20
JOURNAL OF APPLIED DEVELOPMENTAL PSYCHOLOGY 9s 421-440 (1988) Mothers' & Fathers' Explanations of Observed Interactions with Their Children ALAN RUSSELL Flinders University GRAEME RUSSELL Macquarie University Mothers and fathers were asked via an open-ended interview technique to explain specific interactions with their children that had been noted during a period of family observation in the home. Explanations were obtained for behavior in five interac- tional contexts (e.g., interactions involving discipline and interactions encouraging the child to be independent). The explanations were classified first to determine the extent to which they referred to what might have been actual thoughts at the time of the behavior. Second, they were content analysed to determine how much parenls perceived themselves as responding to the child, versus how much they presented their behavior as arising from their own dispositions, purposes, and experiences. Results showed that most of the explanations did not contain what could be consid- ered conscious cognitions at the time of the behavior. This suggests parents might have been responding automatically in the situation. Also, parents did not perceive their behavior as being mainly in response to the child. Instead they primarily saw themselves as the agents of their own actions. The latter trend varied in degree according to the interactional context. Mothers seemed more child centered in their explanations than fathers. Implications are examined for future research on links between parental cognition and behavior, and for the methods adopted to study parental cognitions. Questions are raised about the extent to which parents planfully determine their behavior and the extent to which they are aware of the reasons for their behavior. What causes parents to behave with their children the way they do? In the search for the causes of parental behavior, one factor which has received attention recently is parental cognitions, usually "beliefs." No agreed upon definition of a "belieF' has emerged (Sigel, 1985a), and therefore a wide range of parental cognitions have been studied (e.g., "values", "attitudes", and "ideas," as well as "beliefs"). These cognitions have been seen as possible mediators This research was supported by grants from the AustralianResearch Grants Scheme and from Macquarie and Hinders Universities.The authors wish to acknowledgethe considerableassistance given by Rhondda Rytmeisterand Ann EImsliein the data collection. Correspondence and requests for reprints should be sent to Alan Russell, School of Education, Hinders University,Bedford Park, S.A. 5042, Australia. 421

Mothers' & fathers' explanations of observed interactions with their children

Embed Size (px)

Citation preview

Page 1: Mothers' & fathers' explanations of observed interactions with their children

JOURNAL OF APPLIED DEVELOPMENTAL PSYCHOLOGY 9s 421-440 (1988)

Mothers' & Fathers' Explanations of Observed Interactions

with Their Children

ALAN RUSSELL

Flinders University

GRAEME RUSSELL

Macquarie University

Mothers and fathers were asked via an open-ended interview technique to explain specific interactions with their children that had been noted during a period of family observation in the home. Explanations were obtained for behavior in five interac- tional contexts (e.g., interactions involving discipline and interactions encouraging the child to be independent). The explanations were classified first to determine the extent to which they referred to what might have been actual thoughts at the time of the behavior. Second, they were content analysed to determine how much parenls perceived themselves as responding to the child, versus how much they presented their behavior as arising from their own dispositions, purposes, and experiences. Results showed that most of the explanations did not contain what could be consid- ered conscious cognitions at the time of the behavior. This suggests parents might have been responding automatically in the situation. Also, parents did not perceive their behavior as being mainly in response to the child. Instead they primarily saw themselves as the agents of their own actions. The latter trend varied in degree according to the interactional context. Mothers seemed more child centered in their explanations than fathers. Implications are examined for future research on links between parental cognition and behavior, and for the methods adopted to study parental cognitions. Questions are raised about the extent to which parents planfully determine their behavior and the extent to which they are aware of the reasons for their behavior.

What causes parents to behave with their children the way they do? In the search for the causes of parental behavior, one factor which has received attention recently is parental cognitions, usually "be l ie fs . " No agreed upon definition of a "be l i eF ' has emerged (Sigel, 1985a), and therefore a wide range of parental cognitions have been studied (e.g., "va lues" , "at t i tudes", and " ideas , " as well as "bel iefs") . These cognitions have been seen as possible mediators

This research was supported by grants from the Australian Research Grants Scheme and from Macquarie and Hinders Universities. The authors wish to acknowledge the considerable assistance given by Rhondda Rytmeister and Ann EImslie in the data collection.

Correspondence and requests for reprints should be sent to Alan Russell, School of Education, Hinders University, Bedford Park, S.A. 5042, Australia.

421

Page 2: Mothers' & fathers' explanations of observed interactions with their children

422 RUSSELL & RUSSELL

of child-rearing behavior (Dix & Grusec, 1985; Elias & Ubriaco, 1986; McGil- licuddy-DiLisi, Sigel & Johnson, 1979; Palkovitz, 1984; Sigel, 1985b).

It has been recognized that "research on parental beliefs and attitudes is not easy to conduct" (Segal, 1985, p. 271). Sigel (1986) discusses a range of potential methodologies. The usual methodology (e.g., see the various chapters in Sigel, 1985b) has involved the use of vignettes to measure parental beliefs, and structured observations to assess parental behavior. Relationships between beliefs and behavior have then been investigated.

Two limitations of the vignette procedure have been noted (e.g., Sigel, 1986). The first is that in order to standardize across parents, the vignettes used must involve hypothetical instances of behavior. The vignettes, then, may or may not include real situations experienced by each parent. Second, the interview situa- tion is removed, of necessity, from actual parental behavior. Thus, the inter- viewer has no knowledge of how the parent really would behave in the situation in question. The present research attempted to overcome these limitations by examining beliefs close to those which may have been involved in actual in-

stances of behavior. This was done by first observing parents interacting with their children, and then interviewing them about the reasons for their actions.

There is a third potential limitation of the vignette procedure (along with other interview techniques). This relates to the validity of parental responses and arises from two issues: (a) the extent to which parents are reflective about their own behavior and that of their children, and (b) the extent to which they are knowl- edgeable about or aware of the types of cognitions which mediate their behavior. In short, when parents are questioned about what they do and why, how aware and informed are they? This is an issue that arose in the present research and on which some data were presented. Concerns about parental awareness are relevant not only to researchers, but to any professional (e.g., family therapists, family court counsellors) whose aim is to determine the nature and reasons for parental behavior using interview techniques.

From the research using vignettes it appears to have been concluded that the most important parental beliefs are those dealing with children, childhood, and child development (e.g., see the list of parental beliefs presented in Table 2 by Sigel, 1986). It has then been proposed that it is beliefs about children in general or beliefs about the traits, motives or causes of their own child's behavior that are the principal cognitions mediating child-rearing behavior (Dix & Grusec, 1985 McGillicuddy-DeLisi, et al., 1979; Newberger, 1980; Sameroff & Feil, 1985; Sigel, 1985b). Parents have been presented, therefore, mainly as perceivers of, and responders to, their children.

