases were then repeated for a second situation about which they had a continuing disagreement. Thus, four interaction samples were obtained from each parent-child dyad, two phases for each of two situations.

Selection of Families for Present Study Transcripts of interactions from 62 families, including 40 nondeviant, 12 withdrawn, and 10 aggressive children, were selected from the Martin and Hetherington data using a stratified random-sampling method. Only transcripts of the first parent's interaction with each child were used. (In the Martin & Hetherington study, an experimental manipulation was introduced in the second parent's interaction.) Within each group, equal numbers of father-son, father-daughter, mother-son, and motherdaughter dyads were included, 10 of each for the nondeviant group and 3 of each for the withdrawn and aggressive groups, except that there were only 2 mother-daughter and 2 father-daughter dyads from the aggressive group in the pool. 3 Balancing the trait groups for sex of parent

~Martin and Hetherington studied 221 families, although not all provided complete data. Of these, 78 were studied during the first year of the study (1966-1967) using slightly different procedures. In the second-year (1967-1968) sample, from which transcripts for the present study were selected, 21 were in a group of children with "mixed" aggressive and withdrawn deviant behavior who were not included in the present study, and of the remainder, five families failed to complete the second session, This left 79 nondeviant, 21 withdrawn, and 17 aggressive children in the pool from which the present sample was drawn. In some classifications (e.g., aggressive girls) all eligible families were included in the "random" sample.

Parent - Child Interaction in the Lab

233

and child, together with the matching of groups for child age and IQ, brings this study into approximate conformity with Jacob's (1975) recommendations for controlling these sociodemographic characteristics. VRM Coding According to the VRM taxonomy (Stiles, 1978b), both the grammatical form and the interpersonal intent of each utterance (each independent clause, nonrestrictive dependent clause, compound predicate, or term of acknowledgment) are classified into eight mutually exclusive and exhaustive categories, question (Q), disclosure (D), edification (E), acknowledgment (K), confirmation (C), interpretation (I), reflection (R), or advisement (A), summarized in Table I. Utterances in which form and intent coincide are called "pure modes"; utterances in which form and intent differ are called "mixed modes." To illustrate: "Clean up your room" is a pure advisement, written A(A), whereas "Don't you think you should clean up your room?" is a mixed mode, with question form and advisement intent, written Q(A).

Table I. Verbal Response Modes a Form Question

(Q)

Disclosure

(D)

Edification

(E)

Acknowledgment

(K)

Confirmation

(C)

Interpretation

(I)

Reflection

(R)

Advisement

(A)

Interrogative, inverted subject-verb order Declarative, first person singular ( " I " ) Declarative, third person (e.g., " h e , " "she," " i t " ) Nonlexical or contentless utterances, terms of address and salutation First person plural ("we") where referrent includes other Second person ("you"), verb implies an attribute or ability, terms of evaluation Second person ("you"), verb implies internal experience or volitional action, finishing other's sentence Imperative or second person with verb of permission, prohibition, or obligation

Intent Requests information or guidance Reveals thoughts, feelings, perceptions, intentions States objective information Conveys receipt of or receptiveness to communication Compares speaker's experience with other's; agreement, disagreement Explains or labels the other; judgments, evaluations of other Puts other's experience into words; repetitions, restatements, clarifications Attempts to guide behaviOr; suggestions, commands, advice

aBoth form and intent are coded for each utterance. The form symbol is written first and intent second, in parentheses. For example, K(D) means acknowledgment form with disclosure intent.

234

Stiles and White

Each of the 248 transcripts (62 dyads x 2 problems situations • 2 phases) was unitized and coded according to the VRM manual (Stiles, 1978a) b y one of us (M.L.W.), who was blind to which trait group families were in. Each transcript was then coded independently by two additional volunteer coders (six individuals participated; all were undergraduate psychology students), who were blind to the study's design and who had completed at least two practice transcripts. The three sets of codes were combined into a final composite set on a two-out-of-three basis; forms or intents on which all three coders disagreed were coded "disagreement." Two-out-of-three agreement on all 8,871 utterances was 96.90/o for form and 92.20/o for intent, whereas median pairwise intercoder agreement was 80.9~ for form and 65.8~ for intent. The substantial improvement over the pairwise agreement reflects coders' implicit agreement that utterances were in one of two modes. For example, coders might be unsure whether arl utterance intent was edification or disclosure, but none would call it any of the other six intents. When three coders choose between two modes, at least two must agree.

