Sunteți pe pagina 1din 12

Journal o f Abnormal Child Psychology, Vol. 14, No. 2, 1986, pp.


Indirect Influence of Maternal Social

Contacts on Mother-Child Interactions:
A Setting Event Analysis 1
Jean E. D u m a s 2
University o f Western Ontario

Fourteen mother-child dyads who had sought psychological help for severe
interaction problems took part in a study that investigated the relationship
between maternal social interactions with adults outside the family and
mother-child interactions in the home. Social interactions outside the family were based on maternal self-reports; mother-child interactions in the home
were based on direct observations and included both base-rate and sequential measures. Results indicated that mothers were significantly more aversive toward their children on days in which they had themselves experienced
a high proportion ofaversive interactions with adults than on days in which
they had not. This higher level of aversiveness was evident in their responses
to both aversive and nonaversive child behavior and could not be attributed
to any corresponding change in child behavior. Implications for research and
clinical practice are discussed.

Children develop within a social system, usually the family, that is only a
part of a larger network of social systems. In keeping with an ecological
perspective that stems from the theoretical and applied work of researchers
such as Barker (1968) and Bronfenbrenner (1979), it is now commonly ac-

Manuscript received in final form May 12, 1985.

~This work appeared first as a paper presented at the 92nd Annual Convention of the American
Psychological Association, Toronto, August 1984. The research data reported in this article
were generated by support from the National Institute of Mental Health, Crime and Delinquency Section, Grant No. R01-1068-58, to Robert G. Wahler. His permission to use these
data is gratefully acknowledged.
2Address all correspondence to Jean E. Dumas, Department of Psychology, University of Western
Ontario, London, Ontario, Canada N6A 5C2.

0091-0627/86/0600-0205505.00/09 1986PlenumPublishingCorporation



cepted that a proper understanding of the socialization process requires a

comprehensive analysis of interactions both within and between these systems
(e.g., Hartup, 1978; Lamb, 1977; McCall, 1977; Rutter, 1983).
This ecological perspective has influenced research in two complementary directions. At the two-person or dyadic level, it has led to the direct
observation of family members as they interact with one another in pairs,
with the aim not only of better understanding the process of reciprocal influence in family functioning (Burgess & Conger, 1978; Lobitz & Johnson,
1975; Mash & Johnston, 1982) but also of modifying it when it can be considered deviant (Barkley, 1981; Patterson, 1980; Wahler, 1980). Recent work
has made increasing use of complex sequential analyses of social interaction
in an attempt to disentangle this influence process (Dumas, 1984a; Dumas
& Wahler, 1985; Patterson, 1976; Reid, Taplin, & Lorber, 1981; Snyder, 1977;
Wahler & Dumas, in press a). This methodology provides a significant improvement over more traditional methods of analysis, which typically measure
differences in specific behaviors within individuals and use these differences
to draw conclusions about the nature of the relationship between individuals,
thus centering "on the actions of individuals who are engaged in social interaction, not on social interaction per se" (Hartup, 1979, p. 27).
At the multiperson level, the ecological perspective has led to an increased awareness of the importance of second-order effects (Bronfenbrenner, 1979) or indirect influences in the development and maintenance of
normal and deviant interaction patterns. In other words, it has led to the
recognition that interactions between two individuals (e.g., parent-child) can
be influenced indirectly by the relationships that each of them has with other
members of the environment, whether these persons are part of the immediate
family (e.g., spouse, sibling) or not (e.g., relative, friend, peer). This process of indirect influence is described by Bronfenbrenner (1979) in the form
of a hypothesis:
The capacity of a dyad to function effectively as a context of development depends
on the existence and nature of other dyadic relationships with third parties. The
developmental potential of the original dyad is enhanced to the extent that each of
these external dyads involves mutually positive feelings and the third parties are supportive of the developmental activities carried on in the original dyad. Conversely,
the developmental potential of the dyad is impaired to the extent that each of the
external dyads involves mutual antagonism or the third parties discourage or interfere
with the developmental activities carried on in the original dyad. (p. 77)

This hypothesis seems particularly appropriate to account for findings such

as Pederson's (cited in Lamb, 1979) that parents who are more critical toward
one another tend to express more negative affect toward their infants, or
Hetherington's (1979) that divorced mothers are better able to deal effectively with their young children when their former spouse accepts their approach to discipline and generally supports them in the task of child rearing.

