© 2001 The International Society for the Study of Behavioural Development
International Journal of Behavioral Development 2001, 25 (1), 50–59 http://www.tandf.co.uk/journals/pp/01650254.html
DOI: 10.1080/01650250042000069
Reasoning and action complexity: Sources and consequences on maternal child-rearing behaviour Maria J. Rodrigo
Jan M.A.M. Janssens
Esperanza Ceballos
University of La Laguna, Spain
University of Nijmegen, The Netherlands
University of La Laguna, Spain
In this study, two approaches of complexity of maternal thinking were examined: Newberger’s (1980) analysis of reasoning complexity, and Vallacher and Wegner’s (1985) analysis of action complexity, that was adapted to the parenting eld. Measures for action complexity and for maternal behaviours were derived from hypothetical child-rearing situations and from mother-child observed interactions. Links between reasoning and action complexity to child-rearing actions and to sources, such as socioeconomic status (SES) were explored in a Spanish sample of 60 mothers with a child 7 to 10 years old. Moreover, it was proposed that the relation between SES and maternal child-rearing behaviour was mediated by either form of complexity. A LISREL model for the hypothetical situations showed that the in uence of SES on less positive child rearing was only mediated by Reasoning complexity, whereas Action complexity showed an independent path to positive childrearing behaviour. A LISREL model for the observed situations showed that both maternal reasoning complexity and action complexity played an equivalent mediating role between maternal SES and negative child-rearing behaviour. The theoretical and methodological implications of both aspects of complexity to the analysis of child-rearing behaviour were discussed.
Introduction Within analyses of social cognition one of the major concerns is to nd ways to specify the structural quality of thinking in addition to the content of thought. The same distinction between content and quality of ideas has been put forward in the parenting eld (Goodnow & Collins, 1990). Content refers to differences in the substance of ideas (e.g., the relative contributions of heredity and environment; the qualities parents want to see their child develop), whereas quality refers to those characteristics that may cut across content: The degree of certainty, accuracy, complexity, explicitness, and consistency in the structure of the ideas. In this study, we focus on the complexity of maternal thinking. Complexity is an interesting property to consider because it has been linked to differences in actions and to sources such as socioeconomic status (SES). The links, however, are not strong and the processes underlying the links still need speci cation. To take the next steps, one essential step is to take a closer look at the nature of the quality itself. Two structural approaches of complexity were examined: Newberger’s (1980) analysis of reasoning complexity, and Vallacher and Wegner’s (1985) analysis of action complexity (level of action identi cation). The former approach is part of the cognitive structural tradition already applied to the sociocognitive and sociomoral domains (Kohlberg, 1969; Selman, 1980), and also used in the domain of child rearing (Dekovic & Gerris, 1992). The latter approach has been developed within the social cognition tradition as a new theory of action (Vallacher & Wegner, 1985, 1987; Vallacher, Wegner, & Somoza, 1989; Wegner & Vallacher, 1986, 1987), and has served to identify a new
Correspondence should be sent to Marõ a Jose Rodrigo, Departamento de Psicologõ a Evolutiva, Facultad de Psicologõ a, Campus de Guajara, Universidad de La Laguna, Tenerife, Spain; e-mail:
[email protected]. This research was supported by Ministerio de Cultura y Educacio n de Espan˜a,
dimension of individual variation in personality, the level of personal agency (Vallacher & Wegner, 1989). This approach has not been used in any analysis of parental cognition. Blending developmental and social approaches to the analysis of complexity offers a way to ask whether complexity of thinking is a general quality of parental thinking that may imply both reasoning and action. It offers also a possible way to enrich our current understanding of links between parents’ representations and actions and to clarify the inconsistent results to date between ideas and actions, and between either of these and background characteristics such as SES. Motivations like these prompted the questions that are the focus of the present article: How is complexity to be de ned and measured? To what extent is the quality of mothers’ thinking, de ned as reasoning and action complexity, related to mothers’ childrearing behaviours? To what extent does complexity of either kind vary with SES? To what extent are connections between SES and mothers’ behaviour mediated by either form of complexity?
Complexity as a quality of maternal thinking The central issue is to study the relationship between reasoning complexity as a level of representation of global cognitions, and action complexity as a level of representation of concrete actions. There are conceptual grounds to consider that global and situated cognitions are two distinctive entities. Within the studies on the content of parental thinking there is a distinction between global schemas and praxis schemas (Sigel, 1985). It is claimed that global schemas (e.g., children’s cognitive development is a function of their environment) do not contain
CICYT PB96: 1037 to the rst author. We are particularly grateful to the Institute of Family Studies at the University of Nijmegen for nancing the rst author’s stay at this Institute, and particularly to Jan Gerris for his helpful comments on an earlier version of the manuscript.
INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (1), 50–59
clear-cut direct behavioural expressions, whereas praxis schemas (e.g., it is good to provide contingent reinforcements to child’s behaviour) provide clear guidelines to behaviour. There is a gap, however, between a noncontextual guideline provided by a praxis schema and its appliance to a unique situation (Rodrigo & Triana, 1996). This gap should be lled in with the construction of a situation model that represents the here-andnow of the situation, that is, ‘‘who says what to whom when and where’’ (De Vega, 1996; Zwaan & Radvansky, 1998). Any cognition of the parental action (e.g., action complexity) should be framed to this situation model. In our study, reasoning complexity can be considered analogous to global schemas that contained prototypical knowledge, whereas action complexity is a situated cognition linked to situation models. Although they are two different aspects of complexity they may be related to make beliefs more adapted to everyday situations. Thus, it may occur that complexity is a structural quality that may cut across global and situated cognitions. A detailed account of the two approaches follows before completing the rationale. Reasoning complexity refers to the way parents see themselves as related to the child (i.e., parenthood, parentchild relationship), and about the child as a person (Gerris, Dekovic, & Janssens, 1997; Newberger, 1980). It captures to what extent parental reasoning is engaged in a process of perspective-taking involving four qualitatively different levels of complexity: An egoistic self-orientation (level 1), a conventional norm-orientation (level 2), a subjective child-orientation (level 3), and a mutual-system orientation (level 4). At the rst level, parents focus on their own needs, interests, and feelings and expect their child to understand what their needs and wishes are as a parent. At the second level, parents also consider customs, traditions, rules, and norms prescribed by society. Parenthood is de ned as a set of roles and teaching tasks to help children to conform to desired behaviour. At the third level, parents adopt the subjective child perspective and reason from the position and needs of the child. Parenthood involves building up a good emotional relationship with the child and respecting his/her developmental needs. On the fourth level, parents are able to view both him/herself and the child as two unique subjects, each with their own views, needs and interests, taking part and being part in mutual exchanges and relationships. Action complexity refers to the way parents represent their situated actions (i.e., the meaning of child-rearing behaviours from their own point of view). According to the action identi cation theory developed by Vallacher and Wegner (1985) any action can be identi ed at many levels of complexity, ranging from low level identities (how the action is done) to high level identities (what its effects are). For instance, behind the action of a ‘‘verbal reprimand’’ may lie either a low level representation (e.g., ‘‘shouting at my child’’) that conveyed the details of the action, or a high level representation (e.g., ‘‘having a disciplined child’’) that involved a more general understanding of the action. Each level of representation of an action has its own de ning features. Low level identities permit a mapping of the concrete action, providing a close monitoring of the effective performing of this. The parent describing his/her action as ‘‘shouting at my child’’ is ready to accept any higher level identity made available by the context, because as a tendency people want to have a broad understanding of their actions (Vallacher & Wegner, 1987). Therefore, low level identities provide a nonstable basis that
51
promotes action change. In turn, high level identities permit a broad understanding of what parents think they are doing, providing enduring lines of reasoning and consequences for their actions. The action identi cation theory has served to describe a new dimension of individual variation, the level of personal agency (Vallacher & Wegner, 1989). The evidence came from a wide sample of subjects who completed the 25 items of the Behavioral Identi cation Form (BIF) covering various action domains. Results from BIF were compared with other scales of self-motivation, self-monitoring, and the locus of control. Results indicated that at one extreme is the low level agent, someone who operates on the world with the here-and-now components of the action. At the other extreme is the high level agent, someone who has an abstract sense of what his/her action means in terms of its long-term consequences and social meanings. Low level agents were found to be more impulsive, less self-motivated, less consistent in their behaviour over time, and more external in their locus of control than their high level counterparts. There is no evidence concerning the relationship between reasoning complexity and action complexity. As was said, it is likely that reasoning and action complexity might be related. For instance, mothers who recognise the child’s needs (level 3 of reasoning) would be more able to understand their childrearing behaviour in terms of a rich network of abstract meanings involving reasons and re ections on the long-term consequences for the child development (high level of action identi cation). In contrast, mothers who have an egoistic point of view about the parent-child relationships (level 1 of reasoning) would be more likely to have a narrow understanding of child-rearing behaviour in terms of their here-andnow components linked to the circumstances (low level of action identi cation). To test the relationship between reasoning and action complexity one of the rst tasks is to develop measures that will convert Vallacher and Wegner’s (1985) approach into a form suitable for use with parents given the total absence of studies. Two measures of the level of action identi cation were designed, one taken from hypothetical situations (self-reports of what one would do) and the other from observed motherchild interactions (direct observation of what one actually does). In both cases, mothers are judging context-speci c items instead of noncontextual generalisations, which appears to be a new trend to assess child-rearing attitudes (Holden & Edwards, 1989). Thus, the two action measures developed for this study tapped situated thinking linked to concrete episodes, whereas the Newberger’s measure based on a clinical interview tapped global thoughts about parenting made explicit during the interview.
Complexity and actions Our second research question was to examine the relationships between the quality of maternal thinking (reasoning complexity and action complexity) and mothers’ child-rearing behaviour, either in hypothetical or in observed mother-child situations. One important question to be answered is whether there might be differential links of both aspects of complexity to actions. First, given the situated nature of action complexity we expected closer links to maternal behaviour than reasoning complexity. Second, based on the evidence available we expected that the nature of these links was similar for both
52
RODRIGO, JANSSENS, CEBALLOS / REASONING AND ACTION COMPLEXITY
aspects of complexity (e.g., by yielding positive links with more positive child rearing and negative links with less positive child rearing). There is empirical evidence showing that reasoning complexity is a strong predictor of child-rearing behaviour. Dekovic, Gerris and Janssens (1991) found that perspectivistic parents who reason at a higher level about parent-child relationships are more inclined to display warm and supportive behaviour towards their child than egoistic parents who reason at a lower level. A higher level of reasoning was found to be related to authoritative control, with parents providing explanations and reasons for desired behaviour, appealing to the independence and autonomy of the child, and promoting self-re ections and empathy for victims. A lower level of reasoning was associated to restrictive control, with parents imposing strict limitations, ignoring the child’s initiatives, and using prohibitions. In the same vein, Gerris et al. (1997) found evidence that the higher the level was at which a parent reasoned about parent-child relations, the more authoritatively a parent reared his/her child. Evidence coming from indirect sources indicated that similar results would be expected for the level of action complexity. An interesting distinction has been proposed by Kuczynski (1984) between parental strategies for long-term compliance seeking to promote the child’s internalisation of social norms (e.g., responsivity and demandingness), and strategies for short-term compliance seeking to promote the child’s immediate obedience (e.g., restrictiveness and power assertion). It is likely that mothers with high level meanings of actions (high level agents) may select mainly strategies for long-term compliance because they are more in uenced by the long-term goals of their actions, and wish to establish a more internalised and enduring control over the child’s behaviour. Mothers with low level meanings of actions (low level agents) may select strategies for short-term compliance because they have a narrow understanding of their behaviour, they are more in uenced by situational constraints, and they only want to establish a short-term control over the child’s behaviour.