RESULTS

Effects of Role and Task The composite codes were used to calculate each subject's VRM

profile--the percentage of that person's utterances in each mode--for each phase ("reach agreement" and "tell you how you feel") of each problematic situation. Although most of the 64 modes (form-intent combinations) were used at least once in this sample, only 13 modes met the criterion that they were used an average of at least 3% by parents or by children in either phase 1 or phase 2, averaged across the two situations, and these 13 were selected for further consideration (see Table II). As Table II shows, parents' VRM profiles were clearly different from their children's in both phases, and both parents and children behaved differently in phase 1 (reach agreement) and phase 2 (tell how you feel). In both phases, over 25% of parent utterances were questions, e.g., "What should we do about it?" Q(Q), occasionally with reflection form, e.g., "You don't want to play with him anymore?" R(Q). By contrast, fewer than 5% of children's utterances in either phase were questions. Parents used many more pure advisements, e.g., "You have to wait till you're in junior high school to get your ears pierced," interpretations,

Parent- Child Interaction in the Lab

235

Table II. Mean Verbal Response Mode Profiles for Parents and Children in

Phase 1 (Reach Agreement) and Phase 2 (Tell How You Feel) of a Modified Revealed Differences Taska Parents Mode Q(Q) R(Q) D(D) E(D) K(D) E(E) D(E) K(K) K(C) l(C) I(I) R(R) A(A) Otherg Total Utterances per interaction

Children

Phase 1

Phase 2

Phase 1

Phase 2

23.2 1.8 7.2 1.7 .1 4.7 2.0 4.7 .7 .5 6.1 3.3 5.3 38.7 100.0

24.7 3.2 15.5 4.5 .1 2.0 .5 5.6 .5 .2 4.7 3.8 3.1 31.6 100.0

4.5 .1 24.1 3.3 16.3 6.4 4.9 4.4 7.5 3.5 .9 1.4 2.5 20.2 100.0

1.9 .2 38.0 5.8 19.4 3.3 3.0 5.0 5.6 .4 .6 1.2 .1 15.5 100.0

28.8

19.6

14.0

9.2

Significant differencesb

c,d,f c,d,e c,d,e,f e,f c,d e,f c,d c,d c,f c,d c,d c,d,e,f

aMean percentages for each role and phase were found by averaging across the two interactions (problematic situations) for each parent-child dyad, and then across 62 dyads. Mode abbreviations are question (Q), disclosure (D), edification (E), aknowledgment (K), confirmation (C), interpretation (I), reflection (R), and advisement (A). Form is written first, intent second, in parenthesis. bDifferences significant by t test for correlated samples, p < .05, df = 61, two-tailed tests. Cparents versus children in phase 1. dparents versus children in phase 2. eparents in phase 1 versus phase 2. fChildren in phase 1 versus phase 2. g"Other" category includes modes averaging less than 3~ for both parents and children in both phases, plus utterances coded unscorable or disagreement.

e.g., " Y o u ' r e u s u a l l y the one who starts i t , " a n d reflections, e.g., " y o u d o n ' t like to do the d i s h e s , " t h a n their children. These m o d e s are p r e s u m p t u o u s , i.e., they p r e s u m e k n o w l e d g e o f what the o t h e r ' s i n t e r n a l experience or v o l i t i o n a l b e h a v i o r is or should be (Stiles, 1978b) a n d their greater use b y p a r e n t s t h a n children follows a consistent p a t t e r n f o u n d in other s t a t u s - d i s c r e p a n t dyads, i n c l u d i n g p r o f e s s o r - s t u d e n t dyads (Stiles, W a s z a k , & B a r t o n , 1979), s e n i o r - f r e s h m a n dyads (Cansler & Stiles, in press), p h y s i c i a n - p a t i e n t dyads (Stiles, P u t n a m , James, & W o l f , 1979), a n d p s y c h o t h e r a p i s t - c l i e n t dyads (Stiles & S u l t a n , 1979).