Indirect Influence


However, although this hypothesis makes intuitive sense, the process of indirect influence it proposes is not readily apparent.
We have suggested elsewhere (Dumas, 1984b; Wahler & Dumas, in press
b) that this process of indirect influence may be understood with reference
to the concept of "setting event" (Kantor, 1959). A setting event may be defined as a set of enviromental circumstances that influence which of a variety
of potential stimulus-response relationships will occur. Given a family dyad,
for example, certain environmental circumstances may act to "set up" a
member of the dyad to act toward the other in a specific manner. Of major
importance in the socialization context is how interactions between two adults
can moderate later interactions between adult and child. In two recent studies,
Wahler (1980) and Wahler and Dumas (in press a) showed that, on days in
which highly disadvantaged mothers of oppositional children reported having engaged in aversive interactions with other adults prior to an observation of their interactions with their children at home, they were more likely
to act aversively toward them during the observation than on days in which
they had not engaged in such interactions. However, both studies were limited,
the first one because it did not use sequential measures of mother-child interaction and the second one because its results were based on six dyads only.
The present study was conducted to investigate further the nature of
the relationship described in these studies. Using mother-child dyads who
experienced serious relationship problems, it focused both on sequential
measures of interaction within the dyads and on maternal self-reports of interaction with adults outside the family. Following Bronfenbrenner's (1979)
hypothesis, it was predicted that mothers would be more aversive toward
their children on days in which they had first experienced high levels of aversive interactions with adults in their community than on days in which they
had not; it was predicted further that this aversive response tendency would
not be attributable to a greater likelihood that their children would then respond to them with increased aversiveness.


Fourteen mother-child dyads were selected from a sample of 52 dyads
that had been referred to a therapeutic program for child management problems. AI! data reported here were obtained prior to any therapeutic intervention. Dyads were selected because the mothers reported frequent aversive
interchanges, not only with their children but also with adults outside their
immediate families. These reports were based on an index of maternal corn-



munity contacts (known as "insularity"). This index was obtained, together

with scores on six measures of mother-child interactions, in the course of
home-based observations. From a total of 156 observations, the insularity
measure was used to select two subsets of observations: those preceded by
aversive maternal community contacts and those not preceded by such contacts. The measures of mother-child interaction were then compared between
these two subsets.

All participants had sought psychological help for child behavior problems by contacting the Child Behavior Institute of the University of Tennessee. Results of a semistructured interview (Wahler & Cormier, 1970)
indicated that one family had referred themselves for treatment while all
others had been referred by schools and social agencies (e.g., Department
of Human Services, courts). Referral problems were similar for each family. All children were described as coercive and oppositional at home and/or
school, while their mothers described themselves as having considerable difficulty enforcing rules and instructions and as often yelling, screaming, or
striking their child.
The same interview provided the following sociodemographic data. At
time of referral, mothers ranged in age from 25 to 35 years (M = 28.86,
SD = 4.28). Eight of their children were boys and six were girls. They
ranged in age from 4 to 10 years (M = 6.36, SD = 2.06). Six families had
yearly incomes below $6,000, 7 families below $13,000, and one family below
$25,000. Seven mothers had not completed high school, six had done so,
and one had started a college education. Eight mothers were married or had
been cohabiting with a common-law spouse for 2 years or more, while six
mothers were single parents.

Measurement Procedures
Dyadic Interactions. Biweekly home observation sessions were established during an initial interview with each mother. They were chosen at times
of day when she was most likely to have interaction problems with her referred child. All observations were conducted by undergraduate students who
had successfully completed a 2-week training course on the use of the Standardized Observation Codes (Wahler, House, & Stambaugh, 1976). This
coding system provides a comprehensive (24 codes) picture of interactions
between the target child and other family members. During each observation session, which lasts 30 minutes, the observer is signaled through ear-

