Children's Development of Social Competence Across Family Types/Framework
2.1 Socialization theories 
The hypothesis that children's development of social competence might vary depending on whether a child has one or two parents in the present in the home, and whether those parents are of the opposite sex or of the same sex, has been characterized as rooted in Freudian psychoanalytic theory (Hay & Nash, 2002; Maccoby, in press). Although children were presumed to identify with their same-sex parent, Freud emphasized the importance of children's relationships with both biological parents (e. g., Freud, 1938). Thus, children who were not raised in homes with their biological mothers and fathers were thought of as being at risk for maladjustment and, in particular, of not following typical gender role development. In many ways Freud's theory must be considered within the historical context of the era in which he lived, when it was rarer for children to be raised by non-biologically-related parents. His lack of attention to the socialization of children by non-biological parents and legal guardians is one of the reasons Freud's ideas have been criticized by developmental scientists. There is little empirical support for psychoanalytic theory, and it is generally not considered influential for current theories and research in socialization.
Socialization is a central element of many current theories of child development, including attachment theory (Ainsworth, Blehar, Waters, & Wall, 1978; Bowlby, 1969), social learning theory (Bandura, 1977; Sears, Maccoby, & Levin, 1957), and family systems theory (Cox & Paley, 1997). Attachment theory places primary importance on the sensitivity of parents to the signals and needs of their infants. Parents who respond contingently and appropriately to their infants' needs, and provide a supporting, warm and safe context for early development, will facilitate their children's optimal emotional and social functioning. Their children will internalize mental models of "secure" attachment relationships, in which they believe that they have a safe base with protective caregivers from which they can explore the world, that their relationship partners are trustworthy and good people, and that they themselves are worthy of receiving love and positive attention from others. Conversely, because of their experiences with less sensitive, supportive and warm parents, children with mental models of "insecure" attachment relationships lack these adaptive belief systems and are at risk of experiencing social difficulties.
Social learning theory, and its extension, social cognitive theory (Grusee, 1992), was built from classic behaviourism to explain how children learn in the social world. Behaviourism was focused on direct rewards and punishments of behaviours. Behaviours that were rewarded would be reinforced and would be more likely to be repeated. Behaviours that were punished would be less likely to be repeated. Sears proposed that children did not have to engage in these behaviours directly in order to learn. Rather, they could watch another person engage in the behaviour and be rewarded for it by attaining some valued goals, and thereby learn that the behaviour was effective. Thus, children would model the behaviours of powerful and influential people around them. The implications for the influences of parents' own behaviours on those of their children can be grasped immediately. In the extension to social cognitive theory, Bandura suggested that children actively interpret the meanings of the behaviours that they observe, and the feedback they receive for their own behaviours, and then internalize messages about their characters and who they are as individuals. Thus, a child develops a core sense of self that guides their likely actions in future social situations. Parental statements to children ("You're such a help:" "You're being very mean") become incorporated into how the children see themselves as social actors.
In family systems theory, the relationship between a parent and child is seen as embedded within the entire family unit, which functions as a dynamic whole. Thus, within a two-parent family, a parent is also a partner and has a spousal relationship as well as child-rearing relationship. If there are siblings, a child is also a brother or sister and has sibling relationships. Each individual has multiple roles within the family, and each relationship influences the functioning of each individual. The multiple family relationships provide the context within which specific interactions, such as those between a parent and child, need to be understood. When one relationship within the family changes, such as marital dissolution or divorce, this necessarily impacts every other relationship in the family and leads to redefinitions of individuals' roles. Although it has directly generated relatively little empirical research, family systems theory has considerable appeal as a theoretical perspective from which to examine the literature on family types and children's social competence.
