Psychological Reports, 1990, 66, 675-688. @ Psychological Reports 1990
FAMILY INTERACTION A N D CHILD OUTCOME I N A HIGH-RISK SAMPLE',' DEBORAH F. GREENWALD
Northeastern Universit)r Summary.-This study hypothesized that the quality of family interaction would predict longitudinally children's follow-up functioning. The subjects were 97 boys, functioning on the average within the normal range but considered to be at risk for psychopathology because one parent had previously been hospitalized for psychiatric illness. Family measures of deviant and healthy communication and of activity, balance, and warmth related significantly to children's competent social/ernotional/cognitive functioning at a 3-yr. follow-up. The findings suggest the importance of considering multiple predictors of outcome for children and of employing clinical interventions that attend to more than one aspect of family functioning.
The focus of the present study was the identification of family communication and interaction factors that predict children's cognitive/social/emotional functioning over time. It constitutes part of the University of Rochester Child and Family Study (URCAFS; Wynne, 1984), a longitudinal investigation of multiple aspects of child and parent functioning and family interaction in those intact Caucasian families where at least one parent has been hospitalized for psychiatric disorder. These families were studied to assess factors presumed to create developmental risks for psychological disorder in offspring. In addition, the URCAFS project focused upon those features of parents', families', and children's functioning that are expected to predict children's competent behaviors, which should provide some protection against the risk factors. Included were measures of parents' diagnosis, symptomatology and level of functioning, of family interaction and communication, of physiological responsiveness and attention, and of children's functioning in a variety of situations. Communication and other aspects of interaction in families are hypothesized to have broad effects on the development of children's cornpecence and adaptation. Communication that consistently transmits clear meaning would be expected to help children develop cognitive competencies such as accurate reality-testing, a sense of the stability and predictability of the surrounding world, the ability to focus attention and to integrate new information. Also, 'This research was supported by NIMH Program Project Grant MH-22836, Lyman C. Wynne, Principal Investigator. In addition to the author, other investigatorj on the teams collaborating in this study were Manhal Al-Khayyal, Alfred Baldwin, Clara Baldwin, Robert Cole, Jeri Doane, Lawrence Fisher, Fredric Jones, James E. Jones, Stephen Munson, Lyman C. Wynne, and ?mela Yu. Re uests for reprints should be sent to Deborah F. Greenwald, De artment of Counseling l'sy~ology, Rehabilitation and Special Education, 203 Lake Hall, ~ o r t l e a s t e r nUniversity, 360 Huntington Avenue, Boston, MA 02115.
the same communication factors are expected to affect the social and emotional functioning of the children as well. The ability to comprehend and to predict the environment, including parental messages, is likely to foster selfconfidence, initiative, and a sense of well-being. Such feelings, gained in interaction with close family members, are expected to facilitate successful social skills and mutually gratifying interactions with others. Other elements of family interaction theoretically presumed to affect optimal development of the child include the family affective environment (warmth versus hostility), level of involvement among family members, and the distribution of power and influence. The current study investigated the hypothesis that these characteristics of families would predict children's outcome functioning from initial assessment at Time 1 to functioning at a 3-yr. follow-up (Time 2). The chddren studied were boys presumed to be "at risk" for present and future psychopathology because at least one parent showed severe pathology. Previous research on psychopathological samples provides evidence that disturbed communication in families is associated with an increased likelihood of severe psychopathological outcome for troubled adolescents over a 15-yr. period (Albers, Doane, & Mintz, 1986; Doane, West, Goldstein, Rodnick, & Jones, 1981; Goldstein, 1987; Goldstein, Rodnick, Jones, McPherson, & West, 1978). Comparing families with normal and schizophrenic adult offspring, Mishler and Waxler (1968) and Singer, Wynne, and Toohey (1978) reported differences in communication styles between the two groups. While the association between family communication style and the presence of psychopathology, especially schizophrenia and schizophrenia spectrum disorders, has been supported by these studies, the extent of this connection is not clear. If communication style is seen as either facilitating or hindering the development of clear thought processes, its association with schizophrenia is readily understandable. I t may well be, however, that deviance in communication has a more pervasive effect and is associated not only with severe thought disorder but also with milder forms of disturbance. This idea is consistent with Singer, et al.'s (1978) findings that, when both parents showed disturbed communication, they were likely to produce a schizophrenic offspring. When only one parent showed disturbed communication, the offspring was likely to show a borderline syndrome rather than schizophrenia. Communication may be seen as vital to the development of thought processes, so that poor family communication would teach children to think in disturbed ways (cf. Bateson, Jackson, Haley, & Weakland, 1956). I t may also be that disturbed communication serves as a general inhibitor of the personality processes involved in psychosocial adaptation in general and leads to lowered quality of social/emotional functioning.