Complexity and SES Our third research question was to examine the cultural sources of the two structural approaches of parental cognitive complexity. In our study, socioeconomic status (SES) (level of education and level of occupation) was selected as a potential source of both reasoning and action complexity. It is interesting to know whether the two approaches showed differential links to SES. The results using the Newberger approach have to date been inconsistent. Newberger (1980) assumed that the level of reasoning is constructed from direct personal experience with children (years of parental experience). Newberger (1980) and Newberger and Cook (1983) reported a nonsigni cant relationship between level of reasoning and social class. According to Newberger, this nding rules out the possibility that different levels of reasoning represent different conceptions of parent-child relationships held by different social class groups. However, other results (Dekovic & Gerris, 1992; Gerris et al., 1997; Sameroff & Feil, 1985) showed that lower class parents (low SES parents) tend to reason at less advanced levels. In our study, we expected that reasoning complexity showed stronger links to SES than action complexity. It is likely that cultural sources might be more in uential to global cognitions than to situated cognitions, closely bound to the speci cs of concrete situations.
Complexity as a mediating variable The aim is to determine whether the two forms of complexity act as equivalent mediators between mothers’ SES and their child-rearing actions. Consistent SES-associated differences have been found in child-rearing behaviour both reported or observed. Higher SES homes are more frequently democratic and lower SES homes are more frequently authoritarian (e.g., Hess, 1970; Kelley, Sanchez-Hucles, & Walker, 1993; Skinner, 1985). In interacting with their children, lower SES mothers are more controlling, restrictive, and disapproving than higher SES mothers (e.g., Farran & Haskings, 1980; Hart & Risley, 1992; Hoff-Ginsberg & Tardif, 1995). Moreover, Gerris et al., (1997) found that the relation between SES and child-rearing style was totally mediated by parental reasoning complexity both for mothers and fathers. The higher the level was at which a parent reasoned about parent-child relations, the more childoriented and authoritatively a parent reared his/her child. In this study, we want to replicate this result and see what action complexity may add to this relation. We expected that reasoning and action complexity showed a slightly differential contribution to this relation, that is, reasoning being more closely related to the cultural sources and action complexity being more strongly related to behaviour.
Method Sample The sample consisted of 60 mothers (mean age = 36.5; range 25–52), each with a child (mean age = 8.5; range 7–10 years old) attending either the second (11 girls and 8 boys), the third (12 girls and 9 boys), or the fourth grade (12 girls and 8 boys) of six elementary schools in the island of Tenerife, Spain. About 70% of the assistant mothers to the school meetings agree to participate voluntarily. Mothers come from twoparent families of one to three children. The sample composition was balanced according to the degree of urbanisation (30 mothers from rural and 30 from urban areas), and the level of education (20 mothers with a low educational level, 20 with a medium level, and 20 with a high level of education). Both variables have proved to be very useful as predictors of differential child-rearing attitudes in previous studies (e.g., Hoff-Ginsberg & Tardif, 1995).
Procedure Each mother was given the situational questionnaire of childrearing practices at home during the rst visit and was asked to ll in this questionnaire in the presence of the experimenter. In addition, data were collected for the level of action identi cation ( rst measure) from the hypothetical situations of the questionnaire. During the second visit, the mother-child interaction while performing a task was video recorded. Then, the video was presented to the mother to collect data for the level of action identi cation (second measure). Finally, data from the Newberger interview was collected in the third visit.
Measures Reported child-rearing behaviour. Mothers were asked to ll in the Situational Questionnaire of Child-Rearing Practices and Goals (Ceballos & Rodrigo, 1992). This questionnaire consists
INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (1), 50–59
of 15 hypothetical situations each describing a real-life episode involving a parent and a child. Five situations involve externalising behaviour of a child (e.g., ‘‘One evening your son/daughter is alone at home. He/she is bored and, though he/ she is not allowed to, he/she plays roughly with his/her sister’s radio cassette which gets damaged’’). Five situations involve internalising problems (e.g., ‘‘You go to see your son’s/ daughter’s teacher after class. He tells you that your son/ daughter is very insecure and that he/she is not talking to his/ her classmates’’). Five situations involve a child’s violation of social conventions (e.g., ‘‘One evening you and your partner go to visit a couple of friends who do not have children. After a while your son/daughter gets bored, stands up and says that he/ she wants to go home’’). In each situation, mothers were presented with eight childrearing practices and were asked to indicate on a 5-point scale (range 0–4; never, seldom, sometimes, often, always) how often they would use each child-rearing practice with their own child. The practices were selected according to 16 categories of child rearing: (1) Overprotection (mother is directive in protecting her child against bad in uences from outside); (2) Limiting (places limits to the child with respect to the adults’ world); (3) Punishment (uses reprimands, mild punishment, time-out; (4) Strictness (imposes restrictions and rules on children’s life); (5) Responsibility (makes the child responsible for his/her own actions); (6) Sensitivity (responsive to the children’s problems and opinions); (7) Communication (talks to the child about the problems); (8) Support (promotes initiatives and the child’s decision making); (9) Perspective-taking (uses negotiation skills and other-centred perspectives); (10) Reasoning (uses causal thinking and explanations about the child’s action); (11) Ef cacy (pushes the child to overcome dif culties); (12) Nurturance (promotes correct nutrition and health habits); (13) Openness (stimulates the child to social relationships); (14) Overindulgence (disregards undesired behaviour); (15) Avoidance (avoids problems); (16) Determinism (acceptance of a child’s behaviour). Over the 15 situations, mothers had to rate the frequency of use of 120 instances of childrearing practices (6 to 8 instances for each category) stated in speci c terms with regard to each situation. A mean score was computed for each category of practices. A factor analysis on these 16 mean scores revealed that three factors explained 62.6% of the variance of these measures of child rearing (varimax rotation was used). The rst factor was loaded highly by Sensitivity (.63), Communication (.73), Support (.75), Perspective-Taking (.75), Reasoning (.83), and Ef cacy (.59). These practices may correspond to the part of the Authoritative style that emphasises Responsiveness (Baumrind, 1971; Haaf, Janssens, & Gerris, 1994). The second factor was loaded highly by Overprotection (.68), Limiting (.82), Punishment (.72), Strictness (.86), and Responsibility (.58). These practices may refer to the part of the Authoritative style involving Demandingness and may also involve Restrictive Control, but not power assertion and coercive methods (Lytton, 1980; Maccoby & Martin, 1983). On the third factor, high loadings were found of Nurturance (.61), Openness (.58), Overindulgence (.83), Avoidance (.66), and Determinism (.78). These measures may refer to Permissiveness (Baumrind, 1971; Maccoby & Martin, 1983). Three factor scores were computed for each mother. A high score on the rst factor, Responsiveness, means that a mother gives explanations of rules, offering reasons for desired behaviour, pointing out the harmful consequences of the