These theories and others (e. g. Bronfenbrenner's ecological model; 1989) form a broad basis on which current socialization research rests. However, it would be accurate to say that for the past few decades, the majority of research on family types, parental socialization, and children's social competence is conducted without explicitly or exclusively drawing from any specific theory. Rather, a more integrative and multi-faceted perspective seems to guide most empirical research. A newer omnibus theory, or ‘macro-paradigm’ that reflects this eclectic approach to research is developmental psychopathology (Luthar, Burack, & Cicchetti, 1997). Rather than adhering to the tenets of a single theoretical perspective, developmental psychopathology acknowledges that all theories are likely to contribute important ideas, and drawing from each will give researchers the greatest opportunities for understanding development. This perspective also denies that psychology, or any other field of study, has a unique disciplinary advantage. Child development will be best understood by combining the strengths of work in psychology, education, sociology, anthropology, psychiatry, and other fields. Finally, as already discussed, development is recognized as a series of trajectories that are multiply-determined. A variety of risk and protective factors combine and interact to shape each child's individual pathway of growth (Rutter, 2000).
2.2 Risk and protective factors 
Risk factors are those influences that undermine children's positive functioning and direct children along adverse trajectories toward undesirable or maladaptive developmental outcomes. Poverty, child abuse, parents' spousal conflict, chronic illness, and neurocognitive problems are clear and obvious examples of hardships that are likely to disadvantage children and decrease their chances of developing the skills and abilities of social competence (Mash & Wolfe, 2002). However, risk factors can also be more subtle and harder to detect than these examples might suggest.
Protective factors are those influences that bolster and support children's positive functioning, steer them away from maladaptive behaviours, and guide them along pathways of effective and successful development. Sufficient socioeconomic resources, supportive and attentive parents, stable home lives, mentoring relationship with another concerned and involved adult, good health, and intact intellectual abilities are examples of the normal aspects of life that provide a child with the resources to thrive and achieve social competence (Masten, 2001). Protective factors are not necessarily rare or special qualities; they are the common strengths of human life.
A child's fate is not decided by any one risk or protective factor. There are abused children who grow to be confident, kind, and successful youths, and there are children from advantaged homes who become delinquent and destructive. But development is probabilistic, and multiply-determined. Risk factors often occur in clusters; the presence of one disadvantage may point to others. As a child's exposure to and experiences of risk factors increases, the likelihood that the child will follow positive and adaptive developmental trajectories diminishes. Conversely, the presence of multiple protective factors offers the child a range of strengths and shields to ward off the ill effects of risk factors. Two children who differ in their history of protective factors, perhaps because of their parents' differing socialization approaches, may react quite differently to the same experience of a risk factor, such as a violent episode at school. The child protected by a more supportive and involved parent would be less likely to suffer lasting ill effects from their exposure to the violence, compared to the child who was relatively less protected due to having a more distant and disengaged parent.
Research on child development under conditions of serious risk, such as chronic poverty, has revealed that the majority of children do not develop serious problems or adjustment difficulties (Buckner, Mezzacappa, & Beardslee, 2003). Children who do not develop problems despite the presence of risk factors are described as resilient. A growing body of literature is attempting to address what makes some children resilient. A recent comprehensive survey of the literature (Masten, 2001) points to the primary importance of two protective factors: intact neuropsychological functions and the presence of an effective parent. When both of these factors are present, children appear to be protected from the worst effects of risk factors, and are more likely to follow trajectories of positive development.
This focus on risk and protective factors points to the fundamental importance of recognizing the broader contexts of the lives of parents and children. The ability of parents to raise their children appropriately and effectively is shaped by their own experiences of stress and support. The social context of the family will also affect how children perceive, internalize and respond to the socialization efforts of their parents. Acknowledging and understanding these broader influences is essential for explaining how socialization processes function, and—should any differences in children's functioning across groups be identified—whether those differences are due to parenting and family variables per se, or due to the varying circumstances of the families' lives.
2.3 Selection criteria for research reviewed in this paper 
For the rest of this paper, we will focus on evaluating the research examining how several aspects of parental socialization and family type are associated with children's social competence, with particular attention to evaluating evidence for differences in social competence depending on whether children's parents are of the opposite sex or of the same sex. This review will be restricted to empirical work. We acknowledge from the outset that the research is drawn primarily from the field of developmental psychology, but some research from other disciplines also is examined. Papers that present opinions, beliefs or theories, without accumulating quantitative or qualitative evidence, are not reviewed. Chapters and review papers that do not include new empirical data are not reviewed. Papers that present case studies, which are detailed i descriptions of the unique qualities of individual lives, are not reviewed, as these cannot be presumed to generalize beyond the individuals described. Papers that present retrospective data, such as adults thinking back upon their socialization experiences during childhood, are not reviewed, as such data are notoriously unreliable and subject to the effects of biased recall. Papers that are deeply flawed methodologically through biased recruitment of participants, inadequate measurement of key constructs, flawed or inaccurate analyses, or inappropriate misrepresentations of their results, are not reviewed, except to serve as examples of problematic research.