FAMILY INTERACTION AND OUTCOME
A substantial research literature on disturbed communication in families has focused on deviant communication as a measure of unclear communication, especially the extent to which there is a departure from an optimal shared focus of attention and clarity of meaning. This measure, developed originally for use with individuals (Jones, Rodnick, Goldstein, McPherson, & West, 1977; Singer, et al., 1978; Wynne, Singer, Bartko, & Toohey, 1977) has also been used in family Consensus Rorschachs (Doane, 1977). This procedure involves a family's attempts to agree on Rorschach percepts. Arbitrary, unfocused, nonacknowledging, and incompletely consensual decisions made during the procedure indicate the presence of communication problems. Deviance in communication has been observed in the speech of parents of schizophrenics (Singer, et al., 1978) and parents' deviant communication has correlated with the presence of schizophrenia spectrum disorders at a 5-yr. follow-up (Goldstein, et a / . , 1978; Doane, et al., 198 1). Time 1 assessments with the current (URCAFS) longitudinal sample (Jones, Wynne, Al-Khayyal, Doane, Ritzler, Singer, & Fisher, 1984) have shown that parents' deviant communication, assessed from the Family Consensus Rorschach procedure, is significantly correlated with children's adaptive functioning, measured cross-sectionally by peers' and teachers' blind ratings of the index children. Al-Khayyal (1980) developed a measure of healthy communication (HC) from the consensus Rorschach that was an index of the extent to which verbal interaction clarifies meaning and enhances a shared focus of attention. Healthy communication, in a composite measure that included other communication variables, was significantly related at Time 1 to children's functioning in URCAFS 10-yr.-olds (Wynne, 1984), as assessed by teachers and peers. I n addition to the families' communication style, other characteristics of interaction among family members were also hypothesized to predict children's functioning over time. The affective relationship between parents and children was one of these attributes. Previous research (Hooley, 1987; Vaughn & Leff, 1976) noted that an affective response of high criticism and overinvolvement directed at the patient by a close family member was related to relapse rates for schizophrenics and depressives. These variables also correlated with the development of signs of schizophrenia spectrum disorders at a 5-yr. follow-up of a group of disturbed, nonpsychotic adolescent boys (Doane, et al., 1981). I n the URCAFS study the affective domain was assessed by global measures of warmth expressed in family interaction (Cole, Al-Khayyal, Baldwin, Baldwin, Fisher, & Wynne, 1984). The over-all amount of interaction and the balance of participation among family members were also examined. High levels of interaction were
D. F, GREENWALD
expected to reflect positive family involvement, while balanced interaction was seen as indicating a family environment which had neither highly dominating nor highly withdrawn members. Both activity level and balance of interaction were included in a composite measure together with warmth (activity/balance/warmth). High activity/balance/warmth~relatedat Time 1 to children's competence in the school setting (Baldwin, Baldwin, & Cole, 1982; Cole, et al., 1984). A previous follow-up study (Greenwald, 1989) reported significant correlations between these family interaction measures and children's outcome, as measured by intelligence. The current study expands the examination of the predictions of family variables over time to a multidimensional perspective of competence, including four additional indices of chddren's outcome. These were measures of children's adaptive functioning drawn from the observations of clinicians, parents, school peers, and classroom teachers.
Hypotheses The expectations of the current study were that family interaction at the time of the initial assessment would predict children's outcome as measured by level of functioning at the 3-yr. follow-up. Specific hypotheses were that measures of fathers' and mothers' deviance in communication and healthy communication, composite parents' communication and activity/balancelwarmth from the Free Play and Consensus Rorschach would predict outcome as assessed by clinicians, parents, peers, and teachers.