53
child’s behaviour for others, being sensitive to the child’s problems and offering support, con dence, and communication. A high score on the second factor, Control and Demandingness, means that a mother exercises control by enforcing rules, protects the child from bad in uences, pushes the child to reach high standards of maturity, sets well-de ned limits on the child’s behaviour and uses prohibitions very frequently. A high score on the third factor, Permissiveness, means that a mother avoids con ict with the child, accepts his/ her behaviour, and is indulgent with the child’s misdeeds. Observed mother-child behaviour. To assess child-rearing behaviour, we used video recordings of a shopping task in which mother and child were presented with a map of an imaginary town (Radziszewska & Rogoff, 1988). They had to complete a set of errands doing shopping in this town to prepare an excursion. The dyad was prompted to work jointly to choose the shortest route but only one route was allowed. We derived three measures of maternal behaviour from Wagner and Phillips (1992): Warmth (mother was affectionate towards the child, showed pleasure, was enthusiastic about her child’s performance, laughed a lot, used humour), Stress (mother was impatient, was worried whether her child would nish the task, felt nervous), and Directness (gave her child answers or leading cues, offered unsolicited feedback seeking immediate effect, reacted quickly to what she perceived as an error, was preoccupied with correct task completion). Two observers rated on a 9-point scale (ranging from ‘‘never observed’’ on the video to ‘‘always observed’’) each of these three measures. The rating procedure was done twice: By half of the task completion ( rst), and at the end of the task (second). Inter-rater reliabilities and percentage of agreements were high for all categories and time of measures: Warmth ( rst: .82/100%, second: .82/96.7%); Stress ( rst: .82/96.7%, second: .75/98.3%); and Directness ( rst: .85/100%, second: .73/98%). The rating scores were averaged over raters and measurements. Level of action complexity. Action complexity was measured by means of a level of action identi cation. Two measures were developed. The rst measure was based on the mother’s choices of disciplinary practices in the 15 hypothetical childrearing situations depicted on the questionnaire. Mothers were asked what they would do and why in each situation, in order to have a personal account of their actions. The second measure was based on the mothers’ observed behaviour in the shopping task with their child. Mothers were shown in a postvideo session the recordings of their interaction and were asked at six different points what and why they were doing something. We preferred to register the spontaneous labelling of the actions given by the mothers, instead of proposing labels to be chosen (Vallacher & Wegner, 1989), in order to avoid response bias. The rst action measure was based on 30 answers for what and why questions collected across the 15 hypothetical situations of the questionnaire. Two raters judged independently a total of 1800 responses on a 2-point scale: 0 = low level of action identi cation: 1 = high level of action identi cation. A low level answer contained a concrete description of mother’s behaviour with action details (e.g., ‘‘I tell him to stop that behaviour’’), or indicated here-and-now reasons and short-term goals (e.g., ‘‘I need some rest now’’). A high level answer described a general understanding and
54
RODRIGO, JANSSENS, CEBALLOS / REASONING AND ACTION COMPLEXITY
abstract meaning of an action without the details of the action (e.g., ‘‘I like to have good communication with my daughter’’), or if a mother referred to long-term goals (e.g., ‘‘I want to have a child that acts autonomously’’). Inter-rate reliability was high: .98 for the what- and .97 for the why-question. Mean scores over the 15 answers on the what-question and over the 15 answers on the why-question were calculated. The correlation between these two scores was high (r = .75). Therefore, we calculated one mean score over all 30 answers for subsequent analyses. The second action measure was based on 12 answers for what and why questions collected across the six moments at which the video was stopped. Two raters judged independently a total of 720 responses on a 2-point scale. We followed the same criteria used previously to code the answers as 0 = low level (e.g., ‘‘I am drawing a line to reach the bank’’), and 1 = high-level (e.g., ‘‘I want to help my child to understand his error’’). Inter-rater reliability was high: .98 for the what-, and .97 for the why-question. Mean scores over the 6 answers on the what-question and over the 6 answers on the why-question were calculated. The correlation between these two scores was high (r = .76). Therefore, we calculated one mean score over all 12 answers for subsequent analyses. The what- and why- questioning in both measures was intended to tap low level and high level identities, respectively. Thus, it might be that these questions totally determine the level of identi cation that a mother is going to adopt. However, with the questionnaire measure, 60% of the total answers to the what-question were coded as low level, and 65% of the total answers to the why-question were coded as high level. With the video measure, 67% of the total answers to the whatquestion were coded as low level, and 55% were coded as high level under the why-question. Consequently, the kind of question itself, though giving the opportunity for low and high identities to emerge, did not totally determine the level of Action identi cation adopted by a mother. Level of reasoning complexity. Reasoning complexity with respect to children, mother-child relationship, and motherhood was assessed for each mother individually in a semistructured re ective interview which includes a series of standard questions and follow-up probes about eight issues (Dekovic & Gerris, 1992; Newberger, 1980; Newberger & Cook, 1983): 1. Factors that in uence child’s behaviour and development: How do children learn what is right and what is wrong? What is the most important in uence on the way the child turns out as an adult? Why? 2. Understanding the subjective experience of the child: How do you think your children see you as a parent? Do you think that privacy is important for a child? Why? 3. Qualities and characteristics which de ne personality: How would you describe your child? What kind of person is he or she? How would you describe an ideal child? 4. Establishing communication and trust: Why is communication between parents and children important? How would you describe a good relationship between parents and a child? 5. Resolving con icts: What causes con icts between parents and children? What do you think is the best way to solve con icts? Why?