2.3.1 Explanation of methodological limitations on inferences 
It should be recognized that socialization research in general, and research on socialization in families with heterosexual versus gay or lesbian parents in particular, has frequently been targeted as suffering from widespread methodological limitations, and criticized as misinterpreting or making incorrect assumptions about the meanings of results (Bell, 1968; Cameron, 1999; Harris, 1998; Lerner & Nagai, 2001; Scart, 1992). Many of these criticisms are warranted. Documenting that a relation exists between two variables, for example, that there is an association between children's levels of social competence and parents' enforcement of rules in the home, does not necessarily mean that the variables have any kind of causal influence on each other. Groups of participants selected from a local or convenient community are not likely to be representative of the broader population, so, for example, observing the social behaviours of a group of children selected from one large preschool does not warrant the assumption that their behaviours are typical of all preschoolers. Observations made at only one point in time do not reveal processes and pathways of development, so one cannot assume that if the children of lone heterosexual mothers and lone lesbian mothers do not differ in social competence at preschool-age, then they also will not differ in social competence when they are older. It is also difficult to interpret any results without appropriate standards or points of reference for comparison; if adopted children raised by two gay fathers are found to differ in social competence from children raised by their birth parents in heterosexual-couple families, would adoption status or family type be the probable cause of these differences? These are all valid and unquestionable limitations, at least some of which apply to most of the studies that will be reviewed in the remainder of this paper.
But this recognition of limitations does not mean that the studies reviewed herein are fundamentally flawed, or that their results are invalid, or that we cannot draw from them in order to understand the socialization of social competence across family types. It is extremely difficult, time-consuming and expensive to do research in developmental science. Every kind of research procedure has its strengths, but also its limitations. No single study will ever be able to overcome any and all methodological limitations. The key to having confidence in the results of scientific research is replication. If the same results are documented by independent researchers, using independent samples of participants, then we can have greater confidence that the results are valid. The weaknesses of one methodology can be addressed by the strengths of a second. If two different studies, utilizing two different methodologies, show convergent results, then we can feel more certain that the results are not due to the limitations of a given methodology. When a consistent message is seen across the majority of studies within a large body of research, we can state with confidence that the message should be accepted.
Prior to reviewing these studies, it is useful to begin by briefly explaining the important features, strengths, and limitations of the most common empirical designs and methodological features used in developmental research.
18.104.22.168 Correlational versus experimental design
In correlational research, two or more variables are measured, and the relation(s) between the variables are examined. (A note on language: two variables share a ‘relation’, whereas two people share a ‘relationship’.) If higher levels of one variable are associated with higher levels of another variable, they are said to be positively correlated. For example, if emotionally warm parents were found to have children who were more socially competent than the children of emotionally cool parents, then parental warmth and child social competence would be said to be positively correlated. Conversely, if higher levels of one variable are associated with lower levels of another variable, the variables are negatively correlated. Correlational methods are the most common techniques used in socialization research. Investigators observe and measure various aspects of what they see in families, or at schools, or in communities, and examine how strongly and in which directions these aspects are related to each other. Thus, correlational research tells us whether variables in the world are related to each other. However, it cannot tell us why these relations exist. The fact that two variables are associated with each other does not suggest that the level of the first variable causes the level of the second variable. In the preceding example, we do not know whether more parental warmth makes children more socially competent, or whether being a more socially competent child makes parents feel more warmly toward the child, or whether some third factor—say, the family having greater economic resources—causes both parental warmth and child social competence to be higher.