Subjects Intact families with one parent previously hospitalized for functional psychiatric disorder at either the university or the state hospital in Monroe County, New York, were recruited for the URCAFS project. Individuals with organic brain damage, IQs lower than 80, or severe substance abuse were excluded. The initial assessment was conducted an average of 3.9 yr. after the most recent hospitalnation of the index parent. Following a convergence design (Bell, 1753), researchers conducted the initial assessment when the index offspring were 4, 7, or 10 yr. old, with a follow-up 3 yr. later. This design potentidy permits the examination of developmental trends at four age levels (ages 4 to 13). Subjects who were four at the time of the initial assessment would be seven at follow-up and could be compared with the subjects who were seven at initial assessment. The subjects were 77 boys who comprised the follow-up cohort of the original sample of 145 boys who were studied at the time of initial contact. There were 25 4-yr.-olds, 34 7-yr.-olds, and 38 10-yr.-olds in thls cohort, mostly from middle-class families [M = Class I11 by Hohngshead (1957) two-factor index] and above average in IQ (M = 111.9).
FAMILY INTERACTION AND OUTCOME
Parents' DSM-I11 (American Psychiatric Association, 1980) diagnoses were made from a consensus of two diagnostic teams, consisting of experienced psychiatrists and clinical psychologists, working with data from semistructured interviews (World Health Organization, 1973) focusing on symptomatology, psychiatric history, and social data, systematically collected case history data (Strauss & Harder, 1981), spouses' descriptions of the patient-parent's behavior from the Katz Adjustment Scales (Katz & Lyerly, 1963), and videotaped interviews of each couple (Wynne, 1987). The diagnoses were as follows: schizophrenia, n = 14; schizophreniform, n = 2; bipolar psychosis, n = 19; schizoaffective psychosis, n = 4; unipolar depression, with psychotic features, n = 10; nonpsychotic depression, n = 17; nonpsychotic bipolar disorder, n = 8; paranoid, schizoid, and schizotypal personality disorders, n = 7; other personality disorders, n = 11; nonaffective, nonpsychotic disorders, n = 4, and atypical psychosis, n = 1. The average Global Assessment Scale (Endicott, Spitzer, Fleiss, & Cohen, 1976) score at the time of the study for these index parents was 60.7, with means that ranged from 52 for the schizophrenics to 68 for the milder nonpsychotic depressions. The parents were functioning, on average, at a level just slightly above that of obvious neurotic pathology. For the spouses the mean Global Assessment score on the index parentslpatients at the time of the study was 68.0. Interrater reliabihty of the two clinicians doing most of the GAS ratings was .78, consistent with reliability reports of other studies. Procedure Two experimental procedures, the Consensus Rorschach (Loveland, Wynne, & Singer, 1963; Wynne, 1987), a highly structured task, and the Free Play (Cole, et al., 1984), an unstructured task, were used to derive measures of family interaction. The Consensus Rorschach involved having the entire family, consisting of both parents and all children four years or over, come to agreement on percepts of the Rorschach ink blots. The transcripts of these sessions were coded for deviant communication and for healthy communication of both mother and father. The Free Play (Baldwin, et al., 1984) consisted of a 30-min. period during which each family was asked to play together as they would at home. The procedure was conducted in a large room equipped with chairs, toys, and games, and a one-way mirror through which the family was observed by trained coders. A composite cross-situational measure of parental communication was also created (Jones, et al., 1984) by using averages of individual fathers' and mothers' deviant communication scores from the Consensus Rorschach procedure, individually administered Rorschachs, and individually administered TATS, along with healthy communication scores from the Family Rorschach. These four averages were each assigned a + 1, 0, or -1, depending on whether they fell a half standard deviation above the URCAFS sample mean, within the middle of the distri-