The hypothetical situations used in the vignettes mainly focus on children and their behavior (e.g., a child refusing to share toys with a friend in McGillicuddy- DeLisi, 1982). The follow-up probes show a similar focus (e.g., "How do you think a child comes to know a b o u t . . . ?" Sigel, 1986). Sigel (1986), in fact, describes his interview procedure as aiming to elicit " . . . parents' constructions regarding child behavior" (p. 31) (emphasis added).

Page 3: Mothers' & fathers' explanations of observed interactions with their children

MOTHERS' AND FATHERS' EXPLANATIONS OF OBSERVED INTERACTIONS 423

It appears then, that even though the aim of the research has been to explain parental behavior, attention has mainly been directed at children. Furthermore, while in principle researchers have wanted to identify any cognitions which may mediate parental behavior, most emphasis has been placed on beliefs about children. In the present research, however, the focus was on cognitions relating to the behavior of the parent, in actual interactions the parent had with his or her child. In addition, the interview procedure allowed parents to give any reason as the explanation for their behavior, by using an open-ended technique with inter- viewers blind to subsequent coding categories. An advantage of having inter- viewers blind in this way is that parents are likely to be less restricted by the interviewer and hence are freer to operate on their own theoretical prejudices. Much of the research to date has classified parental beliefs into general the- oretical models (e.g., McGillicuddy-DeLisi, 1982, classified parents according to their beliefs about "constructional developmental processes"). Yet, it seems that parental cognitions and the links between these and behavior or outcomes could be highly specific to the personal characteristics of the parents (mothers vs. fathers; or according to differences in values, and past experience), the indi- vidual child (his or her personality or position in the family), the family context (e.g., social class, number of children), and especially, the particular family interactions being focused on. In the present research, parents were able to draw on any of these possible factors as the explanation for their behavior and hence were not restricted by theoretical constraints set a priori.

In summary, in the research reported here, parents were asked to explain ("why did y o u . . . ?") specific interactions noted during a period of family observation. The explanations were intended to reveal the factors parents believe led them to behave as they did. Usually these factors will be their own thoughts, purposes, beliefs, and so forth. The research advances previous work on parental beliefs in that: (a) it examined cognitions relating to actual rather than hypo- thetical behaviors; (b) its main focus was parental rather than child behavior; and (c) the interview was open rather than directing parents to beliefs about children.

The data were used to test three hypotheses. The first, which arose from the literature on beliefs, was that parents would mainly present their behavior as being a result of, or in response to, their child. To test this, parents' explanations were classified into one of two groups. In the first group were those which suggested that their behavior was in response to the child in some way, for example, the child's personality, need, or behavior. These might be said to be "external" to the parent and draw mainly on the situation or context (in this case the situation or context is almost totally made up of the child and his or her behavior). In the second group were those stemming from the parents them- selves, that is, from their own dispositions, purposes, experiences, and so forth. These might be said to be "internal" to the parent and are consistent with the parent being the agent or locus of his or her own actions.

This type of division is typical in the literature. For example, in Stolz's (1967) model of factors operating on parent-child interaction sequences, child and

Page 4: Mothers' & fathers' explanations of observed interactions with their children

424 RUSSELL & RUSSELL

parent factors provide the key elements. Gans (1962) used child and adult focus as the means of classifying families. Bugental and Shennum (1984) divided parental beliefs into those about their own power and those about child power, and Conger, McCarty, Gang, Lahey, & Krapp (1984) had as two of their three areas of maternal psychological functioning, child-rearing values and percep- tions of the child.

The second hypothesis was that parental explanations would vary according to the type of context within which the behavior occurred. Gjerde (1986), for example, called these "settings" and showed that parental behavior varied ac- cording to the interpersonal setting (e.g., whether or not the spouse was also present). There is also evidence that individual's attributions may vary as a function of the type of behavior being explained, for example, positive versus negative behavior (Dix & Grusec, 1985; Hewstone, 1983; Miller, 1986). In the present research, the perceived goal of the parents' behavior was used as the basis for classifying parent-child interactions into five contexts, for example, where the parent encouraged independence in their child, or where the child was disciplined.

The third hypothesis was that the explanations of mothers and fathers would differ. Previous research in:Vcates that mothers spend more time with their children, are more involved with them, and are generally warmer and more responsive (Belsky, 1979; Belsky, Giistrap & Rovine, 1984; Clarke-Stewart, 1978; Lytton, 1980). In Skinner's (1985) terms, this suggests mothers may be more sensitive to their child, for example, by being more likely to take into consideration the needs and stated desires of the child. It was predicted, there- fore, that mothers will be more likely than fathers to use that group of explana- tions, which suggests that the parent is responding to the child.

RATIONALE FOR THE INNOVATIVE METHODS USED IN THE PRESENT STUDY

The approach taken to the present research was based on a model of human action and its concomitant view of social science and its methods. The central arguments have been presented in a variety of sources over the past thirty years (e.g. Harre & Secord, 1972; Manicas & Secord, 1983; Scriven, 1962; Secord, 1982; Winch, 1958). Manicas & Secord (1983) describe the position on which the present work was based as a "realist theory of science." This view eschews the notion that explanations of behavior based on regular contingencies (e.g., because x always occurs before y, x causes y) are adequate as causal accounts. Such contingency relations represent what is known as a Humean conception of causality and constitutes the position of the "Standard view of science" (Man- icas & Secord, 1983). The alternative is to search for the actual mechanisms involved in the generation of human behavior. For example, an argument that because income loss is associated with punishment by parents, lower income

Page 5: Mothers' & fathers' explanations of observed interactions with their children

MOTHERS' AND FATHERS' EXPLANATIONS OF OBSERVED INTERACTIONS 425

causes parents to punish their children more, follows a Humean conception of causality. Scientific realism requires that the precise mechanisms whereby lower income leads to greater punishment need to be investigated. This may require showing, for example, how the parents' interpretations of particular child behav- iors or their goals or intentions in given disciplinary situations affect their punish- ment and how these interpretations, goals, or intentions are themselves related to lower income and other potential mediating variables such as self-esteem. That is, a realist theory requires that the steps between income loss and punishment be investigated to reveal the actual mechanisms involved in the initiation of the behavior (here punishment).

