Indiscriminate mothering: Empirical findings and theoretical speculations

Indiscriminate mothering: Empirical findings and theoretical speculations

Adv. Behav. Res. Thu. Printed in Great Britain. Vol. 6. PP 13-27. 1984 All rights reserved. INDISCRIMINATE FINDINGS Copyright 0 014M402184 $0...

952KB Sizes 1 Downloads 59 Views

Adv.

Behav.

Res. Thu.

Printed in Great Britain.

Vol. 6. PP 13-27. 1984 All rights reserved.

INDISCRIMINATE FINDINGS

Copyright

0

014M402184 $0.00 + .50 1984 Pergamon Press Ltd

MOTHERING: EMPIRICAL AND THEORETICAL SPECULATIONS Jean E. Dumas*

The University

of Western

Ontario,

Ontario,

Canada

Parent training is a therapeutic strategy which assumes that a problem child can best be helped by considering the social system in which he or she functions, assessing the variables which maintain deviant behavior within this system, and intervening to modify the system accordingly (Wahler & Dumas, 1984). Several process and outcome studies suggest that this systems approach to problem child behavior may be much more complex than was initially believed Bernal, Klinnert, & Schultz, 1980: Dumas & Wahler, 1983: Johnson & (e.g., Christensen, 1975; Patterson, 1980; Wahler, 1980). Specifically, there is growing evidence to indicate that long-term success in parent training depends not only on the patterns of interaction which are characteristic of the parent-child relationship but also on the many contextual variables which indirectly influence this relationship. In other words, success may depend on two classes of events: (1) the immediate stimulus events which parent and child exchange in the course of interaction (typically, therapy will attempt to insure that parents praise rather than criticize, reward rather than ignore, set consistent limits, etc.) and (2) the contextual or setting events which impinge upon parent and child. Setting events are generally definedas environmental conditions (i.e., stimuli or stimulus-response interactions) which influence the function of later stimuli or stimulus-response interactions; in parent training they commonly consist of material and social stressors which may interfere with therapy and limit its effectiveness. Thus, a consideration of the social system in which a problem child functions may require not only a functional analysis of the deviant parent-child relationship but also an understanding of the wider social context in which this relationship is set (Bronfenbrenner, 1979; Griest & Forehand, 1982; Wahler & Dumas, 1984). This article reviews and discusses some of the theoretical and practical implications of a group of studies which attempted to assess the impact of adverse setting events on the behavior of mothers who participated in a structured behavioral parent training program to modify aggressive, . . opposltlonal child behavior. Briefly stated, the program was designed to teach mothers (and fathers when present) to reduce their children's rate of oppositional behaviors through a time-out contingency while increasing their rate of desirable behaviors through a point reward system. The program was *Department London,

of Psychology, Ontario, N6A 5C2,

The University Canada

of

13

Western

Ontario,

J. E. Dumas conducted in three phases (baseline, treatment, follow-up). In each phase, home observations were collected by trained observers, using the Standardized Observation Codes developed by Wahler, House, and Stambauqh (1976). This coding system provided a comprehensive (24 codes) picture of interactions between the target child and other family members. A detailed description this intervention can be found in Wahler (1980) or Wahler and Dumas (1983). Parent