236

Stiles and White

The predominant modes in children's VRM profiles included (1) the four exposition modes, D(D), E(E), D(E), and E(D), so called because they compose the bulk of all expository discourse (see also Stiles, Putnam, Wolf, & James, 1979a), and (2) modes with acknowledgment form, K(D), K(C), and K(K). The exposition modes are either first-person (disclosure form) or third-person (edification form) declarative sentences that express either subjective and private (disclosure intent) or objective and observable (edification intent) information. To illustrate: "I was mad at her" is D(D); "She started it" is E(E); "I didn't hit her" is D(E); and "She made me angry" is E(D). Acknowledgment forms are brief and contentless, e.g., "Mm-hm," "Yeah," " N o , " " O h . " Depending on context, they may convey subjective information (disclosure intent), agreement or disagreement (confirmation intent), or merely receipt of the other's communication (acknowledgment intent). The child mode I(C) fits in this second category; these were brief, evaluative (interpretation) words ("right," " g o o d , " " o k a y , " "fine") used to convey agreement with parent statements rather than evaluation of the parent. From phase 1 to phase 2 both parents and children showed a large increase in the percentage of modes with disclosure intent, particularly D(D) and E(D), with a compensating decrease in various other modes. The increase shows that subjects tended to follow instructions to "tell your feelings and emotions" in the second phase. However, parents did not entirely refrain from interjecting evaluations (interpretations) and directives (advisements).

Compar&on of the Two Problematic Situations VRM use in the two problematic situations (i.e., the repetitions of the task) appeared very similar. Among 52 t tests comparing the percentages of the 13 selected modes in situation 1 versus situation 2 (for parents and for children in phase 1 and in phase 2), only 2 were significant at the .05 level, approximately the number expected by chance. This justified averaging mode use across situations for the other analyses. On the other hand, if VRM percentages in task repetitions are considered as test and retest of measures of subjects' enduring traits, then they are not very reliable measures. For phase 1 (reach agreement), the "reliability coefficients" were significant (p < .05; N = 62 for each correlation) for only four of the parent modes--. 58 for Q(Q), .45 for E(E), .41 for E(D), and .35 for R(Q)--and only two of the child modes-.57 for K(D) and .45 for Q(Q). For phase 2 (tell how you feel) only one

Parent- Child Interaction in the Lab

237

parent correlation--.29 for D(D)--and one child correlation--.37 for K(D)--reached significance.

Effects of Trait Group Multivariate analyses of variance (MANOVAs) were applied to the mean percentages (averaged across the two situations) of parent and child VRM use in each phase. The independent variables were trait group (nondeviant, withdrawn, aggressive), sex of parent (father or mother), and sex of child (son or daughter). These analyses thus tested whether VRM use differed among trait groups, or whether parent's or child's sex, or an interaction of trait group and sex, influenced VRM use. To reduce the number of variables with many zero values, modes averaging less than 3~ in both phases were eliminated from each role's profile, leaving 9 modes in each MANOVA (cf. Table II). Results of these four MANOVAs were within the bounds of chance expectations. Of the 12 multivariate tests of main effects, and 16 tests of two- and three-way interactions, only one main effect and one three-way interaction effect reached the .05 level of significance. Of 108 univariate F tests of main effects (3 effects x 2 roles • 2 phases x 9 modes each) and 144 univariate F tests of two- and three-way interaction effects, only 2 reached the .01 level and 11 more reached the .05 level. The one multivariate main effect was of trait group on parent VRM use in phase 2, Pillai's trace V -- .61, approximate F(18,86) = 2.09, p < .05, reflecting significant univariate effects for parent K(K)--pure acknowledgments--F(2,50) = 3.45, p < .05, and for parent A(A)--pure advisements--F(2,50) = 3.67, p < .05. In phase 2, parents of aggressive, withdrawn, and nondeviant children averaged 6.3o70, 8.3%, and 4.60/o K(K), respectively, and 6.0o70, 9%, and 3.1070 A(A), respectively. Parents' interactions with aggressive children might plausibly involve more advisements (e.g., commands aimed at controlling aggressive behavior) than normal, whereas interactions with withdrawn children might involve fewer. On the other hand, reasons why parents should acknowledge deviant children more than nondeviants, or why these effects should appear in phase 2 but not in phase 1, are not obvious. The one univariate main effect significant at the .01 level was for sex of parent on parent Q(Q)--pure questions--in phase 1. Means were 27.9o7o for fathers and 18.5070 for mothers, F(1,50) = 7.46. This effect deserves mention because Q(Q) was the most common parent mode. It fits common stereotypes that fathers might adopt a more interrogatory

238

Stiles and White

style to reach agreement with their children. However, in the absence of a significant multivariate effect or of any other clear evidence of sex role differences in this task, the finding is best treated as tentative.