The critical assumption made in this alternative position is that much human behavior is intended, purposeful, and goal directed. Therefore, explanations of social behavior which seek the actual mechanisms underlying action should involve such things as the actor's intentions, goals, and so forth. Although there is some disagreement about the extent to which conscious intentions, goals, and other mechanisms cause behavior (Bargh, 1984; Langer, 1978; Nisbett & Ross, 1980), there is now a widely held view that much human social behavior can be explained in terms of such things as intentions, plans, goals, rules, reasons, beliefs, and others of which the actor may be consciously aware (Ajzen & Fishbein, 1980; Chapman, 1982, 1984; Kreitler & Kreitler, 1976; Manicas & Secord, 1983; Read, 1987; Schank & Abelson, 1977; von Cranach 1982.)

If research is to investigate the actual mechanisms involved in behavior, and if these mechanisms represent conscious intentions, goals, beliefs, and so forth, then the method of preference would appear to be one which focuses on the individual's actual behaviors and his or her own explanations for the behavior. This was the approach taken in the research outlined here. The explanations were taken as revealing, among other things, the cognitions (including beliefs) that parents perceive as having led them to behave as they did in the particular context in which they had been observed. The methods are consistent with trends to- wards the acceptability of introspective methods and towards the acknowledge- ment that subjects may be "experts concerning at least their understanding of their own behavior" (Masters & Carlson, 1985, p.440).

The present methods a r e innovative and there seems to be no background literature, especially in developmental psychology, to provide guidance. There appear to be two potential problems, however. The first relates to standardization across subjects. If actual behaviors are to be explained, they must vary from one subject to another. This means that the precise questions asked of each subject will also have to vary. As outlined below, efforts were taken to standardize across subjects, to the degree that this was possible given the requirement to select actual behaviors. The emphasis on actual behaviors clearly places limits on standardization. This can be contrasted with the use of vignettes, where standar- dization is guaranteed, but the behaviors dealt with must be hypothetical and may not even have been experienced by the subject.

Page 6: Mothers' & fathers' explanations of observed interactions with their children

426 RUSSELL & RUSSELL

The second potential problem concerns the nature of parents' explanations. For example, do they really know why they behaved as they do? Can they recall the intentions or goals that were involved at the time? And do they bias their explanations, for example, so as to be seen favourably in some way? This second problem raises issues not easily answered, but data bearing on them are presented below.

METHOD

Sample The subjects were 32 mothers and fathers from 32 intact families in which the eldest child (the target child) was in the second year of school, and aged 6 -7 years (mean age = 6.8 years). Half the children were boys and half were girls. The subjects were a random subsample taken from the 57 involved in a larger project (Russell & Russell, 1987). All families included in the study were se- lected so that the mother was at home from the finish of school in the afternoon. Of the families contacted, 31% agreed to participate.

The mean age of mothers was 32.3 years and fathers, 34.5 years. Thirty-one % of mothers and 59% of fathers had university degrees. All fathers were in full- time employment, with their mean occupational status ranking being 3.1 (Con- galton, 1969; this is similar to that employed by Hollingshead--e.g. , a ranking of two includes university lecturers, engineers, accountants, pharmacists; a rank- ing of three includes schoolteachers, retail managers, computer engineers). One mother was in full-time employment, and 9 were employed part-time. The mean occupational ranking for the mothers' usual paid jobs was 4.1. All couples were married, with the mean length of marriage being 9.8 years.

Procedures and Design The larger project was designed to study family relationship with special atten- tion being given to possible cognitions underlying parents' behavior. Three data collection techniques were used: (a) an initial structured interview with parents; (b) a home observation of about 2 hours--this involved all family members and took place during the early evening, with part of the observation being natu- ralistic (and included the evening meal) and part being semi-structured where family members were asked to play a ring-toss game; and (c) a follow-up flexible interview--this took place immediately after the observation and the parents were interviewed separately. The main purpose of the interview was to obtain parents' explanations of selected specific interactions noted during the observa- tion. In addition, the interview sought parents' own interpretations of the behav- ior coded during the observation. This was done to validate the coding categories and to assess the extent to which observers and parents agreed on the interpreta- tion of behavior.

Page 7: Mothers' & fathers' explanations of observed interactions with their children

MOTHERS' AND FATHERS' EXPLANATIONS OF OBSERVED iNTERACTIONS 427

To investigate the role of the context, parent's explanations were obtained about five "contexts." The "context" was defined in terms of the perceived (to the observer) goal of the parent. Although initially defined from the observers viewpoint, the follow-up interview revealed almost unanimous agreement with parents about the classification of interactions in this way. The five contexts related to the parent: (a) encouraging the child's independence, that is, encourag- ing the child to do things for himself or herself or make his or her own decisions; (b) enabling or encouraging dependence in the child (this usually meant doing things for the child that the observer judged the child was easily capable of doing himself or herself; this view of dependence and the above definition of indepen- dence is consistent with the work of Belier, 1955); (c) fostering achievement or competitive behavior in the child; (d) discouraging competitive behavior or achievement in the child (this usually involved interactions in which the parent "played-down" such things as winning or beating others at school, in games, or in sports); and (e) disciplining, controlling, or correcting the child's behavior.

Explanations about behavior in each of these contexts were obtained from both parents. Insofar as each parent was asked about behavior from each context, the interview procedure was standardized to some degree. However, be~:ause the interview dealt with actual behaviors, the exact actions discussed varied from parent to parent. In a further effort to standardize, where possible an example of the same type of behavior was used with each family (given that it had been observed). Many parents corrected the child's behavior during dinner (although the style might have been different) and supervised the child during a bath or shower. Where these interactions occurred, parents were questioned about them. In practice, then, there were considerable commonalities between interviews.

The observer selected the behaviors to be explained at the completion of the observation, and just prior to the follow-up interview. They were instructed to select up to three examples of behavior in each of the five contexts. This was not always possible, however, as interactions associated with either dependence, competition, or noncompetition did not occur in some families. Consequently, more behaviors in the independence and discipline contexts than in the other three contexts were discussed in the interview.

Two assistants were used in the research and the same person did the observa- tion and interview. Both assistants received extensive training during pilot re- search on 20 families. The training covered the selection of interactions as "pure" as possible within the five contexts, as well as the techniques of inter- viewing in an open-ended style (e.g. in the use of nondirective probing). Inter- view technique was monitored throughout the research by one of the principal investigators.

While explaining pieces of observed behavior, parents sometimes mentioned similar behaviors used at other times. For instance, when explaining his or her emphasis on winning during the ring-toss game, the parent may mention his or

Page 8: Mothers' & fathers' explanations of observed interactions with their children

428 RUSSELL & RUSSELL

her efforts in getting the child involved in football. If the parent outlined a specific behavior in such cases, the interviewer was requested to seek an expla- nation and the data were included in the analysis.