6. Maintaining discipline and authority: On what do you rely to get your child to obey you? What are your reasons for using this approach? 7. Meeting a child’s needs: What do you feel children need the most from their parents? Why is that important for a child? 8. Learning and evaluating parenting: How do people learn to be parents? How does someone know if she/he is a good parent? Each interview lasted 30–45 minutes and was recorded and transcribed. Each mother was given a score (range Level 1 to Level 4) for each interview question. The inter-rater reliability between two independent raters was .91. The separate scores for the eight issues were averaged into a total average score for subsequent analyses. Level of education and occupation. Mother’s level of Education was scored on a 3-point scale; low (primary school), medium (secondary or high school), and high (university or higher vocational education). Mother’s level of Occupation was also scored on a 3-point scale: low (e.g., house assistant, cleaner), medium (e.g., seller, secretary, social worker), and high quali ed work (e.g., manager, engineer).
Data analysis Correlational analysis was performed on level of education, level of occupation, level of reasoning complexity (one measure), level of action complexity (two measures), and child-rearing behaviour (three reported measures: responsiveness, control and demandingness, and permissiveness; three observed measures: warmth, pressure, and directness). In addition, structural equation models (Lisrel VI: Jo¨reskog & So¨rbom, 1984) were used to test the hypothesised mediating role of maternal reasoning and action complexity between SES and child-rearing behaviour with reported and observed data.
Results Preliminary statistics Mean scores of all variables included in this study are presented in Table 1. For the factor scores based on the 15 hypothetical situations we do not present mean factor scores because of the regression method used to compute these scores, the mean of these scores was zero and the standard deviation one. Therefore, we present mean scores over the measures that loaded more than .60 on each child-rearing factor derived from the answers on the hypothetical situations. Mean scores for the measures related to the factor of Responsiveness indicated that mothers ‘‘often’’ used these methods. Mean scores for measures related to the factor Control and Demandingness indicated that mothers ‘‘sometimes’’ used these practices. Mean scores for measures related to Permissiveness indicated that mothers ‘‘seldom’’ reported using these methods. t-Tests for paired observations revealed that these differences were signi cant: Responsiveness versus Control and Demandingness (t = 8.18, df = 59; p < .01); Control versus Permissiveness (t = 20.41, df = 59; p < .01); Responsiveness versus Permissiveness (t = 8.65, df = 59; p < .01). Mothers were rated as rather warm, rather stressful, and
55
INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (1), 50–59
Table 1 Mean scores of SES, maternal cognitions, and child-rearing behaviour Mean
SD
SES Level of education Level of occupation
2.00 1.73
0.82 0.75
Cognitions Reasoning complexity Action complexity (hypothetical situations) Action complexity (observed situations)
2.29 0.52 0.44
0.29 0.25 0.30
Behaviour (hypothetical situations) Responsiveness Control and demandingness Permissiveness
2.70 1.93 1.29
0.50 0.59 0.42
Behaviour (observed situations) Warmth Stress Directness
5.22 5.25 6.50
1.45 1.34 1.40
were inclined to push their child to do what they wanted during the shopping task. All mean ratings of these measures were about 5–7 on a 9-point scale. t-Tests for paired observations revealed that these differences were signi cant for Directness versus Warmth (t = 4.39, df = 59; p < .01), and for Directness versus Stress (t = 4.22, df = 59; p < .01), but not for Warmth versus Stress. The mean scores of Reasoning complexity and Action complexity indicated that mothers tend to reason at the conventional level and showed a moderate level of action identi cation. t-Tests for paired observations indicated that the mean level of action identi cation based on the hypothetical situations was higher than that based on the observation task (t = 1.86, df = 59; p < .05).
than for reported (.26; p < .05) maternal actions. The two measures of action identi cation were related, but the strength of this correlation (.28; p < .05) was lower than might have been expected on the basis of their conceptual similarity. Second, we found relations between maternal cognitions (Reasoning and Action complexity), and measures of maternal child-rearing behaviour based on hypothetical and observed mother-child situations. The higher the level of reasoning, the less Control and Demandingness, and Permissiveness were reported. It also appeared that the higher the level of reasoning, the more mothers displayed Warmth, and the less Pressure and less Directness the mothers showed on the video task. Action complexity based on hypothetical and observed situations were also related to measures of child-rearing practices. The higher the level of action identi cation based on reported actions, the more Responsiveness and the less Control and Demandingness were reported, and the more displayed Warmth during the video task. The higher the level of identi cation based on observed actions, the more Responsiveness and the less Control and Demandingness were reported, and the less Pressure and less Directness they showed during the video task. It appeared also that measures of maternal child-rearing behaviour derived from the shopping task and those derived from the hypothetical situations were fairly independent. Only Control and Demandingness were positively related to Pressure. Third, mothers’ level of education and occupation were related to their cognitions and behaviour. The higher these levels were, the higher mothers’ Reasoning complexity was, and the higher her Action complexity (only for observed situations). The level of education was positively related to Responsiveness, and negatively to Control and Demandingness and to Permissiveness. Mother’s Directness during the shopping task was negatively related to the mother’s level of education. Mother’s Pressure and Directness were also negatively related to her level of occupation. The higher that level was the less pressure and less directness mothers showed at the shopping task.
Correlational analysis Correlations between all variables included in this study are presented in Table 2. First, relations between maternal cognitions (Reasoning and Action complexity) were found. The higher the level of maternal reasoning, the higher the level of action identi cation was. This relation was stronger for observed (.38; p < .01)
Test of Lisrel models In our study, Reasoning complexity and Action complexity were expected to mediate the relationship between SES and child rearing. Two Lisrel models were tested based on the matrix of correlations between the measures derived from the
Table 2 Correlations between SES, maternal cognitions, and maternal behaviour
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
Level of education Level of occupation Reasoning complexity Action complexity (observed situations) Action complexity (hypothetical situations) Responsiveness Control and demandingness Permissiveness Warmth Pressure Directness
* p < .05; ** p < .01.