Indirect Influence


phones to observe for 10 seconds and then record code occurrences for the
following 5 seconds on purpose-made forms. A total of 120 intervals of code
occurrences is thus possible for each observation.
Nine of these codes were used in this study. They were grouped into
four behavior clusters, which provided overall measures of mother and child
nonaversive and aversive behavior as follows: (a) Mother Nonaversive
represented all intervals in which either o f two specific codes reflecting the
mother's nonaversive child-directed behaviors were scored: Nonaversive Instruction was scored for any instance of discrete instruction directed to the
child; Nonaversive Social Attention was scored for any instance of neutral
or positive physical or verbal behavior directed to the child. (b) Mother A vetsire represented all intervals in which either of two specific codes reflecting
the mother's aversive child-directed behaviors were scored: Aversive Instruction was scored for any instance of discrete instruction accompanied by aggressive physical or verbal behavior; Aversive Social Attention was scored
for any instance of aggressive physical or verbal behavior. (c) Child Nonaversive represented all intervals in which either of two specific codes reflecting
the child's nonaversive mother-directed behaviors were scored: Compliance
was scored for any instance of compliance with a discrete instruction given
by the mother; Social Approach was scored for any instance of physical or
verbal contact with the mother initiated by the child and not involving any
aversive behavior. (d) Child Aversive represented all intervals in which any
of the three specific codes reflecting the child's aversive mother-directed
behaviors were scored. Opposition was scored for any instance of noncompliance with a discrete instruction; Aversive Opposition was scored for
any instance of noncompliance with a discrete instruction accompanied by
physical aggression a n d / o r verbal protest; Rule Violation was scored for any
instance of violation of an established household rule.
Insularity Index. Following each observation session, the observer also
conducted a brief structured interview with the mother. The interview format, called the Community Interaction Checklist (Wahler, Leske, & Rogers,
1979), prompted maternal recall of all her adult social contacts outside the
family in the preceding 24 hours. Each mother was asked to recall these contacts within the framework of several categories. Two categories were relevant to the purpose of the present s t u d y - namely, the identity of the contact
person (e.g., friend, relative, helping agency representative) and the valence
of the contact for the mother (from + 3 = very positive, through 0 = neutral,
to - 3 = very aversive). On the basis of these data, a mother had to satisfy
at least one of two conditions to be described as "insular": (a) to report, on
average, at least twice as many of her daily contacts with relatives a n d / o r
helping agency representatives as with friends, or (b) to report, on average,
at least a third of all her daily contacts as neutral or aversive (score of 0 to



3). As required by the participant selection criteria, each mother was insular. The average pattern of extrafamily social contacts reported by the 14
mothers was as follows: number of daily contacts (M = 3.88, SD = 1.59,
range: 2-7), percentage of daily contacts with relatives or helping agents (M
= .45, SD = .26, range: .08-.77), percentage of daily contacts with friends
(M = .38, SD = .26, range: .05-.86), and percentage of daily contacts rated
neutral or aversive (M = .52, SD = .30, range: .08-.94). This pattern is
very similar to insular patterns obtained with other samples (e.g., Wahler,
1980; Wahler & Dumas, in press a).

Measurement Reliability
Dyadic Interactions. Extensive data assessing the reliability of the Standardized Observation Codes have been reported elsewhere (Dumas, 1984a).
They were based on analyses conducted with all 52 families from which the
14 dyads who took part in this study were selected. Briefly stated, measures
of session reliability, which were based on intraclass correlation coefficients
(Guilford & Fruchter, 1978; Winer, 1971), were calculated for all base-rate
probabilities; they ranged from .79 to .99. Measures of interval-by-interval
reliability, which were based on the statistic kappa (Hartmann, 1977; Hubert,
1977), were calculated for all conditional probabilities; they ranged from .34
to .86.
Insularity Index. An intraclass correlation coefficient was also computed to evaluate the reliability of the Community Interaction Checklist. This
measure was obtained by applying the insularity decision rules described above
to the data gathered on each of the last five occasions on which the checklist
was administered during preintervention. Thus, each occasion received either
a noninsular or an insular score. The resulting data matrix was subjected
to a repeated-measures analysis of variance. The reliability of the insularity
index was found to be equal to .82 (F(51,208) = 5.54, p < .0001).