The interview generally took the form of bringing the parent's attention to an incident or behavior that occurred during the observation and then asking the parent "Why did you do x?" or "How do you explain doing y?" Probing of answers occurred when there was a need to clarify inconsistencies or possible confusions. However, the purpose was not to exhaust all the parents' knowledge about the behavior or to probe repeatedly to obtain the ultimate belief, and so forth, underlying the behavior. Instead, the interview occurred at a level close to ordinary conversation where an interested other wishes to clarify reasons for behavior. The interviewers were blind to the way the explanations were coded and to the specific hypotheses tested.

Coding of the Interview The responses were treated by first identifying the behavior being explained and then listing the separate explanations given. The unit coded was always a behav- ior plus an explanation. A behavior without an explanation, or beliefs and at- titudes stated but not tied to any specific piece of behavior were not coded. If the same explanation was given several times for one piece of behavior, it was coded once. If the same explanation was given for a different behavior, it was scored again, as a separate behavior-explanation unit. At times, multiple explanations were given for one behavior. These involved separate identifiable classes of reasons. For example, the parent may first focus on the child ("he is lazy and would have us do things for him all the time") and shift to a parent-based explanation ("we think it important he learn to do things for himself, to be independent"). In coding explanations, an attempt was made to retain ordinary language meaning levels. Thus, each sentence was not segmented into indepen- dent clauses (as done by Miller, 1984, 1986). A complex idea with an example was coded as one explanation. Separate explanations had to contain identifiably different ideas, for example, "he is lazy," " I want him to do things for him- self," and "we have found from experience this is the best way to get him to do things." Decisions about whether or not a parent's answer involved separate explanations were helped by determining whether or not a single code from the coding systems used (see below) could be applied to the answer. Although complex and multiple ideas occurred in some answers, most of the units coded took the relatively straightforward form of " I did x because y" or "The reason I did this was z" or " I did x so that W" (where W is a goal or aim).

The explanations obtained were coded using two coding schemes--the first to investigate the nature or form of the explanations, the second to examine content.

The Nature of the Explanations Given. As noted in the rationale for the methods, one issue associated with the present methods concerns the nature of

Page 9: Mothers' & fathers' explanations of observed interactions with their children

MOTHERS' AND FATHERS' EXPLANATIONS OF OBSERVED INTERACTIONS 429

the explanations given. Two possibilities were apparent: (a) the parent attempts to reconstruct the situation in which the behavior occurred and give his or her actual thoughts, at the time, which gave rise to the behavior, and (b) the parent interprets the task as requiring a general justification or rationale, more for why he or she did (or do) the type of thing that was observed, that is, the parent thinks he or she should give a justification for his or her strategy or role in general terms.

A coding scheme with two categories was devised to determine the extent parents used explanations of these two forms. Situational explanations drew on the here-and-now situation in which the behavior occurred. These explanations covered such things as exactly what the parent was trying to achieve, expressed in a way which did not go beyond the given situation, for example, the parent helped the child get dressed "so she would get dressed quickly and not get cold," or the parent stopped the child from helping cook dinner "because he might burn himself." It is possible to see these as representing the precise thoughts at the time of the action, those which actually led to the behavior.

General explanations ignored the specifics of the situation by drawing on general rules, goals, principles, and so forth, that is, they covered a wide range of situations. This suggests thc action was treated as only an exemplar for a class of similar actions or situations, for example, a parent insisted the child dress himself "because it is important that he be independent and learn to do things for himself," or a father told his child to get a towel for himself when he called for help "because he is relying on us too much." Here the parents seemed to give a general rationale for a class of action. These seemed more consistent with parents attempting to support or justify what they had done than giving a specific cause which occurred at the time of the behavior.

To compare situational versus general explanations, a selection was made from all the explanations given. This was done to ensure that only those explana- tions were analyzed when in principle either type could have been given. This occurred when (a) it was obvious from the transcript that the interviewer had observed the behavior, (b) the behavior was drawn fully and specifically to the attention of the parent so that it was clear both parent and interviewer knew the incident being referred to, and (c) the question asked of parents was exactly of the form "why did y o u . . . ?"

The Content of Explanations. An initial framework for the content analysis of the explanations was provided by Miller's (1984) scheme, where a distinction is made between references to (a) "aspects of the agent undertaking the behav- ior," which in the present case refers to the parent, and (b) "the context." In Miller's scheme, aspects of the agent included such things as the parent's values, interests, personality, abilities, and feelings in the specific situation. This means, for example, a person may explain the fact that he or she punished the child in terms of the priority that parent places on well-behaved children. Agent explana-

Page 10: Mothers' & fathers' explanations of observed interactions with their children

430 RUSSELL & RUSSELL

tions are in many ways "internal." In Miller's classification much of what others refer to as "external" factors are covered by the "context ." The context includes "aspects of persons other than the agent" (p.965), for example, a parent might explain punishing the child in terms of the child being naughty. Clearly, the main element of the context in the present research was the child with whom the parent interacted. In summary, agent explanations mainly draw on factors internal to the parent, while context explanations use factors outside the parent, usually, the child.

Consistent with Miller's division, the present coding scheme was initially divided into three broad categories: (a) the parent, (b) the child, and (c) the situation. "The situation" also represents part of the context. It was separated from those explanations which used the child as the cause of the behavior for the purposes of completeness, although it was used on very few occasions.

The data were divided into separate behavior-explanation units and then coded into one of eight explanation categories (see Table I). After the three-way division into parent, child, and situation types, the further classification was based partly on other analyses of accounts (Harre, 1979), beliefs (Stolz, 1967), values (Rokeach, 1970), and explanations (Antaki, 1981; Miller, 1984). The eight categories were as follows:

(1) Parental characteristic: Description. Here there were no underlying be- liefs, goals, and so forth. The parent virtually said only "This is what I do" or "This is how I do it ." For example, a mother explained not setting up any rules for the ring-toss game by saying "This is the way we first approach all games."

(2) Parental characteristic: Goals, principles, rules. These explanations typ- ically involved statements about parents' aims, wishes, goals, and so forth, general child-rearing principles and/or personality characteristics. The lat- ter refer to things the parents recognize about themselves or their person- ality, for example, they may say " I ' m strongly competitive." Examples are a father who made his child tidy the living room and said, " I believe children should learn responsibility at an early age"; A mother who turned the TV off said, "We have a rule there is no TV watching during meals." These are explanations which draw directly on factors "internal" to the parent. They are stronger than those in category 1, but not as definite or fully-developed as those in category 3.