Training

of

Outcome

Several reviews (Forehand & Atkeson, 1977; Graziano, 1977; O'Dell, 1974; Sanders & James, 1983) indicate that the effectiveness of parent training has been equivocal. Some studies indicate that it can lead to positive therapeutic outcomes that are maintained at follow-ups of one year (Forehand, Sturqis, McMahon, Aguar, Green, Wells, & Breiner, 1979; Patterson, 1974; Patterson & Fleischman, 1979; Weinrott, Bauske, & Patterson, 1979) or more (Baum & Forehand, 1981). However, there is also evidence that parent training can have little or no therapeutic effects (Eyberg & Johnson, 1974; Ferber, Keeley, & Shemberq, 1974; Johnson & Christensen, 19751, that significant effects may not be maintained in a follow-up phase (Bernal and others, 1980: Wahler, 1980: Wahler & Afton, 1980; Wahler & Dumas, 1983). or that some families dropout of treatment before its completion (Fleischman & Szykula, 1981: Johnson & Christensen, 1975; McMahon, Forehand, Griest, & Wells, 1981; Reid & Patterson, 1976). There is growing evidence to suggest that specific factors are commonly associated with outcome. It would appear that parents, and especially mothers, who are repeatedly subjected to multiple material and social stressors are less able to benefit from parent training. Specifically, adverse setting events such as low socioeconomic status (Dumas & Wahler, 1983; McMahon and others, 1981; Patterson, 1974: Wahler, 1980; Wahler & Dumas, 1983), residence in a high-crime neighborhood (Dumas & Wahler, 1983; Wahler, 1980; Wahler & Dumas, 1983), social isolation (Dunas & Wahler, 1983; Wahler, 1980; Wahler & Afton, 19801, absence of father (Dumas & Wahler, 1983; Patterson, 1974; Strain, Young, & Horowitz, 1981), marital conflict (Reisinger, Frangia, & Hoffman, 19761, and parental depression (McMahon and others, 1981; Patterson, 1976) tend to show negative correlations with outcome. Dumas and Wahler (1983) demonstrated this in two identical studies of 67 families who had taken part in the parent training program described above (49 families participated in the original Prior to treatment, scores on several study and 18 in a replication study). measures of socioeconomic disadvantage and social isolation or "insularity" were obtained for each family. They formed the basis of two indices of material and social stress. The first one, the Socioeconomic Disadvantage Index, reflected six variables: family income, maternal education, family mother's marital status at the time of initial interview), composition (i.e., family size, source of referral (i.e., whether a mother had referred herself for treatment or had been referred by a social agency), and area of residence (i.e., whether a family lived in a low- or high-crime neighborhood). Each family was given a score of 0 or 1 on each variable, depending on whether it fell at the advantageous or disadvantageous end of the continuum: the scores were then added to provide the family's standing on the index. The second was based on each mother's self-reports of her index, the Insularity Index, daily contacts with people in her neighborhood and reflected three variables: the number of daily contacts she reported, the identity of each contact person friend, kinfolk, helping agency representative), and her perceived (e.g., valence of each contact (from +3 = very positive, through zero = neutral, to -3 = very negative). A mother was classified as insular if she reported at least twice as many of her daily contacts with kinfolk and/or helping agency representatives as with friends, or a third of all her daily contacts as Treatment neutral or negative. Otherwise, she was classified as noninsular. effectiveness was assessed (on the basis of the behavioral home observations)

Indiscriminate TABLE

1.

Disadvantage Index

Expected Probabilities Failure Given Different and Insularity (after Pergamon Press Inc.,

1.5

and Empirical Proportions of Treatment Levels of Socioeconomic Disadvantage Dumas & Wahler, 1983; copyright reprinted with permission)

Expected of

Insularity Oriqinal Study N=49

Note:

Mothering

Empirical Proportions of Failure

Probabilities Failure

Replication Study N=18

0

0

.oo

.oo

1 2 3 4 5 6

0 0 0 0 0 0

.06 .15 .25 .34 .43 .52

.04 .13 .23 .33 .42 .52

0 1 2 3 4 5 6

1 1 1 1 1 1 1

.45 .54 .63 .72 .81 .90 .99

.47 .56 .66 .75 .85 .94 .99

For example, a score of an expected

a participant in the original 3 on the disadvantage index but probability of treatment failure

Both Samples N=67 -00 .17 .41

.67 .66 1.00

study who obtained was noninsular had of -25.

at a one-year follow-up. Results are summarized in Table 1. They indicated a steady increase, in each study, in the probability of treatment failure in the presence of disadvantage, insularity, or both. A discriminant analysis model which included the two indices as predictor variables accounted for 49% of the variance in outcome and classified over 80% of the families correctly. The usefulness of both instruments in predicting outcome at follow-up was supported by the fact that a comparable increase in the empirical proportion of treatment failure was found when the two samples were grouped together. These results may account in part for some of the equivocal outcome findings mentioned above. For example, Wahler (1980) and Wahler and Dumas (1983) studied two samples of mothers taking part in parent training, assessing treatment effectiveness through home observations before and during training and at a one-year follow-up. In each study, results showed significant improvements in parent-child problems from baseline to intervention. However, in all families, these problems returned to baseline in the follow-up phase. In contrast, Baum and Forehand (1981) reported l- to 4.5-year maintenance of treatment gains in a sample of families who participated in another standardized program. Although direct comparisons between studies are not possible here, it should be noted that all the families Wahler (1980) and Wahler and Dumas (1983) worked with were highly disadvantaged families subsisting on incomes which averaged less than $5,400 a year, but that only one of the families studied by Baum and Forehand (1981) was described as receiving welfare.