DISCUSSION The VRM profiles showed large, clear differences between parent and child roles in both task phases, and clear (though smaller) differences between the "reach agreement" and "tell how you feel" phase in each role's VRM use. On the other hand, the VRM measures failed to clearly distinguish among the interactions of families of aggressive versus withdrawn versus nondeviant children or between males and females in either role. The results are thus consistent with the three-channel hypothesis (Russell & Stiles, 1979)--that measures based on intersubjective categories such as VRMs are primarily sensitive to roles, tasks, and relationships rather than to enduring traits of the speaker or other, such as a tendency to be aggressive or withdrawn. In both phases, the parent role included a preponderance of questions, as well as a substantial number of interpretations, reflections, and advisements directed at their children. The child role consisted mainly of giving information (both personal feelings and objective data) and acknowledging parent utterances. In the "reach agreement" phase, children did most of the explicit agreeing (confirmation intent) in brief utterances such as "yeah" or "uh-huh" (acknowledgment form) or "right" or " o k a y " (interpretation form). Thus, parents gathered information from their child, offered summaries and evaluations, and proposed solutions to the "problems"--which usually concerned the child's chores, rules, or privileges--and the child assented. In the "tell how you feel" phase, this pattern was modified by a substantial increase in the percentage of disclosures (i.e., feeling statements) by both parents and children. Because the VRM system uses the same categories and procedures for coding all kinds of interactions, the profiles in Table II can be directly compared with VRM profiles of other roles and tasks. For example, comparison with VRM profiles of physician-patient interaction (Stiles, Putnam, Wolf, & James, 1979b) shows that physicians, like parents in the present study, used a preponderance of questions and a substantially higher proportion of presumptuous modes (advisements, interpretations, reflection) than their patients, whereas the patients, like children in the present study, used mostly the exposition modes-- D(D), E(D), E(E), and D(E)--and modes with acknowledgment forms (e.g., "yes" and " n o " answers to questions). Comparison of the parent-child profiles with

Parent- Child Interaction in the Lab

239

psychotherapist-client profiles (Stiles & Sultan, 1979) reveals similar parallels. These comparisons begin to quantify the impression that the doctor-patient relationship resembles or recapitulates the parent-child relationship in some ways, independently of the content of the conversations, which varied greatly among these studies. Although content categories were not applied to the present transcripts, members of parent-child dyads appeared (plausibly) to talk about the same things. It seems unlikely that content coding could have distinguished between roles as clearly as intersubjective (VRM) coding. In view of the large number of nonsignificant statistical tests we ran, the relatively small "significant" effects of trait group appear equivocal. (In phase 2 only, parents of withdrawn children used more acknowledgments and parents of aggressive children used more advisements.) Family interaction research seems inevitably to involve many dependent variables and complex, multifactor designs (indeed, these are often recommended by reviewers), which require many tests for significant differences. Probably this approach is justifiable as an exploratory strategy, but it is undoubtedly responsible for a pervasive problem of type I error. The clear differentiation between families with an without deviant children achieved using a simple sociometic self-report measure of family affective structure by Lowman (1980) contrasts with the poor differentiation achieved with intersubjective coding categories in this and previous interaction studies. Content categories might be more sensitive to families' enduring negative attitudes. Or, negative attitudes surrounding a deviant child might influence the affective tone of family interactions as conveyed in the extralinguistic channel--by tone of voice, hesitations, blocking, and so forth. Notably, the best (albeit inconsistent) reported differentiations of interaction in disturbed and normal families have been with rating measures of affect (as opposed to coding measures), such as "hostility," "antagonism," or "warmth" (Jacob, 1975). Of course, it is always possible that different subjects, a different task (see discussions of task selection by Bell, 1964; Hetherington & Martin, 1979), a larger sample, or a different analysis would have demonstrated clearer trait group effects. This study's deviant children were drawn from normal classrooms; a more severly disturbed sample might have shown larger differences. The setting was unfamiliar to the subjects and the interactions were brief; a more naturalistic setting (cf. Patterson, 1976; Reid, 1978) or longer interactions might have yielded stronger effects. The number of subjects in the deviant groups may have been too small to detect subtle effects, particularly if pathological relationships are manifestated differently depending on parent's or child's sex. Or perhaps some sort of Markov chain analysis of sequences and contingencies of coding categories (cf. Brent & Sykes, 1979; Gottman,