1
2
3
4
5
.49** .47** .42** .06 .28* ± .44** ± .23* ± .09 ± .01 ± .26*
.48** .29* .02 ±.02 ±.32** ±.28* .12 ±.23* ±.43**
.38** .26* .15 ± .39** ± .30* .26* ± .34* ± .43**
.28* .27* ± .23* .11 .09 ± .32* ± .28*
.23* ±.20* ±.16 .19* ±.16 ±.08
6
7
8
9
10
.04 ± .15 ± .16
0 ± .08 .28* .14
± .15 ± .04 .05
±.39** ±.26*
.66**
0 0
56
RODRIGO, JANSSENS, CEBALLOS / REASONING AND ACTION COMPLEXITY
hypothetical situations, and from the measures of the observed situations (both sets of measures were fairly independent of each other). Maximum likelihood estimation was used to estimate the strength of the in uences in each model tested. Because the mother’s level of education and occupation were highly related (see Table 2), we constructed for both models a latent variable, SES, with expected loadings of the two manifest variables, level of education, and level of occupation. Reasoning complexity was also included in both models. In the model with reported data, besides SES and Reasoning, we introduced one measure of Action complexity based on reported actions, and three measures of reported child-rearing practices: Responsiveness, Control and Demandingness, and Permissiveness. The test showed that the supposed Lisrel model did not t the data (chi-square = 24.65, df = 11, p = .01; RMSR = .088; GFI = .91; AGFI = .77). Reasoning complexity did not have an effect on Responsiveness, and nonsigni cant effects were observed from Action complexity on Control and Demandingness and on Permissiveness. Moreover, it appeared that SES did not have an effect on Action complexity. A modi ed model was tested, in which nonsigni cant effects were xed at zero, and in which the relation between the unexplained variances of Reasoning complexity and Action complexity was estimated. This model tted the data (chi-square = 22.68, df = 14, p = .066; RMSR = .093; GFI = .92; AGFI = .84). The strength of the relations between variables is represented by the signi cant standard regression coef cients along the arrows in Figure 1. The in uence of SES on child rearing was only mediated by Reasoning complexity. The higher a mother’s SES was, the higher her level of reasoning and the less Control and Demandingness and Permissiveness were reported. Action complexity was linked positively to Reasoning complexity, and
had a positive effect on the use of Responsiveness; the higher a mother’s level of action identi cation was, the more she was inclined to use these methods. In the Lisrel model with observed data, besides SES and Reasoning, we introduced one measure of Action complexity based on observed actions, and three measures of maternal behaviour during the shopping task: Warmth, Pressure, and Directness. Because of the high correlation between Pressure and Directness (.66) we constructed one latent variable, Restrictive control, with loadings of the manifest variables Pressure and Directness. On the second latent variable, Warmth, only the manifest variable Warmth was expected to load. The Lisrel model did not t the data (chi-square = 24.22, df = 11, p = .012; RMSR = .096; GFI = .89; AGFI = .73). The effect of Action complexity on Warmth was not signi cant. A test, in which this nonsigni cant effect was xed at zero, and in which the relation between the unexplained variances of Warmth and Restrictive control was estimated showed a better t (chi-square = 19.18, df = 11, p = .058; RMSR = .079; GFI = .92; AGFI = .79). The strength of the relations between variables is represented by the signi cant standard regression coef cients along the arrows in Figure 2. Mothers’ SES had a positive in uence on her Reasoning complexity and her Action complexity; the higher a mother’s status was, the higher her level of reasoning and her level of action identi cation. The strong effects of SES on both variables suggested that Reasoning and Action complexity did not causally in uence each other. Both complexity measures had a negative in uence on mother’s Restrictive control. The higher these levels were, the less pressure and directness mother showed. Reasoning complexity had also a positive in uence on maternal Warmth; the higher that level, the more mothers exhibited warmth. Warmth and Restrictive
Figure 1. Lisrel model with self-report data.
INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (1), 50–59
Figure 2.
57
Lisrel model with observation data.
control showed negative links; the more mothers displayed warmth the less they behaved restrictively.
Discussion For the rst research question, two aspects of complexity were examined, reasoning and action complexity as de ned by Newberger’s and by Vallacher and Wegner’s approaches, respectively. On conceptual grounds both forms of complexity had a different nature: Reasoning complexity is a global schemata containing prototypical information, whereas action complexity entails a situated cognition linked to the elaboration of situation models (e.g., de Vega, 1996). Results indicated that some relation exists between both aspects of complexity. Mothers reasoning at higher levels tended to have an abstract meaning for their actions and represented them in terms of the long-term goals of socialisation (high level of action identi cation). Mothers who reasoned at the lower level tended to restrict themselves to action details and represented their actions in terms of the here-and-now constituents (low level of action identi cation). Therefore, it seems that complexity is a structural quality of maternal thinking that cuts across global schemas and situated cognitions. The foregoing conclusion suggested that there is a relative interchange between the belief system and the modelling of real situations to provide parents with an adaptive tuning between cognitions and actions. A result in line with this argument is that the statistical relationship between reasoning and action complexity, though signi cant in both cases, was stronger when action complexity was measured in observed rather than hypothetical situations. In fact, we found a moderate correlation between the two action measures, less than was expected given their conceptual and methodological similarities. Both measures used the what- and why-questioning, and were based on the mothers’ spontaneous statements
about their child-rearing actions in concrete episodes. There are, however, some crucial differences: Observed situations are real, factual, recent, and involved the actual mother as a protagonist. By contrast, hypothetical situations are ctitious, nonfactual, anticipatory, and involved a mental simulation of the mother’s role. Moreover, with observed situations we were tapping mothers’ representations of unique actions (what mothers actually did in that situation), whereas with hypothetical situations we were tapping mothers’ representations of averaged actions (what mothers usually do in that kind of situation). The former representations are probably more situated in nature than the latter. These results point to the convenience of moving the topic beyond the contextual or noncontextual character of our measures (e.g., Holden & Edwards, 1989) to the analysis of their situated nature. Our second research question was to examine whether there might be differential links of both aspects of complexity to actions. First, we expected that action complexity as a situated cognition showed closer links to maternal behaviour than reasoning complexity. The results indicated that this was not the case. Both Reasoning complexity and Action complexity based on actual actions were related to a broader set of measures of maternal behaviour than Action complexity based on reported actions. It seems that the global reasoning measure showed structural similarities with the more situated representation of actual actions in their relation to child-rearing actions. Our second expectation was that the nature of the links with action was similar for both aspects of complexity. Results indicated that there were some slight differences in their links to actions but not in the direction of effects. The higher the maternal Reasoning complexity was the less Control and Demandingness, and Permissiveness in hypothetical situations were reported, and the less Pressure and Directness the mothers showed, and the more mothers displayed Warmth in observed situations. These results are clearly consistent with