J. E. Dumas Indiscriminate

Mothering

The results just reviewed are likely to make intuitive sense to many. TO recognize that mothers can be helped or hindered in their childrearing task by events other than the immediate discriminative stimuli provided by their children is to acknowledge the importance of setting events. However, this leaves researchers with the challenging task of explaining how material and social stressors come to exercise their adverse impact on the parent-child relationship, thus presumably interfering with current parent training practices. Although their findings were correlational in nature, Dumas and Wahler (1983) suggested that such stressors may play a causal role in a parent's ability to provide effective caretaking. Specifically, they argued that a mother who is repeatedly subjected to high rates of environmental stress may become inconsistent or indiscriminate (i.e., noncontingent) in her own responses to others, her problem child in particular. The long-term success of parent training obviously requires that a distressed mother will learn to display consistent, prosocial contingencies in her relationship with her child and continue to use them over time and across situations. But can this mother reasonably be expected to adopt and maintain such contingencies if she continues to receive a high rate of aversive social inputs from her environment? If indiscriminate responding is a function of the density of aversive inputs, one would predict that she cannot. Two studies (Dumas, in press; Dumas & Wahler, in press) were conducted to test this prediction with 52 mother-child pairs who had taken part in the program described above. Two types of measures were derived from the observational data collected in each family: (a) Baserate probabilities of non-aversive and aversive mother and child behavior were computed by grouping all observations for each dyad, separately adding the number of observation intervals in which each behavior of interest had been scored, and dividing the results by the total number of intervals. These probabilities were obtained for each dyad separately and 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 observation intervals in which a behavior of interest (e.g., Mother Aversive) had been scored, but only if it had been preceded by another behavior of interest (e.g., Child Aversive) in either of the previous two observation intervals (i.e., 30 seconds) and dividing the results by the baserate probability of the second behavior. These probabilities were also obtained for each dyad separately and averaged across dyads. The reliability of these measures was found to be adequate. The first study (Dumas & Wahler, in press) divided the sample into a noninsular and an insular group according to the insularity decision rules described above and compared the levels of maternal indiscriminate responding within and between groups in baseline. The baserate probabilities indicated that insular mothers and their children displayed higher rates of aversive behaviors than their noninsular counterparts. The conditional probabilities p(mother behavior/child behavior) are presented in Table 2 (upper half). Within-group comparisons indicated that all mothers were far more likely to respond to any child behavior with non-aversive than with aversive behavior. Between-group comparisons indicated that this non-aversive response tendency applied equally to the two groups. However, although the probability of aversive maternal insular mothers were found to be more likely responses remained relatively low, than noninsular mothers to respond to their children's behavior aversively. This was true of their responses to aversive and non-aversive behavior, although, in the latter case, the comparison failed to reach the .05 level of Given these differences, the conditional significance, t(50) = -1.63, E = .07. probabilities of mother aversive behavior given non-aversive and aversive child antecedents were compared to the baserate probability of mother aversive behavior, this to test whether maternal aversiveness depended upon any of the Results are presented in child behaviors which immediately preceded it.

Indiscriminate TABLE

Mother Mother

Mother Mother

Child Child

Child Child

2.

Conditional Probabilities Non-Aversive and Aversive and Between (Rows) Groups copyright Plenum Publishing

Non-Aversive/Child Aversive/Child 4 (19 or Non-Aversive/Child Aversive/Child t (19 or Non-Aversive/Mother Aversive/Mother & (19 or

17

of Mother and Child Behaviors Given Antecedents Compared Within (Columns) (after Dumas & Wahler, in press; Corp., reprinted with permission) Noninsular (n = 21)

Insular (n = 311

.833 .047 13.37***

-827 .091 12.51***

0.09 -1.63

Aversive Aversive

.713 .066 7.30***

.624 .127 2.34***

0.35 -2.10*

Non-Aversive Non-Aversive

.147 .107 2.09*

.157 .180 -1.38

-0.36 -3.00**

.270 .354 -0.88

.301 .515 -4.92***

-0.42 -1.92*

Non-Aversive Non-Aversive 29)

29)

29)

Non-Aversive/Mother Aversive/Mother 4 (19 or

29)

* p-z

**

.05,

Mothering

Aversive Aversive

p < .Ol,

***

4

(50)

2 < .OOOl

Figure 1 (upper half). This figure shows that noninsular mothers significantly increased their level of aversiveness 0.15 in response to aversive child behavior but that insular mothers did so in response to both aversive and non-aversive child behavior. This provided evidence of indiscriminate mothering in insular families by suggesting that the aversive behavior of insular mothers was under the stimulus control of child antecedents, almost irrespective of the valence of these antecedents. In other words, aversive and non-aversive child behaviors acted as if they were discriminative for aversive maternal responding. Dumas (in press) confirmed and extended these findings. This study divided the sample into a successful and an unsuccessful group at follow-up and compared the levels of maternal indiscriminate responding within and between groups in each phase of the program. The conditional probabilities indicated that unsuccessful mothers were more likely to respond aversively to both aversive (baseline and follow-up) and non-aversive (all phases) child behavior when compared with successful mothers. This evidence of higher levels of aversive and indiscriminate responding in unsuccessful mothers was confirmed in within-group comparisons. These suggested that the aversive behavior of unsuccessful mothers was under the stimulus control of child antecedents, irrespective of the valence of these antecedents in baseline and follow-up. Thus, except for short-term changes associated with treatment, aversive and non-aversive child behaviors acted as if they were discriminative for aversive responding, again providing evidence of indiscriminate responding in these mothers. In contrast, successful mothers were likely to react aversively only in response to child aversive behavior, and this only in baseline and treatment. In follow-up, child aversive behavior was unlikely to raise maternal aversiveness above its baserate, suggesting that, by then, these

8

J. E. Dumas

Noninsular Variables

Lagged

dependencies

given Child Aversive

in mother

aversive

behavior

.066

z =

Mother Aversive

IIlSUla~

.127

z = 3.76***

2.20*

.018

.043 7. = 1.62

z = 2.05*

given

Child Non-Aversive .047

Lagged

dependencies

given Mother Aversive

.091 in

child

behavior

.354

.515

z = 3.23*'* Child Aversive

aversive

z = 6.23****

.072

.123 z = 2.81*'

z = 3.24***

given Mother Non-Aversive .107 Fig.

