Elsevier

Social Science Research

Volume 36, Issue 3, September 2007, Pages 1087-1104
Social Science Research

Explaining sibling differences in achievement and behavioral outcomes: The importance of within- and between-family factors

https://doi.org/10.1016/j.ssresearch.2006.09.002Get rights and content

Abstract

Most research on child behavioral and cognitive outcomes focuses on the impact of variables measured across families—holding a number of other characteristics constant. However, this research is limited in that it does not capture variation in child developmental outcomes that occurs within families. To address these limitations, we examine correlations of child outcomes between siblings from the same family. We conduct this analysis for several demographic subgroups. Furthermore, to better understand how these inequalities are generated within families, we also examine the impact of individual level characteristics within families using fixed effects models. Results from our between-family analyses indicate that siblings with fewer family resources are more similar on behavioral outcomes compared to siblings in more privileged families. However, children in two-parent households perform more similarly on age-adjusted achievement tests than do children in single parent households. Results from our within-family sibling comparisons reveal that first born children generally outperform their younger siblings on age-adjusted achievement tests.

Introduction

Sociologists, social epidemiologists and developmental psychologists have long recognized the importance of both family environment and socio-economic conditions for healthy child development. The typical approach of researchers in this vein is to control for a number of key background variables—such as race, maternal age or education—and then focus on the marginal effect of family characteristics or one or more socio-economic inputs such as income, family size, parental age or even neighborhood characteristics (for a review, see, e.g., Hauser, 1994). Child outcomes studied with this methodology run the gamut from birth weight to verbal ability to mortality. Such models generally explore variance between families.

Within-family differences in sibling characteristics are also important predictors of life outcomes. However, fewer studies have looked at child-specific characteristics that may account for sibling disparities in achievement or behavior among young children. A great deal of prior research has considered sibling differences in a variety of adult outcomes, focusing mainly on personality (Ernst and Angst, 1983), but also examining health outcomes (Rosenzweig and Wolpin, 1988), social attitudes (Freese et al., 1999), educational attainment (Travis and Kohli, 1995), and adult socioeconomic outcomes (see, e.g., Conley, 2004 for a review of this literature). Few studies, however, have examined the relationship between sibling-specific characteristics such as birth order, birth weight, or gender and young children’s outcomes, such as math and reading achievement or behavioral problems.

Studies examining the relationship between sibling birth order—for instance—and achievement among young children have yielded mixed results, with some research finding a significant relationship and other research finding no association For example, Hanushek (1992) finds a U-shaped relationship between birth order and achievement. Hanushek (1992) studied first- through sixth-grade reading and vocabulary scores and found that while there were no birth order effects for small families, for large families there was a U-shaped effect—children in the earliest and latest birth positions significantly outperformed those in the middle. Because this study was conducted on a select sample of low-SES black families in the early 1970s, however, it may have limited generalizability. By contrast, Steelman and Powell, 1985 studied children who were ages 6–11 and ages 12–17 in 1963–1965 and 1966–1970, respectively, and found no significant relationship between birth order and children’s teacher-reported academic ability. Although this study is based on a national sample of children, the sample is restricted to intact families. Further, the measure of children’s academic performance is based on a 3-category teacher rating of each child’s intellectual ability and academic performance as “clearly below average,” “average,” or “clearly above average.” These measures, however, may be insensitive to subtle differences in student achievement or intellect.

Like birth order, few studies have examined within-family gender differences in behavior and achievement for young children. Nevertheless, between-family analyses of gender differences are plentiful and consistently suggest that boys are more prone to delinquent behavior and juvenile delinquency than are girls (see, e.g., Piquero et al., 2005). Further, some studies that examine gender differences in cognitive outcomes between families suggest that girls tend to outperform boys on reading achievement assessments in secondary school, while boys tend to outperform girls on mathematics achievement tests (see, e.g., Downey and Yuan, 2005). Although these studies focus on between family differences, to the extent that gender is orthogonal to other important correlates of these outcomes, these studies should come to the same conclusions as within-family analysis.