D. F. GREENWALD
bution, or a half standard deviation below. This process produced a ninepoint variable ranging from + 4 to -4. For the measures of activity/balance/warmth, the family was rated according to the extent to which members were actively involved in the interaction, showed balance in the proportion of times any individual initiated spoken interchange, and showed warmth toward each other compared to the sample mean. These measures were derived from the Consensus Rorschach and the Free Play. The outcome variables used in the current study were selected both for their conceptual relevance to the notions of adaptive functioning outlined above and also for their significant or near-significant relationships at Time 1 with measures of clddren's functioning and/or parents' pathology. These included: two over-all indices of adaptive functioning, one based on clinicians' direct observation of the child (Munson, Baldwin, Yu, Bddwin, & Greenwald, 1984; Yu, Prentky, Fisher, Baldwin, Greenwald, Munson, & Baldwin, 1984), and one based on parent reports to clinicians (Jones, 1977; Yu, et al., 1984); and peers' mean and teachers' mean ratings of each child (Fisher, Harder, & Kokes, 1980; Harder, Kokes, Fisher, & Strauss, 1980). The ratings of peers and teachers were done at school, while the other two procedures were conducted at the university hospital where the research teams were based. All outcome variables were provided by teams that were blind to the measures of family interaction. Each of the two clinical ratings, an over-all measure of adaptive functioning for each child, was provided by an independent research team. One team based its assessments on data provided by the parents in a structured interview, the Rochester Adaptive Behavior Interview or RABI (Jones, 1977). This interview covered each child's social, familial and school functioning, as well as areas of possible symptomatology or pathology. The other over-all measure, COPE, the consensus rating of two or more clinicians, was based on: (I.) videotapes of semistructured interviews with the child, conducted by an experienced child psychiatrist or psychologist and focused on peer and family relationships, school, self-concept and inner life (for the 4-yr.-olds, the interview was a semistructured playroom interview); (2) audio tapes of interviews with the parents, conducted by the clinician who interviewed their child; and (3) a psychological assessment of the child (WISC, TAT, Rorschach, DAP), administered by a child psychologist who was not one of the raters. The interviewers themselves rated the children they interviewed. Interrater agreement on level of C O P E was .90 (Munson, et al., 1984). Assessments in the schools provided teachers' and peers' measures of the index child's functioning (Fisher, 1980). Each child was rated, dong with the other members of his class by classmates and teachers in such a way that
FAMILY INTERACTION AND OUTCOME
the index child was not identified, and raters were not aware that the study focused on a particular child. A factor analysis of peers' and teachers' statements reflecting competent social and cognitive functioning of children in the school setting produced such factors as brightness-compliance, intrusiveness, friendhness, and intellectual dullness (for peers) and cognitive competence, motivation, social competence and compliance (for teachers). These factors were derived and cross-validated prior to the study (Fisher, Schwartzman, Harder, & Kokes, 1984). An averaging of the peers' ratings and of the teachers' ratings created summary variables, which were used in this study. These school measures provide an independent, blind assessment of the index child by two sets of people familiar with him in a context separate from parental psychopathology. I n addition, this method of observation provided a normal control group with whom to compare the "at-risk" sample. The school functioning scores for the at-risk group did not differ significantly from those of their classmates.