1.

.180

of observed dependencies in mother and child aversive behaviors given non-aversive and aversive antecedents. All probabilities to the left of the diagrams are baserate probabilities; all probabilities to the right of the diagrams are conditional upon child or mother behavior (after Dumas & Wahler, in press: copyright Plenum Publishing Corp., reprinted with permission).

Diagrams

* p r: .05,

l

* p < -01,

***

fl L .OOl,

****

E < .OOOl.

a

Indiscriminate mothers learned

had acquired in training.

and

were

maintaining

19

Mothering the

prosocia

1 contingencies

they

had

These results confirm earlier findings. For example, Patterson (1976) compared a sample of clinic-referred antisocial children with a matched control sample of normal children. Home observations indicated, not only that the deviant group displayed higher rates of aversive exchanges, but also that the mothers in that group were less predictable in matching their aversive responses with their children's aversive actions. the deviant children received higher Thus, rates of aversive maternal attention than their normal counterparts, even when they behaved appropriately. In a similar study, Snyder (1977) found that familial, and especially maternal, indiscriminate use of attention to child behavior was much more evident in clinic referred than in control families, so much so that in these families the probability of receiving a positive, neutral, or aversive consequence was almost independent of the behavior displayed by the child. However, the results also extended these earlier findings by demonstrating that, rather than being a characteristic of all mothers of aggressive children, indiscriminate responding was most likely to be found in mothers who were repeatedly subjected to high rates of aversive setting events and commonly failed in parent training. Punishment

Acceleration

Considering mother and child behaviors as both independent and dependent, predictor and predicted variables, the studies reviewed in the previous section also measured consequences provided by children by computing the conditional probabilities p(child behavior/mother behavior). Results of the Dumas and Wahler (in press) study are presented in Table 2 (lower half). Within-group comparisons indicated that noninsular children were more likely to match their response to the maternal antecedent, although this tendency was only significant in their responses to mother non-aversive behavior. Insular children responded to both maternal antecedents by exhibiting more aversive than non-aversive behavior, although this tendency was only found to be significant in response to aversive maternal behavior. Between group comparisons indicated that all children were equally likely to respond with non-aversive behavior to both non-aversive and aversive maternal behavior. However, insular children were found to be significantly more likely than Given noninsular children to respond to any maternal behavior aversively. these differences, the conditional probabilities of child aversive behavior given non-aversive and aversive maternal antecedents were compared to the baserate probability of child aversive behavior. As in the case of mothers, this was done to test for the presence of lagged dependencies in child behavior. Results are presented in Figure 1 (lower half). This figure shows that noninsular children tended to significantly increase their level of aversiveness in response to both aversive and non-aversive maternal behavior. Insular children were found to respond in a similar, though much more aversive manner than their noninsular counterparts. Specifically, they evidenced extremely high increases in aversive behavior in response to maternal aversiveness. Thus, rather than leading to a reduction in child aversiveness, maternal aversive behavior tended to increase it well above baserate, especially in insular families. Although, by reacting with increased aversiveness to any maternal behavior, all children evidenced some degree of indiscriminate responding, they generally responded with approximately three times as much aversiveness to maternal aversive as to maternal non-aversive behavior. This suggested that they remained able to discriminate between maternal antecedents and usually responded accordingly. As expected, comparable results were obtained in the second study (Dumas, in press) which compared the patterns of responding in successful and unsuccessful children.

20 TABLE

J. E. Dumas 3.

Baserate and Child Aversive

Conditional Behaviors

Probability Under Three

Measures Observation

of

Mother and Conditions.

Conditions

Observations

Behaviors

All

observations

preceded

by

Positive maternal contacts

Aversive maternal contacts

t(135)

Mother

Aversive

.044

.034

.062

-1.e5*

Mother Child

Aversive/ Non-Aversive

.088

.065

.122

-1.83*

Mother Child

Aversive/ Aversive

.112

.079

.159

-2.35*

Child

Aversive

.lll

.107

.llO

-0.23

Child Mother

Aversive/ Non-Aversive

.182

-201

.176

Child Mother

Aversive/ Aversive

.397

.389

-453

*p < .05,

Setting

Events

and

**p

0.76 -0.95

< .Ol.

Mother-Child

Interaction.

A Direct

Link?