Data Analyses
Two types of measures were derived from the observational data: (a)
Base-rateprobabilities of nonaversive and aversive mother and child behaviors
were computed. This was done by grouping all observations for each dyad
separately, adding the number of observation intervals in which each behavior
cluster had been scored, and dividing the results by the total number of intervals. These probabilities were then averaged across dyads. (b) Conditional
probabilities were computed to establish how mothers and children responded
to each other's behavior. This was done by counting the number of observa-

Indirect Influence


tion intervals in which a behavior cluster of interest (e.g., Mother Aversive)

had been scored, but only if it had been preceded by another behavior cluster
o f interest (e.g., Child Aversive) in either of the previous two observation
intervals (i.e., 30 seconds) and dividing the result by the base-rate probability of the second behavior cluster. These probabilities were also Obtained for
each dyad separately and averaged across dyads.

Base-Rate and Conditional Probabilities
Base-rate and conditional probability measures of mother and child
behaviors were computed on three sets of observations: (a) all observations
(n = 156); (b) all observations preceded by maternal self-report of positive
contacts on the Community Interaction Checklist, i.e., less than a third of
all contacts rated as neutral or aversive (n = 76); and (c) all observations
preceded by maternal self-report of aversive contacts on the checklist, i.e.,
more than two-thirds of all contacts rated as neutral or aversive (n = 61).
These results are presented in Table I, where comparison is made between
the two contact conditions. As there were no difference between these conditions for the base-rate and conditional measures of nonaversive behavior,
only the measures of aversive behavior are reported here. Comparisons indicate that the probabilities of child behavior did not differ under the two
contact conditions. The children were not more likely to act in an aversive
Table I. Base-Rate and Conditional Probability Measures of Mother and Child
Aversive Behaviors Under Three Observations Conditions
Observations preceded by



Child aversive
Child aversive/
mother nonaversive
Child aversive/
mother aversive
~p < .05.
bp < .01.








- 1.85"




- 1.83"




-2.35 ~















manner under either condition, whether their behavior was measured as a

base rate or was conditional upon a maternal antecedent. However, the probabilities of mother behavior were found to differ systematically. Mothers
were more likey to act in an aversive manner toward their children when they
had experienced a large proportion of aversive contacts with adults in their
community prior to an observation than when they had not. As their conditional probabilities indicated, this response tendency applied irrespective of
the child behavior antecedent. In other words, the mothers were more likely
to respond aversively to both aversive and nonaversive child behavior when
they had themselves experienced high levels of aversive interactions before
an observation.

Observed Dependencies in Maternal A versiveness

Given this response tendency, the base-rate and conditional probability measures of maternal aversiveness computed on all observations were compared to the same measures computed only on the observations preceded by
"fable II. Diagrams of Observed Dependencies in Mother Aversive Behaviors Given
Positive or Aversive Community Contacts as Antecedents*

Mother aversive

z =


aversive/ ~
Motherchildnonaversive ~




z =


Mother aversive/
child aversive ~



z = - 2.49"

"Probabilities to the left of the diagrams are based on all observations; probabilities
to the right of the diagram are based only on observations preceded by aversive
or positive maternal community contacts.
bp < .05.
cp < .01.

Indirect Influence


positive or aversive contacts. This was done, with the help of a z statistic
(Allison & Liker, 1982), to test for the presence of dependencies in the
mothers' aversiveness toward their children on the prevailing valence of their
prior contacts. Results are reported in Table II. This table shows that in five
out of six comparisons the probabilities of maternal aversiveness were
significantly related to the valence of their contacts; while aversive contacts
had a facilitating effect upon such aversiveness, positive contacts had an inhibitory effect upon it. In line with the previous analysis, these comparisons
indicated that this facilitating or inhibitory effect occurred irrespective of
the child behavior antecedent.

The results indicated that mothers who manifested severe interaction
problems with their children tended to behave more aversively toward them
on days in which they reported to have themselves engaged in a high proportion of aversive social contacts with adults in their community. Specifically,
it would appear that aversive social contacts exercised a facilitating effect
upon maternal aversiveness, while positive social contacts exercised an inhibitory effect upon it, this irrespective of the behavior exhibited by the child
in the course of interaction.
Although these findings support the predictions at the origin of this
study and, in so doing, emphasize the usefulness of the concept of setting
event or second-order effect, they should be interpreted with caution. The
relationships described here are correlational in nature and therefore do not
demonstrate that aversive social contacts play a causal role in aversive
mother-child interactions. (One could in fact argue that the causal chain might
have its source in the mother herself, who may somehow affect, from day
to day, the valence of all her interactions with others.) Similarly, the results
are based on group averages, which may mask the fact that these relationships may not always apply to individual dyads. Finally, the findings are
limited by methodological issues that were not addressed here. Future studies
will need to consider the possibility that sequential data of the type used in
this study may be autocorrelated (Gardner & Hartmann, 1984) and that the
degree of autocorrelation may vary within and across dyads. Given these reservations, however, the results raise some major issues that must be discussed
Why do insular, distressed mothers behave more aversively toward their
children when they themselves report having experienced high proportions
of aversive social contacts? Because mothers were asked to recall their adult
social contacts after each observation session, it could be argued that aversive child interactions evoked an angry mood in mothers on some days, which