Finally, research also confirms that birth weight is important for young children’s cognitive outcomes. For instance, when examining the impact of birth weight on the Peabody Individual Achievement Tests of Math and Reading, Boardman et al. (2002) found that, among children ages 6–14, low birth weight exerted a negative effect on both scores, and the effect was especially pronounced for the younger children in their sample. Although this study is unique in that it included siblings from the same family, they were pooled together with children from different families making it impossible to differentiate the between-family from the within-family effects. Birth weight also appears to exert a significant effect on behavioral outcomes of elementary school students. Klebanov, Brooks-Gunn, and McCormick, 1994 found that children with low birth weight were rated by their teachers as having poorer classroom behavior, as well as lower attention and higher daydreaming and hyperactivity scores than students with normal birth weight. However, this study did not examine the relationship between birth weight and childhood outcomes among siblings in the same family.

In the present study, we will build on this work in several ways. First, we will assess whether within-family variance is constant across family types by examining family background differences in sibling correlations on behavioral and achievement outcomes. We use the intra-class correlation procedure in SPSS, which is essentially a one-way ANOVA analysis that will calculate within-family sibling correlations for achievement and behavioral outcomes. These correlations will assess the global impact of family background on our outcome measures, rather than the marginal impact of measured background variables. If this global impact varies by demographic group, it will give us a better understanding of the system of stratification as it impacts children. For example, previous research (Blau and Duncan, 1967) has indicated that family background exerts a weaker effect on the adult socioeconomic outcomes of whites vis-à-vis blacks; however, less research has examined whether these patterns are also present in early childhood measures.

Second, we will adopt a sibling fixed effects approach to assess child cognitive and behavioral measures. In addition to examining differences between families, it is important to examine the individual characteristics that might generate inequalities within families. The within-family variables we will model will not be parental characteristics that vary over time—as has typically been done in the literature—but rather child-specific factors, including birth weight, gender and birth order.

While the fixed effects approach to modeling sibling variability has not been extensively used in the literature on child development, it has been applied to adult outcomes such as IQ or educational success (see, e.g., Kuo and Hauser, 1995, Solon et al., 1991). Some researchers have used a similar approach to parcel out genetic and environmental influences on child outcomes and have met with various criticisms. That is not the agenda of the current study, which views the common family background (genetics, culture, economic resources and so on) as a single unit for study. From these over-arching objectives, specific research questions emerge:

  • (1) First, how do sibling differences in young children’s behavioral and achievement outcomes vary across different family types, based on demographic characteristics, including family structure and family resources?

  • (2) Second, how do child-specific factors, including birth weight, gender, and birth order, affect within-family sibling differences in young children’s developmental outcomes, including cognitive achievement and behavioral problems?

What types of developmental outcomes tend to be unique to individuals regardless of their family of origin (high within-family variance) and which tend to cluster within families (high between-family variance)? And do certain family structures exhibit more “clustering” of sibling behavioral and developmental outcomes than others do? For example, do children from single-parent households emerge relatively closer in behavioral and developmental outcomes than do those from two-parent households? McLanahan and Sanderfur (1994) have shown that children from single-parent families have lower average levels of cognitive functioning than those from two-parent families (though much of this effect is due to lower economic resources in these households). But this is different from asking whether the outcomes of children from single-parent homes are more similar to each other than are those of children from dual parent families.1 This effect may be due to a lack of social control and stability in the form of the second parent (resulting in greater outside influences and randomness in lone parent families—see, e.g., Teachman et al., 1998). On the other hand, with a single role model, children may turn out more cognitively similar when changing family environment is held constant.

The same can be asked of family size: Does a larger family lead to greater family socialization—or, on the other hand—more randomness due to a lack of social control resulting from divided parental attention? (It could also be the case that the more complex social networks within larger families lead to greater differences.). Prior research has generally found an inverse relationship between family size and educational achievement, net of family background (e.g., Downey, 1995), although some find no family size effect (Guo and Van Wey, 1999). However, as most research has found a significant relationship, we predict that siblings from larger families will have lower siblings correlations in childhood outcomes.