The data presented so far indicate that mothers who exhibit high levels of aversive and indiscriminate behavior toward their children and fail to maintain long-term gains in parent training are commonly subjected to high levels of Although it is impossible to demonstrate, in the aversive setting events. absence of an experimental design, that such events play a causal role in this association, Dumas (1984) obtained evidence which clearly suggests that there may be a direct link between setting events and mother-child interaction. This study replicated and extended Wahler's (1980) findings that insular, disadvantaged mothers were more likely to respond to their children aversively on days in which they had first engaged in aversive interactions with other It was conducted with 14 mothers adults than on days in which they had not. who evidenced severe management problems with their antisocial children and reported consistently high levels of aversive interactions with adults in their community (see the Insularity Index described above). Baserate and conditional probability measures of mother and child behaviors were computed on three sets of baseline observations: (a) all observations; (b) all observations preceded by maternal self-report of positive contacts (i.e., less than a third of all and (c) all observations preceded by contacts rated as neutral or aversive), maternal self-report of aversive contacts (i.e., more than two thirds of all these measures were then compared contacts rated as neutral or aversive); Differences between these conditions were between the two contact conditions.

Indiscriminate

Mothering

.062

given Aversive Mother

z =

1.72*

Aversive <::::Positive contacts

.034

given Aversive contacts Mother Child

Aversive/ Non-Aversive

<

given Positive contacts given Aversive contacts

Mother Child

Fig.

Aversive/ Aversive

2.

.159 -7. =

2.92**

.112 <

given Positive Contacts

<

5 = -2.49* .079

Diagrams of observed dependencies in mother aversive behaviors given positive or aversive community contacts as antecedents. Probabilities to the left of the diagrams are based on all observations; probabilities to the right of the diagrams are based only on observations preceded by aversive or positive maternal community contacts. *E < .05,

**p

c -01.

only found for the measures of aversive behavior. As Table 3 indicates, the probabilities of child behavior did not differ under the two contact conditions. The children were not more likely to act in an aversive manner under either condition, whether their behavior was measured as a baserate or was conditional upon a maternal antecedent. However, the probabilities of mother behavior were found to differ systematically under the two conditions. The mothers were more likely to act in an aversive manner toward their children when they had experienced a large proportion of aversive contacts with adults As their in their community prior to an observation than when they had not. this response tendency applied conditional probabilities indicated, irrespective of the child behavior antecedent. Given this response tendency, the baserate and conditional probability measures of maternal aversiveness computed on all observations were compared to the same measures computed only on the observations preceded by positive or aversive contacts, to test for the presence of dependencies in the mothers' aversiveness toward their children on the prevailing valence of their prior contacts. Figure 2 shows that in five out of six comparisons the probabilities of maternal aversiveness were

22

J. E. Dumas

significantly related to this valence; while aversive contacts facilitating effect upon such aversiveness , positive contacts effect upon it. These comparisons indicated again that this inhibitory effect occurred irrespective of the child behavior

had a had an inhibitory facilitating or antecedent.

Discussion The reader should be warned against an overinterpretation of the findings presented here, as they are limited by important methodological issues common to much applied research in the field of family psychopathology. For example, this work was based on samples of families broadly defined as distressed or dysfunctional. However, it has not been demonstrated that these families can be placed in a single group for research purposes: there may be important clinical differences among them which may in fact limit the generality of the findings. This generality is also limited by an absence of control groups to provide measures of mother-child interaction in adequate families. Similarly, this work did not control for potentially important variables such as the age and sex of the children; thus, the results may mask systematic developmental differences in family interaction. Finally, these results relied on behavioral (direct observation) and statistical (sequential analysis) techniques which remain fairly crude at this time. Developments in these techniques, improvements in subject selection, and control of potentially important variables should at least improve the precision of these findings, if they do not modify them in fundamental ways. Given these limitations, however, the results raise some major issues for a proper understanding of family psychopathology; these issues are discussed in the context of a model put forward to account for the role played by adverse setting events in the maintenance of aversive parent-child interaction. Why do insular, disadvantaged mothers provide indiscriminate consequences for their children's prosocial and aversive behaviors? There are at least two complementary ways of accounting for these findings from an operant perspective. On the one hand, they may be understood in terms of a contingency analysis. It is well-known that aversive behaviors can function as social reinforcers in family interactions (e.g., Patterson, 1976; 1980). In this one could expect that a mother might react aversively to a child's perspective, non-aversive approach and thus increase the likelihood that the child would then behave aversively toward her. This possibility cannot be ruled out in the absence of an investigation of the probabilities of response chains (e.g., child aversive/mother aversive/child non-aversive; child aversive/mother aversive/child aversive). However, although the analyses did not include response chains, it is unclear how a contingency analysis would account for the fact that: (1) insular children responded with approximately three times as much aversiveness to maternal aversive as opposed to non-aversive behavior when their mothers responded with comparable levels of aversiveness to any of their behaviors; (2) unsuccessful mothers evidenced higher levels of aversive and indiscriminate responding in baseline than successful mothers even though their children were not more likely to be aversive in baseline than their successful or (3) insular mothers evidenced higher levels of aversive and counterparts; indiscriminate responding on days in which they had themselves experienced high levels of aversive social contacts in the absence of corresponding changes in their children's behavior. Because of these limitations, the results may also be understood in terms of a contextual, or setting event, analysis. As mentioned in the introduction, a setting event is an environmental condition (i.e., a stimulus or stimulus-response interaction) which influences the function of later stimuli We have argued elsewhere (Dumas & Wahler, or stimulus-response interactions. disadvantaged mothers, when in press; Wahler & Dumas, 1984) that insular, responding to their children's behavior, may be influenced not only by that setting events in their behavior but also by other, usually aversive,