in turn biased their recall of their earlier adult interactions. It would appear
that this is unlikely, however, since no changes in child aversiveness were
observed under the three conditions. Rather, a setting event analysis suggests that, when insular, distressed mothers respond to any one of their aversive partners (e.g., child, spouse, relative, social agency representative), their
responses are influenced not only by that person but also by many other persons with whom they commonly have aversive interchanges (Wahler &
Dumas, in press b). In other words, when responding to their children, these
mothers may in fact be responding to both the children's behavioral cues
and a broader pattern of cues that includes those provided by other social
agents in other settings. If this is correct, their daily variations in their
responses to their children may reflect not only the valence of the latter's
behavior but also the aversive or positive nature of many of the contextual
stimuli to which they are exposed. Under such circumstances, the finding
of high levels of aversive and indiscriminate or inconsistent behavior in insular mothers reported here and elsewhere (Dumas & Wahler, 1985) makes
sense. Indeed, it would be difficult to expect an insular mother to interact
with her child in a nonaversive and consistent fashion if her behavior is under
the control of environmental events that bear little contingent relationship
to the child's behavior.
Although they remain tentative at this stage, these results should encourage further research in this area if we are to understand better the role
of indirect influences in the development and maintenance of normal or
pathological behavior processes in children. Among other things, research
should attempt to develop and use better measurements of both adult social
contacts (e.g., use of multiple measures) and dyadic interactions (e.g., use
of response chain analyses) than the measurements used in this study. The
results should also encourage clinicians who seek to help distressed families
to pay close attention not only to relationships within the family but also
to the wider environmental context that provides the setting in which these
relationships take place. The fact that child behavior did not covary with
maternal aversiveness suggests that, in some family referrals at least, a parent
may need more help than a child. If this help is to modify a parent's pattern
o f social contacts, psychologists will probably need to join forces with other
professionals (e.g., social workers, physicians, and psychiatrists) and community leaders in an interdisciplinary attempt to provide comprehensive services to distressed families.


Allison, P. D., &Liker, J. K. (1982). Analyzingsequentialcategoricaldata on dyadicinteraction: A commenton Gottman.Psychological Bulletin, 91, 393-403.

Indirect Influence


Barker, R. G. (1968). Ecologicalpsychology. Stanford: Stanford University Press.