(3) Parent child-rearing value. A child-rearing value was taken as a substantial set of ideas, ideals, beliefs, goals, and so forth, that came together into a consciously planned strategy for child rearing. A child-rearing value was something which linked a range of parental behaviors in the area of the value. In many ways, a value here was like what Rokeach (1970) has called a value system. It was scored when parents made strong statements about the significance of the behavior in question, for example, "I t is impor- t a n t . . . " or " I v a l u e . . . " . Statements such as " I like . . . " or " i

Page 11: Mothers' & fathers' explanations of observed interactions with their children

MOTHERS' AND FATHERS' EXPLANATIONS OF OBSERVED INTERACTIONS 431

w a n t . . . " were not seen as strong enough and were classified as (2) above. A second factor taken into account in scoring a value was whether or not a variety of support statements were made around the central idea. Although a variety of statements may be made, only one value was scored unless the same value was used to explain different pieces of behavior.

Explanations (l), (2), and (3) fall on a continuum and at times it was difficult to distinguish (1) from (2), and (2) from (3). Some difficulty of this kind is to be expected (Leddo, Abelson, & Gross, 1984).

(4) Parent: experience. This was scored when a parent explained his or her behavior in terms of something learned from a book, formal education, the parent's own childhood, or prior experience with the child, for example, " I always give him a warning first because we have found that works best with him."

(5) Parent: need. These were explanations where a short- or long-term need from the parent was used to account for his or her behavior, for example, fostering independence to "free the mother," or doing things for the child because it is "quicker and easier."

(6) Child: characteristic. Here the behavior was explained in terms of some- thing about the child, his or her behavior, personality, what he or she had done, for example, a child was allowed to be independent because "he likes doing things for himself."

(7) Child: need. Here the parent implied that he or she had little choice but to behave as he or she did because of some child need, for example, the child's safety, well-being, and so forth. For example, the parent ran the bath for the child " to make sure the water wasn't too hot ."

(8) Situation. This applied when a parent gave some transitory or special fea- ture of the circumstance (for example, " I brushed his teeth for him because the dentist said I should") as the reason for the parent's behavior.

Reliability. The reliability for the division into situational and general expla- nation was based on data from a second coder who coded 48 randomly selected explanations. Cohen's kappa for the coding into these two categories was .96. This shows a clear separation between these two forms of explanation. For the content analysis, a second person coded eight randomly selected families. This involved a total of 213 behavior-explanation units. The reliability coefficients (Cohen's kappas) are given in Table 1. They show moderate to high reliabilities.

RESULTS

The results from the situational versus general coding are presented first, fol- lowed by those from the content analysis.

Page 12: Mothers' & fathers' explanations of observed interactions with their children

432 RUSSELL & RUSSELL

TABLE 1 Mean Number end Percentage of Explanations for Mothers and Fathers,

and Reliabilitles

Mothers Fathers M

Explanation N % N % N % Reliability

1. Parent: Descriptive 1.84 11.7 1.62 11.0 1.74 11.5 .97 2. Parent: Goal/Rule 3.97 25.3 4.37 29.8 4.16 27.4 .83 3. Parent: Value 1.04 6.6 1.04 7.1 1.04 6.9 .94 4. Parent: Experience 0.56 3.6 0.40 2.7 0.48 3.2 1.00 5. Parent: Need 1.37 8.7 1.12 7.6 1.25 8.2 .77 6. Child: Characteristic 2.57 16.4 1.89 12.9 2.23 14.7 .96 7. Child: Need 3.88 24.8 3.82 26.0 3.85 25.4 .94 8. Situation 0.43 2.7 0.43 2.9 0.43 2.8 1.00

TOTAL 15.67 14.69 15.18

The Nature of the Explanations Given. There were 260 behaviors which satisfied the criteria for situational versus general coding. For these behaviors, a total of 409 explanations were given. 155 (37.9%) were of the situational type and 254 (62.1%) were of the general type. Of the 155 situational explanations, 97 (62.6%) were given in the form of a single explanation for a behavior, while of the 254 general explanations, only 74 (29.1%) were given as a single explana- tion. That is, most of the general explanations were used in giving multiple explanations for an action.

The Content of Explanations. The mean number of explanations, and per- centages, for mothers and fathers in each of the eight content categories are given in Table 1. It can be seen that on average parents were scored for between 15 and 16 behavior-explanation units per interview. The most common explanation was Parent: goal/rule, followed by Child: need. Parents used a child-rearing value on average about once per interview. A total of 971 explanations were given. Of these, 57% were of the "Parent" type (explanations 1-5), 40% were of the "Child" type (explanations 6 and 7), while 3% were based on the "Situation." The data on interactional context are presented in Table 2. Here the eight expla- nation types have been grouped into types 1-5 (Parent), types 6 and 7 (Child), and type 8 (Situation).

It can be seen in Table 2 that the proportionate use of the three explanation groups (Parent, Child, Situation) varied according to the interactional context. In the context of dependence interactions, for example, parents gave more "chi ld" explanations, while for noncompetition and discipline interactions they gave more "parent" explanations. The mean percentage use of "Parent" explana- tions in Table 2 is 54.6%. This compares with an overall percentage of 57%

Page 13: Mothers' & fathers' explanations of observed interactions with their children

MOTHERS' AND FATHERS' EXPLANATIONS OF OBSERVED INTERACTIONS

TABLE 2 Proportionate Use of Explanations Besed on the Perent, the Child

and the Situation for Each Interactional Context

433

Explanation Type

interaction Context Parent Child Situation

Independence 56.5 42.3 1.2 Dependence 35.2 60.0 4.8 Achievement/competition 50.9 46.6 2.5 Non-Achievement/competition 64.3 34.6 1.1 Discipline/control 66.3 30.7 3.0

Mean percentage 54.64 43.64 2.53

based on the raw scores. The figure in Table 2 is lower because the procedure of calculating percentages for each interactional context, and then a mean percent- age, gives equal weight to each of the five contexts. However, it emerged that fewer explanations were given in the dependence context (dependence-type in- teractions did not occur in all families during the observations) so that the bias away from "parent" explanations shown in the dependence context has less impact on the overall percentage from the raw scores.

To test for differences apparent in Table 2 a repeated measures multivariate analysis of variance was conducted comparing sex of parent, interactional con- text, and explanation type. The proportions were transformed using the arcsin transformation (Winer, 1962). There was no effect for sex of parent. However, the effects for explanation type [F(2,30) = 306.52, p < .01)] and the interaction of explanation type and interactional context [F(8, 24) = 4.98, p < .01)] were significant. Follow-up t tests showed that overall, significantly more "parent" explanations were used than "chi ld" explanations (t = 5.13, p < .01). For the individual interactional contexts, more "parent" than "chi ld" explanations were used for independence (t = 2.67, p < .01), non-achievement/competition (t = 4.05, p < .01) and discipline/control (t = 6.17, p < .01). The effect for "parent" versus "chi ld" explanation types was not significant for the depen- dence context, despite the apparent magnitude of the difference.