That kind of causal information can be derived from experiments. When conducting an experiment, a researcher is interested in testing whether changes in one variable cause changes in a second variable. The researcher manipulates the first variable in some way, in order to create at least two different conditions. The researcher then randomly assigns research participants (e. g., parents or children) into one of the conditions. Thus, one group of participants experiences the first variable as it exists in one manipulated condition, and a second group of participants experiences the first variable as it exists in a different manipulated condition. The researcher then measures the level of the second variable for all participants. If the level of the second variable differs for the group in one condition compared to the group in the other condition, then the researcher can infer that manipulating the first variable caused the level of the second variable to change. Thus, an experiment can show that a relation can be produced between two variables, and the variables are related in a causal way.
Importantly, the fact that a researcher can manipulate conditions and produce a causal relation between variables does not mean that the causal relation exists naturally and operates in the real world. Thus, experimental and correlational procedures are complementary. Observations and measurements in the real world can tell us that a relation between variables does exist, and experiments can tell us whether that relation appears to be causal. However, true experiments are very, very rare in socialization research. For both ethical and practical reasons, investigators do not have the ability to randomly assign children to parents, or parents to family types. When an investigator measures whether two pre-existing groups differ, for example, whether children with opposite-sex parents differ in social competence from children with same-sex parents, the investigator has used a correlational procedure, not an experimental one. If a difference were to be found between the two groups of children, it could not be assumed that living in different family types produced or caused that difference. It is worth noting, though, that the absence of differences can be informative about the absence of causal relations. Causation cannot exist in the absence of correlation. If family type really does affect social competence in a causal way, and if families are carefully matched on all other potentially relevant factors (age, race, education, income, number of children, etc.) such that the only difference between groups is family type, then some differences across family types should be observed. Observing those differences using correlational techniques would not allow researchers to definitively say that variations in family type were the unquestionable cause of the differences in social competence. Failing to find differences, though, would support the argument that variations in family type do not cause variations in social competence.
22.214.171.124 Single time-point/cross-sectional versus longitudinal design
One way in which socialization researchers have attempted to overcome their limited opportunities to conduct meaningful experiments and still gain some insight into probable causal relations between parent and child variables is by conducting longitudinal research. Most correlational studies are static. They measure how things are "now," that is, what the relations between variables are at a single point in time. This does not provide information about change and growth. Some correlational studies include two or more groups of participants at different ages, such as preschool-aged children and elementary school-aged children; these are called cross-sectional studies. However, these are still static, single time-point studies. Finding that younger and older children differ on some variable suggests that the difference is associated with age and maturation, but it does not reveal how or why that apparent developmental change occurs.
Longitudinal studies can provide this information. In longitudinal research, the same group of participants is followed over an extended period of several months or years, and the same variables are measured repeatedly. Thus, it is possible to chart the nature of change over time. Importantly for socialization research, both parent and child variables can be measured, and changes in the relations of those variables can be observed. Longitudinal techniques also allow researchers to progress from stating whether one variable is related to another variable, to stating whether one variable predicts another variable over time. If the earlier levels of one variable, such as parental involvement, predict the later levels of a second variable, such as children's prosocial behaviour, this provides greater evidence that there might be a causal relation of the if first variable to the changes produced in the second variable. Even longitudinal studies do not produce incontrovertible proof of causal relations, but they offer more support for them than do single time-point correlational studies.
Longitudinal studies are expensive, time-consuming and difficult to conduct. However, they are being used with increasing frequency by socialization researchers. Several of the studies on the relations between parenting practices and children's social competence that are reviewed in the subsequent section are longitudinal studies. However, as it is a newer and smaller field of work (Patterson, 2002), few longitudinal studies of children with gay or lesbian parents have been completed. Therefore, conclusions about the possible causal influences of parenting practices on children's social competence are likely to be more defensible than conclusions about whether family type impacts social competence.