Barkley, R. A. (1981). The use of psychopharmacology to study reciprocal influences in parentchild interaction. Journal of Abnormal Child Psychology, 9, 303-310.
Broufenbrenner, U. (1979). The ecology of human development. Cambridge, Massachusetts:
Harvard University Press.
Burgess, R. L., & Conger, R. D. (1978). Family interaction in abusive, neglectful and normal
families. Child Development, 49, 1163-1173.
Dumas, J. E. (1984a). Interactional correlates of treatment outcome in behavioral parent training. Journal of Consulting and Clinical Psychology, 52, 946-954.
Dumas, J. E. (1984b). Indiscriminate mothering: Empirical findings and theoretical speculations. Advances in Behaviour Research and Therapy, 6, 13-27.
Dumas, J. E., & Wahler, R. G. (1985). Indiscriminate mothering as a contextual factor in
aggressive-oppositional child behavior: "Damned if you do, damned if you don't." Journal
of Abnormal Child Psychology, 13, 1-17.
Gardner, W., & Hartmann, D. P. (1984). On Markov dependence in the analysis of social interaction. Behavioral Assessment, 6, 229-236.
Guilford, J. P., & Fruchter, B. (1978). Fundamental statistics in psychology and education (6th
ed.). New York: McGraw-Hill.
Hartmann, D. P. (1977). Considerations in the choice of interobserver reliability estimates. Journal
of Applied Behavior Analysis, 10, 103-116.
Hartup, W. W. (1978). Perspectives on child and family interaction: Past, present, and future.
In R. M. Lerner & G. B. Spanier (Eds.), Child influences on marital and family interaction. A life-span perspective (pp. 23-46). New York: Academic Press.
Hartup, W. W. (1979). Levels of analysis in the study of social interaction: A historical perspective. In M. E. Lamb, S. J. Suomi, & G. R. Stephenson (Eds.), Social interaction analysis:
Methodological issues (pp. 11-32). Madison: University of Wisconsin Press.
Hetherington, E. M. (1979). Divorce: A child's perspective. American Psychologist, 34, 851-858.
Hubert, L. (1977). Kappa revisited. Psychological Bulletin, 84, 289-297.
Kantor, J. R. (1959). Interbehavioral psychology. Granville, Ohio: Principia Press.
Lamb, M. E. (1977). Father-infant and mother-infant interaction in the first year of life. Child
Development, 48, 167-181.
Lamb, M. E. (1979). The effects of the social context on dyadic social interaction. In M. E.
Lamb, S. J. Suomi, & G. R. Stephenson (Eds.), Socialinteraction analysis:Methodological
issues (pp. 253-268). Madison: University of Wisconsin Press.
Lobitz, W. C., & Johnson, S. M. (1975). Parental manipulation of the behavior of normal and
deviant children. Child Development, 46, 719-726.
Mash, E. J., & Johnston, C. (1982). A comparison of the mother-child interactions of younger
and older hyperactive and normal children. Child Development, 53, 1371-1381.
McCall, R. B. (1977). Challenges to a science of developmental psychology. Child Development, 48, 333-344.
Patterson, G. R. (1976). The aggressive child: Victim and architect of a coercive system. In
E. J. Mash, L. A. Hamerlynck, & L. C. Handy (Eds.), Behaviormodification andfamilies.
1. Theory and research (pp. 267-316). New York: Brunner/Mazel.
Patterson, G. R. (1980). Mothers: The unacknowledged victims. Monographs of the Society
for Research in Child Development, 45(5, Serial No. 186).
Reid, J. B., Taplin, P. S., & Lorber, R. (1981). A social interactional approach to the treatment of abusive families. In R. B. Stuart (Ed.), Violent behavior: Social learning approaches to prediction, management, and treatment (pp. 83-101). New York:
Rutter, M. (1983). Stress, coping, and development: Some issues and some questions. In N.
Garmezy & M. Rutter (Eds.), Stress, coping, and development in children (pp. 1-41).
New York: McGraw-Hill.
Snyder, J. J. (1977). A reinforcement analysis of intervention in problem and nonproblem
children. Journal of Abnormal Psychology, 86, 528-535.
Wahler, R. G. (1980). The insular mother: Her problems in parent-child treatment. Journal
of Applied Behavior Analysis, 13, 207-219.



Wahler, R. G., & Cormier, W. H. (1970). The ecological interview: A first step in out-patient
child behavior therapy. Journal of Behavior Therapy and Experimental Psychiatry, 1,
Wahler, R. G., & Dumas, J. E. (in press a). Stimulus class determinants of mother-child coercive interchanges in multidistressed families: Assessment and intervention. In J. D. Burchard & S. Burchard (Eds.), The prevention of delinquent behavior. Beverley Hills, CA:
Wahler, R. G., & Dumas, J. E. (in press b). Family factors in childhood psychopathology: Toward
a coercion-neglect model. In T. Jacob (Ed.), Family interaction and psychopathology.
New York: Plenum Press.
Wahler, R. G., House, A. E., & Stambaugh, E. E. (1976). Ecological assessment of child problem behavior. A clinicalpackage for home, school, and institutional settings. New York:
Pergamon Press.
Wahler, R. G., Leske, G., & Rogers, E. S. (1979). The insular family: A deviance support system
for oppositional children. In L. A. Hamerlynck (Ed,), Behavioral systems for the
developmentally disabled: 1. School and family environments (pp. 102-127). New York:
Winer, B. J. (1971). Statistical principles in experimental design. New York: McGraw-Hill.