Likewise, do siblings spaced more closely together tend to resemble each other more in behavioral and developmental outcomes than do those spaced further apart? Previous research has found that siblings spaced closer together have lower average cognitive functioning (e.g., Powell and Steelman, 1993). One explanation for these results is that close spacing may yield less variability since siblings experience critical family transitions (such as income shocks and the like) at similar times in the course of their development (e.g., Elder and Caspi, 1988). There also may be more interdependence (i.e. cross-socialization) among those who are not too far apart in age. An alternative hypothesis for this finding is that the closer siblings are in age, the more they are in direct competition with each other (Stafford, 1987) and the more they will diverge in their outcomes since small differences in abilities become magnified (Sulloway, 1996). Closely spaced siblings may also seek out different niches within the family.

And what about the age of parents? Prior research suggests that children in households with older mothers have higher cognitive test scores and fewer behavioral problems, although this relationship is greatly reduced after controlling for background characteristics (e.g., López Turley, 2003). One explanation for this finding is that younger parents may be less equipped to handle the rigors of childrearing, both financially and emotionally. If this is the case, children in these households may receive less consistent parenting. As a result, we’d expect siblings in families with younger parents to have lower correlations than siblings in families with older parents.

Finally, how do family resources affect the developmental “spread” between siblings? Previous qualitative work (Conley, 2004), which examined adult outcomes, suggests that among disadvantaged households, sibling disparities tend to increase since limited opportunities and resources may evince parenting strategies that accentuate sibling differences by directing family resources to the one (or few) sibling(s) for whom upward mobility is most likely. This research also suggests that among families that are well-endowed with class resources (and are racially privileged as well), parents often invest more heavily in those offspring they see as having the worst chances for success in the education system and/or labor market—in a compensatory fashion. Put another way, disadvantaged families are seen to be behaving efficiently (investing more in the offspring for whom they expect higher returns) thus reinforcing sibling differences, whereas better-off families appear to be behaving inefficiently (investing more in the kid for whom they expect lower returns), thus trying to bring about more equity in the outcomes of offspring. Thus, our motivating hypothesis is that among families that are disadvantaged, we should observe greater sibling disparities (i.e. lower correlations). With this in mind, we hypothesize that blacks have lower sibling correlations than non-blacks and that siblings whose mothers have completed fewer years of education have lower correlations than do those whose mothers have completed more years of schooling.2

Such a hypothesis stands in contrast to the theoretical prediction of Becker and Tomes (1986) who posited that with capital constraints, low-income parents may not be able to optimally invest in their children’s human capital. Such under-investment may lead to higher degrees of sibling resemblance at lower incomes since “high ability children from poor families may receive the same low level of education as a sibling with lower academic ability, compressing their earnings compared with similarly different siblings from a prosperous family” (Mazumder, 2003, p. 16). This prediction, however, is based on the notion of capital (and credit) constraints with respect to the cost of formal schooling. For our early childhood developmental measures this may be less of an issue than it would be for ultimately, completed schooling.

Given patterns of sibling resemblance and difference in child outcomes, what predicts which siblings thrive early in childhood and which do not? Here we will rely on a number of individual characteristics that have been shown to matter to varying degrees in other contexts to individual outcomes both within and between families. First we will examine the effect of birth weight on cognitive and behavioral outcomes. As reviewed, previous research has found a significant relationship between birth weight, educational outcomes, and behavioral problems. Since our measures of young children’s achievement and behavior come at a time when premature or small birth weight babies have had limited time to “catch up,” effects may be more pronounced for the early childhood measures we will examine. On the other hand, early childhood measures demonstrate greater degrees of measurement error, suggesting an attenuation of coefficients. However, given the strength of birth weight effects in the literature, we hypothesize that children with higher birth weight will have higher cognitive outcomes and fewer behavioral problems.