To examine the third hypothesis, that mothers are more child-centered, moth- er's and father's use of "chi ld" explanations were compared (transformed pro- portion scores). Mothers gave proportionally more "chi ld" explanations (t = 2.02, p < .05, 1-tailed). Separate tests on the two "chi ld" explanation types revealed mothers used more Child: characteristic explanations (t = 1.79, p < .05, 1 tailed), but not more Child: need explanations. Inspection of the data showed mothers used proportionally more Child: characteristic explanations in each of the five interactional contexts except that dealing with discipline.

Page 14: Mothers' & fathers' explanations of observed interactions with their children

434 RUSSELL & RUSSELL

DISCUSSION

There appear to be three major findings from the present study: (a) explanations were more often of the "parent" type than the "chi ld" type: (b) the type of explanation given varied according to the nature of the interaction that was being explained; and (c) the explanations given were more usually in a form which suggested the parent was not attempting to reconstruct the situation mentally and give actual thoughts, beliefs, intentions and so forth, which might have gener- ated the behavior. Insofar as the last finding appears to have implications for how the present data are to be interpreted, it will be considered first.

The "situational" versus "general" analysis of the explanations showed that the parents largely provided a nonspecific justification or rationale for their behavior. This was in spite of the interviewer drawing a specific behavior to their attention and asking "why did y o u . . . ?" There are two possible reasons for parents giving general explanations:

1.

.

Because most of the behaviors asked about are likely to occur often, the parent interpreted the task as one in which he or she should justify his or her behavior in a broad way, to explain the kind of relationship the parent has with his or her child. That is, parents could have reinterpreted the question in the belief that that was the intention of the interviewer. Why they would do that most of the time, but at other times answer the question directlt as a specific one about a precise behavior, is not clear. They were not consciously monitoring what they were doing at the time of the behavior. When asked for an explanation then, there were no specific thoughts, beliefs, intentions and so forth which were present at the time of the behavior and which could be recalled and outlined. This assumes that most routine or day-to-day interactions (the kind that dominated during the family observation) occur fairly automatically (cf Langer, 1978; Bargh, 1984), that is, the parent behaves without the conscious involvement of things such as an assessment of the situation, consideration of beliefs, and the evaluation of goals. A similar distinction was made by Kuczynski (1984) when he argued that in some situations parents choose strategies to obtain child compliance planfully, while in other situations they "react less con- sciously" or "automatically" (P. 1071).

When parental behavior is of the automatic type, it can be questioned in what sense it is cognitively mediated. Clearly, it is not mediated by conscious cogni- tions. One possibility is that many of the mediating cognitions (e.g., beliefs) could be implicit and accessible via special probing or other techniques. Some of the general explanations given by parents might represent beliefs and other cognitions which were used in the original formulation of the behavior (cf Lan- ger, 1978) and at times throughout the course of their relationship with the child,

Page 15: Mothers' & fathers' explanations of observed interactions with their children

MOTHERS' AND FATHERS' EXPLANATIONS OF OBSERVED INTERACTIONS 435

but are no longer conscious. Techniques other than conscious recall of mediating cognitions would be needed to determine these beliefs. Others of the general explanations might never have been actually used in the selection of behavior.

When a behavior is automatic and an explanation is subsequently given for it, the explanation appears open to conscious and unconscious bias. There is evi- dence from the attribution literature of a tendency to give attributions which present the individual in a favourable light (Cashmore & Goodnow, 1986; Miller & Ross, 1975). Equally, Robbins (1963) has shown that inaccuracies in parents' reports of child rearing "tended to be in the direction of the recommendations of experts in child r e a r i n g . . . " (p.261). It is possible, then, that the general explanations given here could be biased in some way, for example, towards culturally accepted viewpoints. The fact that concrete instances of behavior that had occurred very recently were used might reduce the bias, but bias is still possible and difficult to assess.

Concerning the content of explanations, parents were more likely to draw on their own dispositions, purposes, goals, experiences, and so forth, than factors associated with the child and his or her behavior. At first glance this suggests that the predominant focus in the research to date, that is, on parents responding principally in terms of their beliefs about children or their interpretations of the child's behavior, may reflect on a priori bias towards this type of finding. It seems instead, from the present data, that there is a wider diversity of factors leading parents to behave the way they do than is implied by the view that parents mainly respond to their children. Certainly more recognition needs to be given to the notion that parents can be agents or initiators of their own actions. Much of the literature seems to portray parents as reactors, when in fact they are mainly interactors.

A cautionary note about the finding that parents used more "parent" than "child" explanations is engendered by the fact that the use of the various explanation types varied as a function of the kind of interaction being explained. Parents were more "parent" oriented in their explanation when explaining in- teractions associated with independence, non-achievement/competition and dis- cipline. There was no significant difference between "parent" and "chi ld" explanations for interactions associated with achievement/competition, while there were more "chi ld" explanations for dependence interactions (but this difference was not statistically significant). It appears, then, that not only might parent-child interaction vary as a function of the context (e.g. Gjerde, 1986), but the reasons for parental behavior might also shift according to the context.

In considering the fact that parents (overall) used more "parent" than "child" explanations, the interview context itself needs to be taken into account as a possible factor. Lalljee (1981) and Cantor & Brown (1981) have claimed that the context within which explanations are given is an important influence on the nature of the explanation, for example, " . . ordinary explanations are given for a variety of reasons and take on a wide range of forms" (Canter & Brown,

Page 16: Mothers' & fathers' explanations of observed interactions with their children

436 RUSSELL & RUSSELL

1981, p.221). In the present case, the procedure could be seen as creating a "self-focusing" situation for parents, one where the emphasis is on themselves qua parents, rather than on the child. It has been found that " in situations that have been structured to direct the individual's attention inward, on the self, such internal sources of information as attitudes, traits, and values will tend to become s a l i e n t . . . " (Snyder & Ickes, 1985, P.908). Although parents were free to provide any explanation, and many of the questions in the follow-up interview focused on the child (data not presented here), it is possible that the results partly reflect a tendency for the procedure to make dispositional explanations more salient. The fact that parents gave fewer dispositional explanations for depen- dence interactions, of course, creates some difficulties for this interpretation of the results.

The research context could also have placed parents in a situation of perceived evaluation of their parenting. This could have encouraged a tendency to give explanations based on cultural norms and expectations about the behavior of "good" or "competent" parents (a tendency suggested by Bernard, KiUworth, Kronenfeld & Sailer, 1984, p.508, about how informants respond to questions about social situations they had recently been in). Such cultural norms might imply that a parent should be seen as taking the initiative in matters like disci- pline and that a "good" parent has dispositional reasons for not encouraging dependence in 6- and 7-year-olds. Whether or how much individual parents were responding because of such norms it is impossible to answer. When a parent says " I make him eat properly with his knife and fork because I want him to grow up with good manners," for example, it is difficult to determine the extent to which the parent genuinely has this goal or said it (and did it) so as to present him or herself more favourably as a "good" parent.