Indiscriminate

Mothering

23

environment. In effect, their responses to their children may reflect, not only the valence of the child's behavior, but also the aversive nature of many of the contextual stimuli to which they are regularly subjected, such as economic pressures and conflicts with other adults. Although this analysis is speculative, it has major theoretical implications for an operant approach to We present these implications in the form of an expanded family interaction. operant model aimed at accounting for the maintenance of aversive parent-child interactions. As readers familiar with the experimental analysis of behavior literature will notice, this model draws heavily upon evidence on the relativity of reinforcement (Premack, 1971), schedule control (Morse & Kelleher, 1977). and choice in concurrent schedules (Badia, Harsh, & Abbott, 1979; De Villiers, 1977; Herrnstein, 1961, 1979; Imada & Naqeishi, 1982). The model. The model assumes that, in order to understand a child's aggressive behavior at home, two classes of events need to be considered: the immediate discriminative stimuli exchanged by parent and child and, further removed in time, the setting events which characterize these individuals' material and social environment. When responding to each other, mother and child do not only respond to the behavioral cues they exchange but also to a broader pattern of cues provided by their environment. They do so by evaluating the reinforcing property of a specific stimulus event in relationship to the overall reinforcement provided by the environment and the contingent nature of that reinforcement. Specifically, the model makes the following assumptions: (1) The reinforcing or punishing function of stimulus events is not an intrinsic property of these events: rather, it depends upon the relationship of these events to their setting context. Thus, a stimulus event can only be defined as a "reinforcer" or a "punisher" in a given context. (2) If the setting context is aversive in nature (i.e., characterized by a high ratio of aversive to positive setting events), the reinforcement function of many stimulus events presented in this context is likely to be aversive irrespective of the apparent valence of these stimulus events. (3) If the setting context is inconsistent in nature (i.e., unpredictable, noncontingent), the reinforcement function of many stimulus events presented in this context is likely to be aversive irrespective of the apparent valence of these stimulus events. (41 If an individual within an aversive setting context produces a behavior that is followed by a decrease in the aversiveness of a stimulus event, that consequence will prove to be reinforcing. The critical feature of this consequence is its non-aversive nature when compared to the setting context. As long as it is less aversive than its context, it will prove reinforcing. (5) If an individual within an inconsistent context produces a behavior that is followed by the contingent presentation of a stimulus event, that consequence will prove to be reinforcing. The critical feature of this consequence is its contingency or predictability when compared to the setting context. As long as it exceeds its context in predictability, it will prove reinforcing. In essence, the model just outlined is relational in nature. It assumes that the impact of stimulus events on behavior is largely determined by the context in which they are presented: it is the relational quality of these events to their context which determines their reinforcing or punishing value. Thus, in distressed families at least, immediate interactions between mother and child cannot be fully understood in the absence of historical or contextual information. Consider again some of the results presented above in light of this model. The first assumption of relativity of reinforcement that, while child non-aversive behavior tended to aversiveness well above its baserate (a definition disadvantaged, insular families who commonly failed

was illustrated by the fact increase maternal of reinforcement) in in parent training, the