RESULTS To test the hypothesis that parental communication deviance, healthy communication, and activity/balance/warmth would correlate with children's TABLE 1 CORRELATIONS OF FANILYINTERACTION VARIABLES AND CI~ILDREN'S OUTCOME AT 3-YR.FOLLOW-UP Interaction Predictors COPE Deviant Communication Fathers
Ou tcorne Variables RABI M Rating Peers Teachers
.23* 79 .24* 79
-.I8 92 -.23* 92
-.34$ 85 -.29$ 85
Healthy Communication Fathers
,353 64 .27* 64
67 -.26* 71 -.26t 91
f Free Play, A/B/W
-.24 54 -.41$ 59 -.27t 79
.34$ 60 .35$ 64 .23' 86
D. F, GREENWALD
social-emotional and cognitive functioning at a 3-yr. follow-up, bivariate correlates were calculated between the family interaction variables and children's outcome measures. These data are presented in Table 1. The results of a previous study (Greenwald, 1989), in which these same predictors were used with IQ as the outcome measure, are included for comparative purposes. TABLE 2 OF OUTCOME VARLABLES SIGNIFICANT MULTIPLEREGRESSION PREDICTORS Interaction Predictors COPE Deviant Communication Fathers
Outcome Variables M Rating Peers 'Teachers
f Healthy Communication Fathers
f Composite r
f Rorschach, A/B/W
f Free Play, A/BjW
.24* 1,56 .24*
Five of the seven predictors correlated significantly and in the predicted direction with at least one of the four outcome variables (other than IQ). The activity/balance/warmth measure from the Rorschach procedure was the only predictor significantly related to all four outcome measures. Activity/balance/warmth based on the family Free Play, both fathers' and mothers' deviant communication, and the family's composite communication measure all correlated significantly with at least two outcome measures. The two measures of healthy communication, however, were not significantly associated with any of the outcome indices focused on in this study. The correlations were all modest, ranging in absolute magnitude from .23 to .41. To examine the overlap of the predictors in pointing towards particular outcomes and to assess the best combination of predictors for each outcome measure, a series of stepwise multiple regressions were performed, with the
FAMILY INTERACTION AND OUTCOME
seven predictors as the independent variables and each of the four outcome variables, in turn, as a dependent variable. Once again, IQ was included in this analysis for comparative purposes. Only regression equations in which the individual predictors showed p values of .05 or less were considered significant; see Table 2. With COPE as the outcome measure, only composite communication predicted independently (P = -.28, F,,,, = 5.35, p < .05). When the score on the Rochester Adaptive Behavior Inventory provided the outcome measure, Rorschach activity/balance/warmth was the only variable predicting independently and significantly (0 = -.36, F,,,, = 7.87, p < .01). For the peers' mean rating composite communication was the only significant independent predictor (P = .30, F,,,, = 6.22, p < .02). No predictor appeared to be related independently at the .05 level or better to the teachers' mean rating. Only 65 cases were left in the analysis for the teachers' rating, as cases were deleted for the multiple regressions when any data were missing. For this reason the Beta of -.22 for mothers' communication deviance showed a significance level of .09, while a bivariate correlation of similar magnitude (-.24) had been significant at the .05 level. With WISC IQ as an outcome criterion, three predictors contributed significantly and independently to the multiple regression equation: fathers' deviant communication (0 = -.26, F,,,, = 4.83, p < .05), activity/balance/warmth from the Free Play (0 = .24, F,,,, = 4.28, p < .05), and activity/balance/warmth from the family Consensus Rorschach (P = .24, F,,,, = 4.03, p < .05).
DISCUSSION The results of the current study demonstrate that f a d y interaction factors predict significantly to the son's follow-up functioning, as assessed by a broad array of outcome measures across different situations. Independent variables, which measured multiple aspects of functioning, were assessed by a broad array of outcome measures across different situations. Although the word prediction is used, the correlations between Time 1 and Time 2 variables do not necessarily indicate causality, as it may be that both sets of variables are the product of a third variable or set of variables. One such possibility would be severity of parental disorder. However, in this sample psychotic-nonpsychotic status of parents at Time 1 did not predict child functioning. Another possible variable would be the sex of the patient-parent, but this variable was unrelated to outcome. I t should be noted that this sample was composed of Caucasian boys from largely middle-class families, who were functioning at a slightly above-average intellectual level. Generalizability of the findings may therefore be restricted. Independent variables which assessed multiple aspects of family interaction were the most successful predictors. The measure of activity/balancelwarmth from the Consensus Rorschach was the best predictor of
D. F. GREENWALD