Insofar as mothers used more explanations within the category child: charac- teristic, the results provided some support for the hypothesis that mothers were more child-oriented than fathers. Why they did not also use the category child: need more is not clear, although the "need" category did relate to more obvious needs of the child while the "characteristic" category dealt with subtle features of the child such as personality, likes and dislikes, and so forth. That mothers used the child: characteristic category more than fathers could be due to more experience or contact with the child and hence greater knowledge of the child (Goodnow, Knight & Cashmore, 1985). However, the finding on mother-father differences was not strong and needs further investigation.

The present methodology seems to have advantages and disadvantages in comparison to the vignette technique. The major disadvantage is the lack of standardization, so that no two parents are questioned about precisely the same behavior. This makes some comparisons between parents difficult, and detailed assessment of how and why parents would respond in given situations (e.g., child refusal to go to bed when told) virtually impossible. The advantages arise from the ecological validity of the entire procedure (parents were questioned

Page 17: Mothers' & fathers' explanations of observed interactions with their children

MOTHERS' AND FATHERS' EXPLANATIONS OF OBSERVED INTERACTIONS 437

about actual observed behaviors) and the freedom given to parents to present any

reason for their behavior. It was found that there may be a wider range of factors (especially factors internal to the parent) causing parents to behave in the way they do than appears to have been recognized in the literature to date. It may be that the vignette procedure has led to a premature narrowing of the research focus before sufficient exploratory research of the kind outlined here has been con- ducted.

Although the present research was designed to examine the content of parental explanations, the nature and validity of the parental responses emerged as an issue during data collection. The subsequent coding of responses into situational versus general explanations was done to clarify the degree to which parents present conscious cognitions at the time of their behavior as the explanation for their actions. The data, together with supporting ideas from the literature, prompted a conclusion that much parental behavior is reasonably automatic, with no mediating cognitions of a conscious kind. This means that in many situations parents may not know what, if any, beliefs or cognitions mediate their behavior.

If the latter conclusion is accepted, it poses difficulties for research directed at the link between behavior and beliefs. It also has implications for anyone work- ing in applied settings who endeavours to use interviews to gain information from parents. Any assumption that parents always deliberately and planfully determine their behavior with their children, and then are able accurately to reflect on their behavior and the reasons for it, needs to be questioned. Re- searchers and other professionals should be cautious and possibly interpret as- pects of parental accounts of their own behavior as little more than constructs generated retrospectively so as to provide culturally plausible or acceptable rea- sons which justify or make sense of their behavior (Bernard, Killworth, Kronen- feld, & Sailer, 1984; Gilbert & Mulkay, 1983).

REFERENCES

Aj~n , I°, & Fishbein, M. (1980). Understanding attitudes and predicting social behaviour. En- glewoud Cliffs, NJ: Prentice-Hall.

Antaki, C. (Ed.). (1981). The psychology of ordinary explanations of social behavior. London: Academic.

Bargh, J.A. (1984). Automatic and conscious processing of social information. In R.S. Wyer, Jr. & T.K. Srull (Eds.), Handbook of social cognition (Vol. 3). Hillsdale, NJ: Edbanm.

Belier, E.K. (1955). Dependency and independence in young child~n. Journal of Genetic Psychol- ogy, 87, 25-35.

Belsky, J. (1979). Mother-father-infant interaction: A naturalistic observational study. Develop- mental Psychology, 15, 601-607.

Belsky, J., Gilstrap, B., & Rovine, M. (1984). The Pennsylvania infant and family development project. I: Stability and change in mother-infant and father-infant interaction in a family setting at one, three and nine months. Child Development, 55, 692-705.

Bernard, H.R., Killworth, P., Kronenfeld, D. & Sailer, L. (1984). The problem of informant accuracy: The validity of retrospective data. Annual Review of Anthropology, 13, 495-517.

Page 18: Mothers' & fathers' explanations of observed interactions with their children

438 RUSSELL & RUSSELL

Bugental, D.B., & Shennum, W.A. (1984). "'Difficult" children as elicitors and targets of adult communication patterns: An attributional behavioral transactional analysis. Monograph of the Society for Research in ChiM Development, 49( I, Whole No. 205).

Cantor, D., & Brown, J. (1981). Explanatory roles. In C. Antaki (Ed.), The Psychology of ordinary explanations of social behavior. London: Academic.

Cashmore, J.A., & Goodnow, J.J. (1986). Parent-child agreement on attributional beliefs. Interna- tional Journal of Behavioral Development, 9, 191-204.

Chapman, M. (1982). Action and interaction: The study of social cognition in Germany and the United States. Human Development. 32, 295-302.

Chapman, M. (Ed.) (1984). International action as a paradigm for developmental psychology: A symposium. Human Development, 27, 113-144.

Clarke-Stewart, K.A. (1978). And daddy makes three: The father's impact on mother and young child. Child Development. 49, 466-478.

Congalton, A.A. (1969). Status and prestige in Australia: Studies in Australian society. Melbourne, Australia: Cheshire.

Conger, R.D., McCarty, J.A., Yang, R.K., Lahey, B.B., & Krapp, J.P. (1984). Perception of child, child-rearing values, and emotional distress as mediating links between environmental stressors and observed maternal behavior. Child Development, 55, 2234-2247.

Dix, T.H., & Gnasec, J.E. (1985). Parent attribution processes in the socialization of children. In I.E. Sigel (Ed.), Parental belief systems: The psychological consequences for children. Hillsdale, NJ: Erlbaum.

Elias, M.J., & Ubriaco, M. (1986). Linking parental beliefs to children's social competence: Toward a cognitive-behavioral assessment model. In R.D. Ashmore and D.M. Brodzinsky, (Eds.), Thinking about the family: Views of parents and children. Hillsdale, N J: Erlbaum

Gans, H.J. (1962). The urban villagers: Group and class in the life of ltalian-Americans. New York: Free Press.

Gilbert, G.N., & Mulkay, M. (1983). In search of action. In G.N. Gilbert & P. Abell, (Eds.), Accounts and action. Aldershot: Gower.

Gjerde, P.F. (1986). The interpersonal structure of family interaction settings: Parent-adolescent relations in dyads and triads. Developmental Psychology, 22, 297-304.