J. E. Dumas same behavior never had the same effect in less disadvantaged noninsular families who usually succeeded. The relativity of reinforcement and the importance of contextual information were also illustrated in the last study (Dumas, 1984) which showed that mothers behaved as if their aversive encounters with adults were predictive of later child aversiveness, even though their children did not behave differently on days in which their mothers had or had not experienced aversive community contacts. In terms of assumption 2, the reinforcement function of many child stimulus events differed for the mothers according to the context in which they were presented, even though the nature of these events did not change significantly as the context changed. Tentative support for assumption 3 was found in the fact that children who functioned in inconsistent environments were systematically more aversive than their counterparts who did not, whether their behavior was measured as a baserate or conditionally upon a maternal antecedent. Thus, the reinforcement function of many maternal stimulus events differed for these children according to the context in which they were presented, even though the nature of these events did not apparently change as the context changed. Stronger support for this assumption would be available, however, if similar changes in behavior could also be demonstrated within, rather than between, children as a function of contextual changes. Although no data in support of assumption 4 were presented above, such data could be collected. For example, the assumption predicts that a distressed mother may be expected to react differently to the same child transgression on days in which she has first experienced aversive contacts with other adults than on days in which she has not. Thus, she may "give in" to noncompliance (e.g., child refuses to put toys away and whines) on aversive, but not on nonaversive days. On former days, the consequence of such behavior would presumably be reinforcing (i.e., child stops whining) when compared to the aversive context, even though this consequence has not removed the aversive stimulus (i.e., toys have not been put away). On nonaversive days, however, the same consequence might not be reinforcing and the mother would presumably be more likely to use some of the techniques she learned in parent training to insure her child's compliance. A comparison of the probabilities of response chains (e.g., mother consequence/child noncompliance/mother instruction) between aversive and nonaversive days would provide data relevant to this assumption. Finally, the punishment acceleration findings described above can be understood in terms of assumption 5. Data indicated that children were most likely to respond to maternal aversiveness with aversive behavior in a context in which their nonaversive behavior commonly led to an unpredictable maternal response. The fact that such behavior in turn led to a high probability of maternal aversiveness suggests that some children acted coercively at least in part to generate predictability. Sporadic aversive exchanges could presumably be reinforcing in an unpredictable social context. Response chain analyses will again be needed to investigate this possibility. In the meantime, the assumption suggests that predictability may play a major role in the organization of behavior and could, therefore, be sought after even at the cost of long or intense aversive stimulation. Although they remain tentative of this stage, the results reported here and the model just outlined should encourage further research. Research is urgently needed in this area if we are to understand better the complex interplay between parent-child interaction and setting events which appears to exercise a major influence on the maintenance of aggressive , oppositional behavior in children. 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 the wider environmental context which provides the setting in which these relationships take place. As suggested by the fact that parent clinicians ought to be careful and child behaviors do not necessarily covary, not to focus their attention mainly on children in family referrals: in some cases at least, a parent may need more help than a child. If this help is to modify a parent's pattern of aversive setting events, psychologists will need to join forces with other professionals and community leaders in an interdisciplinary attempt to provide comprehensive services to distressed families.

Indiscriminate

Mothering

25

References Badia,

Baum, Bernal,

P., Harsh, unpredictable 1107-1131. C. G., training

J., & Abbott, B. shock conditions:

& Forehand, by use of

Choosing and theory.

Mass.

:

U. Harvard

(1979). The University

Choice DeVilliers, P. (1977). formulation of the law of (Eds.), Handbook of operant Prentice Hall. J. E. behavioral

(In

Press). parent

between predictable Psych. Bull.,

R. (1981). Long term follow-up assessment Behav. Ther., multiple outcome measures.

M. E., Klinnert, M. D., & Schultz, L. A. of behavioral parent training and client-centered J. of App. children with conduct problems.

Bronfenbrenner,

Dumas,

(1979). Data

ecology Press.

of

human

Dumas,

J. E. (1984). Indirect mother-child interactions submitted for publication.

Dumas,

J. E., & Wahler, R. G. (1983). Mother insularity parent training: Behav. Assess., 2, 301-313.

Dumas,

J. E., & Wahler, contextual factor if you do, damned

Eyberg,

S. M., & Johnson, S. M. modification with families: order of treated problems.

Ferber,

H., Keeley, S. M., & Shemberg, behavior modification: Outcome Behav. after Patterson's work.

of

correlates Consult.

influence of in distressed

Behav.

development.

and a quantitative L J. E. R. Staddon Englewood Cliffs:

of treatment outcome and Clin. Psych.

maternal families.

social contacts Manuscript

Indiscriminate mothering child behavior: of Abnor. Child Psych.

K. M. (1974). Training of problems encountered Ther., 2, 415-419.

on

in

as a "Damned

in

parents in a program

A community setting replication (1981). for aggressive children. Behav. Ther.,

Generality M. (1977). A review of assessment Ther., 2, 575-593. B.

of

treatment effects and implementation

Forehand, R., Sturgis, E. T., McMahon, R. J., Aguar, D., Green, K., Wells, K. C., & Breiner, J. (1979). Parent behavioral training Treatment generalization across time and child noncompliance. to school. Behav. Modif., 2(l), 3-25. P.

in

Multiple assessment of behavior (1974). Effects of contingency contracting and J. of Consul. & Clin. Psych., 42, 594-606.

Fleischman, M. J., & Szykula, S. A. of a social learning treatment 12, 115-122.

Graziano, A. M. (1977). R. M. Eisler, and vol. 4. New York:

parent 643-652.

Cambridge,

Predictors of treatment outcome and socioeconomic disadvantage.

R. G. (In Press). in aggressive-oppositional J. if you don't."

Forehand, R., & Atkeson, parents as therapists: Behav. procedures.

of 12,

Outcome evaluation parent counseling for Anal., 13, 677-691.

(1980).

in concurrent schedules In W. K. Honig effect. behavior (pp. 233-287).