outcome and the only one to relate significantly to aU four outcome measures. Composite communication, covering both healthy and deviant communication, also was a good predictor, relating significantly to three of four outcome measures. Similarly, it proved important to measure analogous aspects of family interaction in more than one setting. For example, the two activity/balancelwarmth measures each predicted significantly the children's outcomes, though they did not correlate significantly ( r = . l 3 ) with each other. This was probably because the procedures from which each was obtained were different. Together, the results noted here underline the importance of multidimensional assessments of family functioning, which focus on several aspects of interaction across a variety of settings. The multiple regression analyses, with only one useful independent predictor for each outcome variable (except in the case of IQ), might be seen as indicating that one variable is ordinarily sufficient for maximal prediction. However, different interaction measures emerged as the best single predictors for different measures of outcome, which also supports the need for multiple predictors in research concerning high risk. The two healthy communication variables did not predict to any of the outcome measures in the current study, though both predicted I Q s~gnificantly in this same sample (Greenwald, 1989) and have shown significant relationships with cross-sectional measures of children's functioning (AlKhayyal, 1780). It is not clear from these data whether this relatively new construct is really a less useful predictor or whether the way in which it is measured needs to be refined. It might be questioned whether assessments of children's functioning based on parental reports, vulnerable as they might be to distortion, would be valid. The Rochester Adaptive Behavior Inventory, however, which relied on parental responses, was as highly correlated with the predictor variables as any other outcome measure (except IQ), supporting the validity of parental reports. In the current study at most four independent variables predicted any one outcome measure. However, when I Q had previously been the measure of outcome, all seven variables had shown significant correlations. One likely reason for this is that I Q is an objective measure, with excellent reliability. If so, then the significance of these results is probably not that family interaction predicts best to intellectual capacity, which I Q purports to measure, but rather that family functioning relates to competent, adaptive functioning in children, of which I Q is a highly reliable indicator. The present longitudinal study is consistent with the perspective that an interaction between family factors and children's vulnerability is important in . with psychopathothe development of children's s y m p t o m a t ~ l o g ~Research
FAMILY INTERACTION AND OUTCOME
logical populations has provided evidence for such linkages with offspring already showing signs of disturbance. This research extends such findings to offspring who are not clinically disturbed. While the subjects in the current study are presumed to carry vulnerabilities for future disturbance, at both initial and follow-up assessments these boys scored within the average range on those variables for which norms were available (peers' and teachers' ratings and IQ). The findings of this study suggest, then, that the longitudinal impact of family interaction factors on children's competence are relevant for chddren's nonpathological functioning as well. The relatively modest correlations obtained in the current study may reflect the fact that the majority of the sample of boys indeed is "normal." Had some of the boys been demonstrating overt indications of disturbance and consequently greater variability on the outcome measures, family interaction factors might have shown more predictive power. These findings are consistent with those in the developmental literature, in which family factors such as warmth, consistent discipline, and parental involvement (Baurnrind, 1971; Parker, Tupling, & Brown, 1979; Watson, 1957) have been correlated with children's functioning. I n this way, this study forms a link between research on the effects of family interaction in pathological samples and research on normal developmental patterns. Communication problems have typically been seen as relating to thought disorder and have not been much investigated in nonpathological samples. The result presented here that deviant communication relates to children's competence in boys who are currently functioning on a par with other boys their age suggests broader effects not limited to thinking difficulties. That is, communication, like other farmly-interaction factors, affects how the child grows and develops in multiple areas of social/emotional and cognitive functioning. There are several ways in which family interaction can be viewed as contributing to the formation of pathology. I t may be that family factors affect the adaptive functioning of all children and play a part in determining the extent to which development will proceed optimally, or not. From this perspective, pathological development is seen as being on a continuum from good to poor to pathological adaptation. Alternatively, pathology may be viewed as distinct from a lack of adaptive functioning and as occurring in those individuals who are particularly vulnerable, whether because of inborn factors or because of poor adaptive competence. In this view, vulnerable children who develop in the context of a family with interaction that inhibits, or fails to facilitate, optimal development, will be likely to develop pathology, much as a child who is malnourished or has a defective immune system will be more vulnerable to