Goodnow, J., Knight, R., & Cashmore, J. (1985). Adult social cognition. Implications of parents' ideas for approaches to development. In M. Perlmutter (Ed.), Minnesota Symposia on Child Development (Vol. 18). Hillsdale, N J: Erlbaum.

Harre, R. (1979). Social being: A theory for social psychology. Oxford, England: Basil Blackwell. Harre, R., & Secord, P.F. (1972). The explanation of social behavior. Oxford, England: Basil

Blackwell. Hewstone, M. (Ed.). (1983). Attribution theory: Social and functional extensions. Oxford, England:

Basil Blackwell. Kreitler, H., & Kreitler, S. (1976). Cognitive orientation and behavior. New York: Springer. Kuczynski, L. (1984). Socialization goals and mother-child interaction: Strategies for long-term and

short-term compliance. Developmental Psychology, 20, 1061-1073. Lalljee, M. (1981). Attribution theory and the analysis of explanations, in C. Antaki (Ed.), The

psychology of ordinary explanations of social behavior. London: Academic. Langer, E.J. (1978). Rethinking the role of thought in social interaction. In J.H. Harvey, W.J. Ickes,

& R.D. Kidd (Eds.), New directions in attribution research (Vol. 2). Hillsdale, N J: Erlbaum. Leddo, J., Abelson, R.P., & Gross, H. (1984). Conjunctive explanation: When two reasons are

better than one. Journal of Personality and Social Psychology, 47, 933-943. Lytton, H. (1980). Parent-child interaction: The socialization process observed in twin and sin-

gletonfamilies. New York: Plenum. Manicas, P.T., & Secord, P.J. (1983). Implications for psychology of the new philosophy of

science. American Psychologist, 38. 399-413.

Page 19: Mothers' & fathers' explanations of observed interactions with their children

MOTHERS' AND FATHERS' EXPLANATIONS OF OBSERVED INTERACTIONS 439

Masters, J.C., & Carlson, C.R. (1985). Children's and adults' understanding of the causes and consequences of emotional states. In C.E. Izard, J. Kagan, & R.B. Zajonic (Eds.), Emotions, cognition and behavior. Cambridge, England: Cambridge University Press.

McGillicuddy-I)eLisi, A.V. (1982). Parental beliefs about developmental processes. Human Devel- opment, 25, 192-200.

McGillicuddy-DeLisi, A.V., Sigel, I.E., & Johnson, J.E. (1979). The family as a system of mutual influences: Parental beliefs, distancing behaviors, and children's representational thinking. In M. Lewis & L.A. Rosenblum (Eds.), The child and its family. New York: Plenum.

Miller, D.T., & Ross, M. (1975). Self-serving biases in the attribution of causality: Fact or fiction? Psychological Bulletin, 82, 213-225.

Miller, J.G. (1984). Culture and the development of everyday social explanation. Journal of Person- ality and Social Psychology, 46, 961-978.

Miller, J.G. (1986). Early cross-cultural commonalities in social explanation. Developmental P.Ev- chology, 22, 514-520.

Newberger, C.M. (1980). The cognitive structure of parenthood: Designing a descriptive measure. In R.L. Selman & R. Yando (Eds.), New directions for child development, No. 7. San Francis- co: Jossey-Bass.

Nisbett, R., & Ross, L. (1980). Human inference: Strategies and shortcomings of social judgement. Englewood Cliffs, NJ: Prentice-Hall.

Palkovitz, R. (1984). Parental attitudes and fathers' interactions with their 5-month-old infants. Developmental Psychology, 20, 1054-1060.

Read, S.J. (1987). Constructing causal scenarios : A knowledge structure approach to causal reason- ing. Journal of Personality and Social Psychology, 52, 288-302.

Robbins, L.C. (1963). The accuracy of parental recall of aspects of child development and of child rearing practices. Journal of Abnormal and Social Psychology, 66, 261 - 270.

Rokeach, M. (1970). Beliefs, attitudes and values: A theo~" of organization and change. San Francisco: Jossey-Bass.

Russell, G., & Russell, A. (1987). Mother-child and father-child relationships in middle childhood. Child Development, 58, 1573-1585.

Sameroff, A.J., & Feil, L.A. (1985). Parental concepts of development. In I.E. Sigel (Ed.), Parental belief systems: The psychological consequences for children. Hillsdale, NJ: Erlbaum.

Schank, R.D., & Abelson, R.P. (1977). Scripts, platte, goals and understanding. Hillsdale. NJ: Erlbaum.

Scriven, M. (1962). Explanations, predictions and laws. In H. Feigl & M. Scriven (Eds.), Minnesota studies in the philosophy of science, Vol. 1. Minneapolis: University Minnesota Press.

Secord, P.J. (Ed.). (1982). Explaining social behavior: Consciousness, human action and social structure. Beverly Hills, CA: Sage.

Segal, M. (1985). A study of maternal beliefs and values within the context of an intervention program. In I.E. Sigel (Ed.), Parental belief systems: The psychological consequences for children. Hillsdale, NJ: Erlbaum.

Sigel, I.E. (1985a). A conceptual analysis of beliefs. In I.E. Sigel (Ed.), Parental belief systems: The psychological consequences for children. Hillsdale, N J: Erlbaum.

Sigel, 1. E. (Ed.). (1985b). Parental belief systems: The psychological consequences for children. Hillsdale, NJ: Erlbaum.

Sigel, I.E. (1986). Reflections on the belief-behavior connection: Lessons learned from a research program on parental belief systems and teaching strategies. In R.D. Ashmore & D.M. Brodzinsky, (Eds.), Thinking about the family: Views of parents and children. Hillsdale, NJ: Erlbaum.

Skinner, E.A. (1985). Determinants of mother sensitive and contingent-responsive behavior: The role of child rearing beliefs and socioeconomic status. The I.E. Sigel (Ed.), Parental belief systems: The psychological consequences for children. Hillsdale, NJ: Erlbaum.

Page 20: Mothers' & fathers' explanations of observed interactions with their children

440 RUSSELL & RUSSELL

Snyder, M., & Ickes, W. (1985). Personality and social behavior. In G. Lindzey & E. Aronson (Eds.), Handbook of Social Psychology (3rd ed.), Vol. H: Special Fields and Applications. New York: Random House.

Stolz, L.M. (1967). Influences on parent development. Stanford, CA: Stanford University Press. von Cranach, M. (1982). The psychological study of goal-directed actions: Basic issues. In M. von

Cranach & R. Harre (Eds.), The analysis of action: Recent theoretical and empirical ad- vances. Cambridge: Cambridge University Press.

Winch, P. (1958). The idea of a social science. London: Routledge & Kegan Paul. Wirier, B.J. (1962). Statistical principles in experimental design. New York: McGraw-Hill.