Second, we will examine the impact of birth order. As previously mentioned, few studies have examined early childhood cognitive and behavioral measures, and the results of prior research are largely mixed. Moreover, these studies are limited in the use of a select sample (Hanushek, 1992) or teacher-reported measures of children’s educational performance, which might be unreliable (Steelman and Powell, 1985). Even more important is that although the studies we review control for family size when examining the effects of birth order, many studies that do not use within-family (i.e. sibling fixed effects) comparisons typically conflate sibship size effects with parity effects. Because previous research has yielded mixed results and has not used sibling fixed effects, our findings may differ from prior studies.

Finally, how does a sibling’s gender affect children’s cognitive achievement or behavioral outcomes? Between-family studies of boys and girls suggest that boys generally exhibit more behavioral problems and have lower reading achievement test scores. Therefore, we hypothesize that brothers will have significantly lower reading achievement and significantly more behavioral problems than their sisters. Prior studies also suggest that boys tend to outperform girls in math achievement. Consequently, we also hypothesize that sisters will tend to have lower math scores compared to their brothers.

Section snippets

Data and methods

Our analysis draws on data from the Panel Study of Income Dynamics (PSID), which is a nationally representative sample of households and individuals in the United States. The PSID began in 1968 and presently continues to follow families and individuals. The PSID collects information from both individuals and families, primarily focusing on economics and demographics, including income, employment, family composition, and residential location. Research indicates that when weights are used, the

Results

Table 1 presents the sibling correlations for the various achievement and behavioral problem outcome variables, as well as age and birth weight. As evident in the table, there is quite a bit of variation between siblings on the achievement and behavioral indicators, with correlations ranging from .284 on the Passage Comprehension scores to .441 for the Broad Reading scores. In the following models, we attempt to explain how such within-family sibling variation in achievement and behavioral

Discussion

Most research on child behavioral and cognitive outcomes focuses on the impact of variables measured across families—holding a number of other characteristics constant. However, this research is limited because it does not capture variation in child developmental outcomes that occurs within families. To address these limitations, we examine sibling correlations on behavioral and achievement outcomes across various family types. Results show that, overall, sibling correlations in early childhood

References (30)

  • G.S. Becker et al.

    Human capital and the rise and fall of families

    Journal of Labor Economics

    (1986)
  • P.. Blau et al.

    The American Occupational Structure

    (1967)
  • J.D. Boardman et al.

    Low birth weight, social factors, and developmental outcomes among children in the United States

    Demography

    (2002)
  • D. Conley

    The Pecking Order: Which Siblings Succeed and Why

    (2004)
  • D.B. Downey

    When bigger is not better: family size, parental resources, and children’s educational performance

    American Sociological Review

    (1995)
  • Douglas B Downey et al.

    Sex differences in school performance during high school: puzzling patterns and possible explanations

    The Sociological Quarterly

    (2005)
  • G.H. Elder et al.

    Economic stress in lives: developmental perspectives

    Journal of Social Issues

    (1988)
  • C. Ernst et al.

    Birth Order: Its Influence on Personality

    (1983)
  • J. Fitzgerald et al.

    An analysis of sample attrition in panel data: the Michigan panel study of income dynamics

    Journal of Human Resources

    (1998)
  • J. Freese et al.

    Rebel without a cause or effect: birth order and social attitudes

    American Sociological Review

    (1999)
  • G. Guo et al.

    Sibship size and intellectual development

    American Sociological Review

    (1999)
  • E.A. Hanushek

    The trade-off between child quantity and quality

    The Journal of Political Economy

    (1992)
  • R.M. Hauser

    Measuring socioeconomic status in studies of child development

    Child Development

    (1994)
  • Hofferth, Sandra L., Pamela Davis-Kean, Jean Davis, Jonathan Finkelstein. 1998. User guide for the Child Development...
  • P. Klebanov et al.

    Classroom behavior of very Low birth weight elementary school children

    Pediatrics

    (1994)
  • Cited by (0)

    The authors acknowledge support from the National Science Foundation Social and Behavioral Sciences Division’s CAREER Award and The Robert Wood Johnson Investigator Award.

    View full text