58
RODRIGO, JANSSENS, CEBALLOS / REASONING AND ACTION COMPLEXITY
those found by Dekovic et al. (1991), and Gerris et al. (1997). Mothers who reason at the higher levels did not show restrictive or permissive parenting. In our study, however, we did not nd a positive relationship between Reasoning complexity and Responsiveness on the hypothetical situations. We did nd a positive relation with maternal warmth in the shopping task. Mothers reasoning at higher levels were affectionate towards the child, showed pleasure, were enthusiastic about their child’s performance, laughed a lot, and used humour. Concerning the action complexity, the higher the level of identi cation was the more mothers were Responsive, less Controlling, and showed Warmth, less Pressure, and less Directness during the shopping task. Mothers who had a broad understanding of their child-rearing behaviour (high level agents in both hypothetical and observed situations) were those who claimed to be sensitive to the child’s needs, used reasons and explanations, promoted empathy with victims, and were less inclined to be controlling. These results are quite consistent with those found by Kuczynski (1984) on the strategies of long-term compliance (e.g., Responsivity and Demandingness) or short-term compliance (e.g., Restrictivity and Power assertion). That is, it seems that mothers who were high level agents wanted to promote the child’s internalisation of social norms and values by using strategies of long-term compliance, whereas those who were low level agents wanted to promote the child’s immediate obedience by using strategies of short-term compliance. The only discrepancy was that Demandingness was considered a strategy for long-term compliance by Kuczynski, and our results showed that it was related to lower levels of action complexity. A functional distinction should be made between Responsiveness and Demandingness in terms of the kind of representation that lies behind the actions. To be responsive to the child’s needs goes along with a rich network of meanings involving reasons and re ection on the long-term consequences of the action. To be demanding and expecting mature behaviour from the child goes along with a narrow understanding of the parental behaviour, and with the wish to establish a short-term control over the child’s behaviour. Our third research question was to examine the cultural sources of reasoning complexity and action complexity. We expected that cultural sources might be more in uential to global cognitions than to situated cognitions, closely bound to the speci cs of concrete situations. On the contrary, results indicated that both global and situated measures of complexity were strongly related to SES. Mothers with higher SES were mothers who represented their relation to the child at more perspectivistic levels (Dekovic & Gerris, 1992; Gerris et al., 1997; Sameroff & Feil, 1985). Again, Newberger’s (1980) and Newberger and Cook’s (1983) expectation that social class was not related to the level of reasoning was not con rmed. As for action complexity, mothers with higher SES had a broader understanding of their child-rearing actions and de ned their actions in terms of long-term consequences in the shopping task. Again, it is the more situated measure of action complexity that showed a similar trend of results with the global measure of reasoning. In addition, consistent SESassociated differences were also found in the child-rearing behaviour either reported or observed. Higher SES mothers claimed to be more responsive and less restrictive and permissive (Hess, 1970; Kelley et al., 1993; Skinner, 1985). In interacting with their children, mothers with higher
occupational levels showed less pressure and directness toward the child (Farran & Haskings, 1980; Hart & Risley, 1992). Consequently, cultural sources seem to shape the way parents organise their thinking, their conceptualisation of actual parental actions, and the content of these actions. Our fourth research question was to determine whether the two forms of complexity act as equivalent mediators between mothers’ SES and their child-rearing actions. We expected that reasoning and action complexity showed a slightly differential contribution to this relation, reasoning being more closely related to the cultural sources and action complexity more strongly related to behaviour. Results indicated that reasoning complexity was the stronger mediator in both Lisrel models. With reported data, it was clear that the in uence of SES on less positive child rearing (Control and Demandingness, and Permissiveness) was only mediated by Reasoning complexity. Gerris et al. (1977) also found this sort of mediation in a sample of mothers and fathers. Instead, level of action complexity, though linked to reasoning, showed an independent path to positive maternal child-rearing behaviour (Responsiveness). Thus, Reasoning and Action complexity based on reported actions showed a nonequivalent mediating role. When the action measure entailed the representation of actual actions, the in uence of SES on Restrictive control (somehow similar to Control and Demandingness) was equally mediated by both Reasoning and Action complexity. In this case, both constructs may be considered to be culturally determined to a larger extent and contributed jointly to predict negative child rearing. To conclude, in this study we have explored in more detail the complexity of maternal thinking by bringing Vallacher and Wegner’s analysis of the Action identi cation to the parenting eld. We have shown that complexity is a structural quality of maternal thinking that cuts across two distinctive entities: global schemas (reasoning complexity) and situated cognitions linked to situation models (action complexity). The increasing complexity of reasoning enables the mother to set a broader context of meaning for her actual actions in terms of their longterm consequences on child development. Two measures of action complexity drawn from hypothetical and observed situations have been proposed. From our results it is clear that a distinction should be made between a representation of unique actions linked to a situation models and a representation of averaged actions. Only the former action representations shared many structural features with Reasoning complexity as a global cognition: Both were related to SES, and were as good predictors of similar negative maternal behaviours. We hope that the introduction of these situated measures in our studies would increase our current understanding of the links between parents’ representations and actions. Manuscript received 14 April 1998 Revised manuscript received 24 June 1999
References Baumrind, D. (1971). Current patterns of parental authority. Developmental Psychology Monographs, 4 (No. 1, Pt. 2). Ceballos, E.M., & Rodrigo, M.J. (1992, September). A situational questionnaire of child-rearing practices and goals. Paper presented at the Vth European Conference on Developmental Psychology, Seville, Spain.
INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (1), 50–59 Dekovic, M., & Gerris, J.R.M. (1992). Parental reasoning complexity, social class, and child-rearing behaviors, Journal of Marriage and the Family, 5, 675– 685. Dekovic, M., Gerris, J.R.M., & Janssens, J.M.A.M. (1991). Parental cognitions, parental behavior, and the child’s understanding of the parent-child relationship. Merrill-Palmer Quarterly, 37, 523–541. De Vega, M. (1996). Spatial and interpersonal models in the comprehension of narratives. In M. Carreiras, J.E. Garcõ a-Albea, & N. Sebastõ an-Galle s (Eds.), Language Processing in Spanish (pp. 275–304). Mahwah, NJ: Erlbaum. Farran, D.C., & Haskings, R. (1980). Reciprocal in uence in the social interactions of mothers and three-year-old children from different socioeconomic backgrounds. Child Development, 51, 780–791. Gerris, J.R.M., Dekovic, M., & Janssens, J.M.A.M. (1997). The relationship between social class and childrearing behaviors: Parents’ perspective taking and value orientations. Journal of Marriage and the Family, 59, 834–847. Goodnow, J.J., & Collins, W.A. (1990). Development according to parents: The nature, sources, and consequences of parents’ ideas. Hillsdale, NJ: Erlbaum. Hart, B., & Risley, T.R. (1992). American parenting of language-learning children: Persisting differences in family-child interactions observed in natural home environments. Developmental Psychology, 28, 1096–1106. Haaf, P.G.J. ten, Janssens, J.M.A.M., & Gerris, J.R.M. (1994). Child-rearing measures: convergent and discriminant validity. European Journal of Psychological Assessment, 10, 111–128. Hess, R.D. (1970). Social class and ethnic in uences upon socialization. In P.H. Mussen (Ed.), Carmichael’s manual of child psychology (3rd ed., Vol. 2, pp. 457–557). New York: Wiley. Hoff-Ginsberg, E., & Tardif, T. (1995). Socioeconomic status and parenting. In M.H. Bornstein (Ed.), Handbook of parenting: Vol. 2. Biology and ecology of parenting (pp. 161–188). Mahwah, NJ: Erlbaum. Holden, G.H., & Edwards, L.A. (1989). Parental attitudes toward child-rearing: instrumens, issues, and implications. Psychological Bulletin, 106, 29–58. Jo¨reskog, K.G., & So¨rbom, D. (1984). LISREL VI: Analysis of linear structural relationships by the method of maximum likelihood, instrumental variables, and least squares methods. Mooresville, IN: Scienti c Software. Kelley, M.L., Sanchez-Hucles, J., & Walker, R. (1993). Correlates of disciplinary practices in working to middle-class African-American mothers. Merrill-Palmer Quarterly, 39, 252–264. Kohlberg, L. (1969). Stage and sequence: The cognitive-developmental approach to socialization. In D.A. Goslin (Ed.), Handbook of socialization theory and research (pp. 347–480). Chicago, IL: Rand McNally. Kuczynski, L. (1984). Socialization goals and mother-child interaction: Strategies for long-term and short-term compliance. Developmental Psychology, 20, 1061–1073. Lytton, H. (1980). Parent-child interaction: The socialization process observed in twin and singleton families. New York: Plenum. Maccoby, E.E., & Martin, J.A. (1983). Socialization in the context of the family: Parent-child interaction. In P.H. Mussen (Series Ed.) & E.M. Hetherington
59
(Vol. Ed.), Handbook of child psychology: Vol 4. Socialization, personality, and social development (pp. 1–102). New York: Wiley. Newberger, C.M. (1980). The cognitive structure of parenthood: Designing a descriptive measure. In R.L. Selman & R. Yando (Eds.), Clinical-developmental psychology. New directions for child development (No. 7, pp. 45–67). San Francisco, CA: Jossey-Bass. Newberger, C.M., & Cook, S.L. (1983). Parental awareness and child abuse: A cognitive developmental analysis of urban and rural samples. American Journal of Orthopsychiatry, 53, 512–524. Radziszewska, B., & Rogoff, B. (1988). In uence of adult and peer collaborators on children’s planning skills. Developmental Psychology, 24, 840–848. Rodrigo, M.J., & Triana, B. (1996). Parental beliefs about child development and parental inferences about actions during child-rearing episodes. European Journal of Psychology of Education, 11, 55–78. Sameroff, A.J., & Feil, L.A. (1985). Parental concepts of development. In I.E. Sigel (Ed.), Parental belief systems: The psychological consequences for children (pp. 83–105). Hillsdale, NJ: Erlbaum. Selman, R.L. (1980). The growth of interpersonal understanding. New York: Academic Press. Sigel, I.E. (1985). Parental belief systems: The psychological consequences for children. Hillsdale, NJ: Erlbaum. Skinner, E.A. (1985). Determinants of mother sensitive and contingentresponsive behavior: The role of child-rearing beliefs and socioeconomic status. In I.E. Sigel (Ed.), Parental belief systems: The psychological consequences for children (pp. 51–82). Hillsdale, NJ: Erlbaum. Vallacher, R.R., & Wegner, D.M. (1985). A theory of action identi cation. Hillsdale, NJ: Erlbaum. Vallacher, R.R., & Wegner, D.M. (1987). What do people think they’re doing? Action identi cation and human behavior. Psychological Review, 94, 2–15. Vallacher, R.R., & Wegner, D.M. (1989). Level of personal agency: Individual variation in action identi cation. Journal of Personality and Social Psychology, 56, 660–671. Vallacher, R.R., Wegner, D.M., & Somoza, M.P. (1989). That’s easy for you to say: Action identi cation and speech uency. Journal of Personality and Social Psychology, 56, 199–208. Wagner, B.M., & Phillips, D.A. (1992). Beyond beliefs: Parent and child’s behaviors and children’s perceived academic competence. Child Development, 63, 1380–1391. Wegner, D.M., & Vallacher, R.R. (1986). Action identi cation. In R.M. Sorrentino & E.T. Higgins (Eds.), Handbook of motivation and cognition (pp. 550–582). New York: Guilford Press. Wegner, D.M., & Vallacher, R.R. (1987). The trouble with action. Social Cognition, 5, 179–190. Zwaan, R.A., & Radvansky, G.A. (1998). Situation models in language and memory. Psychological Bulletin, 123, 162–185.
Copyright of International Journal of Behavioral Development is the property of Sage Publications Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.