D. F. GREENWALD
physical diseases. High-risk children who, on the contrary, are in families with an interaction style conducive to the optimal development of competence may be able to compensate for their vulnerability. In terms of therapeutic interventions, these findings support the need for family treatment when a child is showing signs of disturbance. In addition, such treatment will d o best to address multiple aspects of family interaction, such as the family's level of mutual involvement and affective warmth, as well as the clear sharing of attentional focus and of meaning. REFERENCES ALBERS,L. J., DOANE,J. A,, & MINTZ, (1986) Social competence and family environment: 15-yeu follow-up of disturbed ad!descents. Family Process, 25, 379-389, AL-KHAWAL,M. (1980) Healthy parental commuaication as a predictor of child competence in families with a schizophrenic and psychiatrically disturbed nonschizophrenic parent. Unpublished doctoral dissertation, Univer. of Rochester. ~ M E I U C A N PSYCHIATRIC ASSOCIATION. (1980) The diagnostic and statistical manz~alof mental disorders. (3rd ed.) Washington, DC: Author. BALDWIN, A. L., BALDWIN,C. I?, & COLE,R. E. (1982) Family free play interaction: setting and methods. In Parental athology, family interaction, and the competence of the child i n school. Monographs o r t h e SocieQ for Research in Child Development, 47, No. 5, 36-44. (Serial No. 197) BATESON,G., JACKSON, D. D., HALEY,J., & WEAKLAND, J. (1956) Toward a theory of schizophrenia. Behavioral Science, 1, 251-264. BAUMRIND, D. (1971) Current patterns of parental authority. Developmental Psychology Monographs, No. 1, 1-103. BELL,R. Q. (1953) Convergence: an accelerated longitudinal approach. Child Development, 24, 145-152. COLE,R., AL-KFIAWAL, M., BALDWIN, A,, BALDWIN, C., FISI-IER,L., & WYNNE,L. (1984) A cross-setting assessment of family interaction and the prediction of school competence in children at risk. In N. F. Watt, E. J. Anthony, L. C. Wynne, & J. Rolf (Eds.), Children at risk for schizophrenia: a longitudinal perspective. New York: Cambridge Univer. Press. Pp. 388-392. DOANE,J. A. (1977) Parental communication deviance as a predictor of child competence in families with a schizo hrenic and nonschizophrenic parent. Unpublished doctoral dissertation, Univer. of ~ o c f e s t e r . DOANE, J. A,, WEST, K. L., GOLDSTEIN, M. J., RODNICK, E. H., &JONES,J. E. (1981) Parental communication deviance and affective style: predictors of subsequent schizophrenia spectrum disorders in vulnerable adolescents. Archives of General Psychiatry, 38, 679-685. ENDICOTT, J., SPITZER,R. L., FLEISS,J. L., & COHEN,J. (1976) The global assessment scale: a procedure for measuring overall severity of psychiatric disturbance. Archives of General Psychiatry, 33, 766-771. FISHER,L. (1980) Child competence and psychiatric risk: 1. Model and method. ]ournal of Nervous and Mental Disease, 168, 323-331. FISHER,L.,HARDER, D. W., & KOKES,R. F. (1980) Child competence and psychiatric risk: 3. Comparisons based on diagnosis of hospitalized parent. Jottrnal of Nervous and Mental Disease, 168, 353-355. FISHER,L., SCHWARTZMAN, P., HARDER,D. W., & KOKES,R. F. (1984) A strategy and methodology for assessing school competence in high-risk children. In N. F. Watt, E. J. Anthony, L. C. Wynne, & J. E. Rolf (Eds.), Children at risk for schizophrenia. Cambridge, Eng.: Cambridge Univer. Press. GOLDSTEIN, M. (1987) Family interaction patterns that antedate the onset of schizophrenia and related disorders: a further analysis of data from a longitudinal, rospective smdy. In K. Halweg & M. Goldstein (Eds.), Understanding mental disorLrs. New York: Family Process Press.
D. F. GREENWALD
child competence. In K. Hahlweg & M. J. Goldstein (Eds.), Understanding major mental disorder: the contribution o//amily interaction research. New York: Family Process Press. Pp. 57-73. WYNNE,L. C., SINGER, M. T., BARTKO, J. J., & TOOHEY,M. L. (1977) Schizophrenics and their families: recent research on parent communication. In J. M. Tanner (Ed.), Developments in psychiatric research. London: Hodder & Scoughton. Pp. 254-286. Yu, P., PRENTKY,R., FISHER,L., BALDWIN, A. L., GREENWMD,D. F., MUNSON,S., & BALDWIN,C. P. (1984) Child competence as assessed by clinicians, parents, teachers, and peers. In N . F. Watt, E. J. Anthony, L. C. Wyme, & J. E. Rolf (Eds.), Children at risk for schizophrenia: a longitudinal perspective. New York: Cambridge Univer. Press. Pp. 365-370. Accepted March 7 , 1990.