Interactional J. training.

and 86,

Parents as behavior M. Miller (Eds.), Academic Press.

therapists. Progress

in

with

to modify from home

In M. Hersen, behavior modification,

26 Griest,

J. E. Dumas D. L., & Forehand, R. (1982). How can I get any parent training with all these other problems going on?: The role of family variables in child behavior therapy. Child and Family Behav. Ther., 4, 73-80.

Hermstein, R. J. (1961). function of frequency 4, 267-272.

Relative and of reinforcement.

Hermstein,

Derivatives

Imada,

R.

J.

(1979).

H., & Nageishi, experiments using

Y. (1982). aversive

Johnson, S. M., & Christensen, behavior modification 135-154. McMahon, R. J., Forehand, out of therapy during 1, 79-85.

O'Dell,

S. Psych.

g,

of

The concept stimulation.

(1977). and J. Cliffs:

Training parents 418-433.

Patterson, G. R. (1974). Multiple settings, Psych., -,42 471-481.

Intervention treatments

in

and

Patterson, G. R. (1976). The aggressive a coercive system. In E. J. Mash, (Eds.), Behavior modification and Patterson, of

the

Patterson, effects: data.

G. R. Sot.

Reid,

(1980). for Res.

Mothers: in Child

G. R., & Fleischman, Some considerations Behav. Ther., lo,

Premack, D. (1971). generalization: The nature of

M.

of Exp.

Psych.

response Anal. of

Rev.,

of uncertainty Psych. Bull.,

in 91,

Multiple criteria J. of Abnor. Child & Wells, training?

K.

Determinants E. R. Staddon Prentice-Hall. behavior

for boys criteria. child: L. A. families.

with J.

as a Behav.,

as,

486-495.

animal 573-588.

follow-up Psych.,

of 3(2),

C. (1981). Behav. Coun.

Who drops Quar.,

of reinforcement (Eds.), Handbook

modification:

and of

A review.

conduct problems: of Consul. and Clin.

Victim and architect of Hamerlynck, and L. C. Handy New York: Bruner/Mazel.

The unacknowledged Develop., -45 (5,

victims. Serial No.

e 186).

Maintenance of treatment family systems and follow-up

J. (1979). concerning

168-185.

Catching up with common sense, or two sides In R. Glaser Reinforcement and punishment. reinforcement. New York: Academic Press.

of

a (Ed.),

J. B., & Patterson, G. R. (1976). The modification of aggression and In A. Bandura & E. Ribes stealing behavior of boys in home settings. Experimental analyses of aggression (Eds.), Behavior modification: Hillsdale, N-J.: Lawrence Erlbaum Associates. and delinquency.

Reisinqer, J. J., Frangia, G. W., & Hoffman, training: Generalization and marital Exp. Psych., 1, 335-340. Sanders, M. R., & James, J. A review of generalization Snyder,

matching.

R., Griest, D. L., parent behavioral

W. H., & Kelleher, R. T. punishment. In W. K. Honiq operant behavior. Enqlewood

strength of the

J.

A. (1975). families.

with

Morse,

(1974). Bull.,

absolute

done

J. J. nonproblem

(1977). children.

E.

E. H. status.

(1976). J. of

(1983). The modification and maintenance. Behav.

A reinforcement J. of Abner.

analysis Psych.,

Toddler management Behav. Ther. and of parent behavior. Modif., 1, 3-27.

of intervention 86, 528-535.

in

problem

and

Indiscriminate Strain,

P. change family

27

Mothering

S., Young, C. C., & Horowitz, J. (1981). Generalized during oppositional child training. An examination Behav. Modif., 2, 15-26. demographic variables. The insular Appl. Behav.

R. G. treatment.

(1980). J. of

Wahler,

R. G., noninsular

& Afton, mothers.

Wahler,

Stimulus class determinants of R. G., & Dumas, J. E. (1983, June). mother-child coercive interchanges in multidistressed families: Assessment and intervention. Paper presented at the Vermont Conference on the Primary Prevention of Psychopathology, Bolton Valley, VT.

Wahler,

R. G., & Dumas, J. E. (1984). of insular and noninsular mothers: In R. F. Dangel training effects. training: Foundations of research Press.

Wahler,

R. G., House, A. E., of child problem behavior. institutional settings.

Weinrott,

Systematic M. R., Bauske, B. W., & Patterson, G. R. (1979). replication of a social learning approach to parent training. In P. Sjoken, S. Bates, and W. S. Dockens (Eds.), Trends in behavior Academic Press. therapy. New York:

D. (1980). Child Behav.

Her problems 13, 207-219.

Attentional Ther., z(2),

in

and

Wahler,

A.

mother: Anal.,

behavior of child

processes 25-41.

parent-child in

Changing the observational A step toward maintenance and R. A. Polster (Eds.), and practice. New York:

& Stambaugh, A clinical New York:

E.

E. (1976). package for Pergamon Press.

insular

and

coding styles of parent Parent Guilford

Ecological home, school

assessment and