126.96.36.199 Convenience versus random/representative sampling
Most socialization studies also utilize convenience samples. These are groups of voluntary participants that the researchers obtain from the local community. Convenience samples cannot be assumed to be representative of the overall community, or of the broader population. For example, in North America, most convenience samples typically are comprised of mostly White families from middle- to upper-middle socioeconomic levels who live in towns or cities with universities. Families from lower socioeconomic groups, from ethnic or racial minority groups, and from rural areas are relatively under-represented. The results of studies conducted with a select and limited sample simply might not apply for families living in very different cultural, social, and physical realities. Some socialization researchers have designed their studies to examine processes specifically within these under-represented groups, and those studies will be reviewed in order to examine whether the links between parenting and social competence appear to vary depending on socioeconomic status (SES) or other group characteristics.
A smaller number of studies have been completed that include samples of families that are representative of the general population. These are typically large studies of several hundred to several thousand participants. Epidemiological (popular-based) techniques, such as utilizing census data, are used to recruit samples that are as diverse as the province, state, or country within which they live. The results of the studies therefore can be generalized broadly. Two recent studies have used data from large, representative samples in order to study the social competence of children living in families with same-sex parents (Golombok et al., 2003; Wainright, Russel, & Patterson, 2004).
188.8.131.52 Sources of information
A final consideration for understanding the results of research is knowing about the sources of information. Most researchers use questionnaires. Parents, children, and/or teachers are asked to complete pencil-and-paper measures of the variables of interest. Thus, these researchers are examining the reported levels of the variables. Many researchers use direct observations of behaviour. Parents are watched while they interact with their children or children are watched while they interact with their peers. Thus, these researchers are examining the observed level of the variables. Each has its advantages and disadvantages.
Researchers cannot be with participants twenty-four hours a day, seven days a week, and therefore they cannot see participants' behaviours in all circumstances. Participants' reports of their behaviour may be more complete and representative than researchers' direct observations. And of course, some things cannot be observed directly, such as beliefs about childbearing, or interpersonal perceptions; participants need to be asked to provide the information on those kinds of variables. However, participants' reports may or may not be completely honest and accurate. People may not be consciously aware of their levels of skills or abilities, relative to others. Parents may be reluctant to report parenting practices that could be considered less than ideal. Children may not have the maturity or cognitive capacity to reflect upon their own actions and describe them accurately. Reports may be biased, and must be interpreted with some caution.
An even greater concern emerges when questionnaires measuring different variables are obtained from a single respondent. Such data are not independent of each other, and this lack of independence limits what researchers can infer about the nature of relations between variables. For example, adolescents might be asked to describe both their own capacity to influence their peers and the quality of their relationships with their parents. The fact that the two measures of peer and family relationships are both obtained from adolescents might result in apparent relations between these variables because (a) individuals have a style of answering questionnaires that is used for both measures, such as preferentially endorsing the extreme ends of rating scales, or (b) the adolescents project their beliefs that their relationships are or are not related, or (c) the peer and family relationships truly are related.
Behaviour that is directly observed is free from the possible biases that can affect reports on questionnaires. The researchers are examining what can actually be seen in the home, at school, in the playground, in the laboratory, or wherever they observe their participants. Researchers cannot watch people forever or in all contexts, though; therefore observations are not efficient ways of measuring behaviours that occur rarely or that are only presented in specific, difficult-to-access locations. Participants also usually know that they are being observed, and this might cause them to hide or selectively present certain actions. As mentioned above, some variables of interest, such as beliefs and other cognitive processes, may not be directly observable behaviours. Finally, it is much more expensive to directly observe behaviours than it is to distribute questionnaires, and it is more work to process observed data.
Both reported and observed data are useful, and both have their limitations. It is inaccurate to infer that one kind of data is always superior. When used correctly, both can provide informative, objective information. Some of the best socialization studies combine these approaches, using both reported and observed measures of variables. In the following review, we will point out the constraints on conclusions or inferences that can be drawn from specific studies because of the nature of data obtained by researchers.
|The copyright holder of this work allows anyone to use it for any purpose, provided that due diligence is exercised in ensuring the accuracy of the materials reproduced; that the Department of Justice Canada is identified as the source department; and that the reproduction is not represented as an official version of the original report.|
|Please check that the conditions given above are compliant to the Wikisource copyright policy. Most importantly, derivative work and commercial use must be allowed.
This is not an official version of the original report.