240

Stiles and White

1979; Raush, 1972) would discriminate among trait groups. (This analysis would require either a much larger interaction sample or drastic collapsing across VRM categories to obtain the necessary frequencies of category cooccurrences.) The weakness of the trait group discriminations does not seem attributable to a general insensitivity of the VRM system, since the VRM codes clearly discriminated between roles and between task phases. In conjunction with the very low "test-retest reliability," this suggests that subjects' VRM use may be construed as "sampling" from a common prototypical VRM profile for a particular task and role, and that families' ability to engage in role-appropriate interactions under laboratory conditions may not be a sensitive indicator of child behavior pathology.

REFERENCES Bell, R. Q. Structuring parent-child interaction situations for direct observation. Child Development, 1964, 35, 1009-1020. Brent, E. E. & Sykes, R. E. A mathematical model of symbolic interaction between police and suspects. Behavioral Science, 1979, 24, 388-402. Cansler, D. C., & Stiles, W. B. Relative status and interpersonal presumptuousness. Journal of Experimental Social Psychology, in press. Frank, G. The role of the family in the development of psychopathology. Psychological Bulletin, 1965, 64, 191-205. Gottman, J. M. Marital interaction: Experimental investigations. New York: Academic Press, 1979. Hetherington, E. M., & Martin, B. Family interaction. In H. C. Quay & J. S. Werry (Eds.), Psychopathological disorders of childhood (2nd ed.) New York: Wiley, 1979. Jacob, T. Family interaction in disturbed and normal families: A methodological and substantive review. Psychological Bulletin, 1975, 82, 33-65. Lowman, J. Measurement of family affective structure. Journal of Personality Assessment, 1980, 44, 130-141. Martin, B., & Hetherington, E. M. Family interaction and aggression, withdrawal, and nondeviancy in children. Terminal Progress Report on Grant No. MH12474, National Institute of Mental Health, 1971. Patterson, G. R. The aggressive child: Victim and architect of a coercive system. In E. Mash, L. Hamerlynck, & L. Handy (Eds.), Behavior modification and families. L Theory and Research. New York: Brunner/Mazel, 1976. Peterson, D. R. Behavior problems of middle childhood. Journal of Consulting Psychology, 1961, 25, 205-209. Raush, H. L. Process and change--A Markov model for interaction. Family Process, 1972, 11, 275-298. Reid, J. B. A social learning approach to family interaction. Vol. IL A manual for coding family interactions. Eugene, Oregon: Castalia Press, 1978. Riskin, J., & Faunce, E. An evaluative review of family interaction research. Family Process, 1972, 11, 365-456. Russell, R. L., & Stiles, W. B. Categories for classifying language in psychotherapy. Psychological Bulletin, 1979, 86, 404-419. Stiles, W. B. Manual for a taxonomy of verbal response modes: Chapel Hill: Institute for Research in Social Science, University of North Carolina at Chapel Hill, 1978. (a)

Parent-Child Interaction in the Lab

241

Stiles, W. B. Verbal response modes and dimensions of interpersonal roles: A method of discourse analysis. Journal of Personality and Social Psychology, 1978, 36, 693-703. (b) Stiles, W. B., Putnam, S. M., James, S. A., & Wolf, M. H. Dimensions of patient and physician roles in medical screening interviews. Social Science and Medicine, 1979, 13A, 335-341. Stiles, W. B., Putnam, S. M., Wolf, M. H., & James, S. A. Interaction exchange structure and patient satisfaction with medical interviews. Medical Care, 1979, 17, 667-681. (a) Stiles, W. B., Putnam, S. M., Wolf, M. H., & James, S. A. Verbal response mode profiles of patients and physicians in medical screening interviews. Journal of Medical Education, 1979, 54, 81-89, (b) Stiles, W. B., & Sultan, F. E. Verbal response mode use by clients in psychotherapy. Journal of Consulting and Clinical Psychology, 1979, 47, 611-613. Stiles, W. B., Waszak, C. S., & Barton, L. R. Professorial presumptuousness in verbal interactions with university students. Journal of Experimental Social Psychology, 1979, 15, 158-169. Wiggins, J. S. , & Winder, C. L. The peer nomination inventory: An empirically derived sociometric measure of adjustment in preadolescent boys. Psychological Reports, 1961, 9, 643-677.