Do Parents’ Social Skills Influence Their Children’s Sociability?

In an exceptional work by Dohmen et al. (2012), the researchers investigate the intergenerational transmission of risk and trust attitudes using the 2003 and 2004 waves of the German Socio-economic Panel (GSOEP).

Psychological studies examining the intergenerational link in traits and behaviors use homogenous subsamples. However, the disadvantages of using homogenous subsamples over large and representative population samples are the following: (1) homogenous subsamples suffer from attenuation bias, (2) they focus on maternal rather than paternal characteristics (Duncan et al. 2005), and (3) the estimated effect of parents’ sociability on children’s wages might not represent the population as a whole.

The assumption that workers hold occupations that match their traits and personalities also corresponds to the following observation by Robert Hauser (1998, 5): “Job-holding tells us about the technical and social skills that we bring to the labor market. … As market labor has become nearly universal among adult women as well as men, it is increasingly possible to characterize individuals in terms of their own current or past jobs.”

Spearman (1904) proposes the existence of general intelligence, termed g, which is a single general factor that governs an individual’s level of intelligence.

The intergenerational link in sociability estimated in this article is not necessarily causal, if there remain unobserved factors that affect children’s sociability and that are correlated with parents’ people skills.

Lundberg (2005) reviews the literature on son preferences and documents that fathers spend more time and are more involved with their sons than with their daughters. This paternal behavior could explain why we find evidence of a stronger sociability link between fathers and sons than between fathers and daughters.

The most widely accepted taxonomy of personality traits is called the Big Five or the five-factor model (FFM). The Big Five factors are openness to experience, conscientiousness, extroversion, agreeableness, and neuroticism. Sociability is included under extroversion, which is characterized by facets such as gregariousness, assertiveness, activity, and outgoingness. See McCrae and John (1992) and Borghans, Duckworth, Heckman, and ter Weel (2008) for details.

Autor, Levy, and Murnane (2003) and Bacolod and Blum (2010) also utilize textual definitions to classify the DOT variables and thereby analyze changes in skill requirements and skill returns in the United States. Ingram and Neumann (2006) perform a factor analysis on the revised fourth-edition DOT data to reduce the data to a smaller set of dimensions. We also implement a factor analysis to corroborate our choice of skill categories. Most of our skill categorizations are consistent with the grouping from the factor analysis.

If information on the parents’ occupations when the respondent was 14 is unavailable, we use their occupation in 1978, when the respondent was between the ages of 13 and 21.

We report robust standard errors clustered by mothers.

Duncan et al. (2005) estimate standardized regression coefficients for mother–child links in sociability (both at age 6) for a sample that includes all racial groups. They cluster the answers on the mothers’ sociability at age 6 into two values: zero for shyness (answers 1 or 2) and one for outgoing (answers 3 or 4). They find that the estimate for sociability is 0.13 for both mother–daughter and mother–son pairs (standard errors are not reported in their article). When we restrict the age range of children to 6, as in Duncan et al., the estimate for sociability is 0.039 (0.034) for the mother–daughter link and 0.004 (0.030) for the mother–son link. The effect for daughters is greater than the effect for the sons, although the difference is insignificant.

For people skills, the first principal component explains 62.8% of the variance in the matrix correlations for fathers and 55.8% of the variance for mothers. For cognitive skills, they are 71.7% for fathers and 68.8% for mothers, respectively; for motor skills, they are 43.1% for fathers and 33.8% for mothers; and for physical strength, they are 67.2% for fathers and 42.1% for mothers.

The estimated factor loadings for each DOT variable are displayed in Appendix Table 4, Panel A for the sample of NLSY fathers and in Panel B for the NLSY mothers. For fathers, the first, second, and third factors are identified as cognitive skills, motor skills, and physical strength, respectively. The fourth and fifth factors are identified as people skills; the fourth factor has a relatively higher loading on “interpret feelings” and “influence people”; and the fifth factor has a relatively higher loading on “deal with people,” “communicate data,” and “talking/hearing.” We label the fifth factor as people skills, because when using the sample of NLSY respondents, we find DOT people-skill variables that are highly loaded on the fifth factor are more strongly linked with their own sociability measures (see Appendix Table 2). For mothers, the first, second, third, and fourth factors are labeled cognitive skills, motor skills, people skills, and physical strength, respectively.

The first principal component explains 32.2% of the variance in the matrix of correlations for fathers and 46.5% of the variance for mothers.

The fathers’ “verbal” skill (the DOT cognitive-skill variable) has a positive effect of 0.210 (0.111) on their sons’ sociability in early adulthood (result not reported in Table 5). “Verbal” skill is likely to be related to both cognitive and people skills because according to the DOT variable description in Appendix Table 1, this skill refers to the ability to understand the meaning of words and use them effectively, to comprehend language, to understand the relationships between words, and to understand the meanings of whole sentences and paragraphs. Therefore, both the adjusted people skills (which do not overlap with “verbal”) and “verbal” skill of fathers have a positive effect on their sons’ sociability.

The correlation coefficient between the Rotter scale and early-adulthood sociability is 0.083. The correlation coefficient between the Rosenberg Self-esteem Scale and early-adulthood sociability is 0.157.

For the regression on the sons’ sociability at age 6, the coefficient of the Rotter score is 0.060 (0.025), and the coefficient of the Rosenberg Self-esteem Scale is 0.101 (0.027). For the regression on the sons’ early-adulthood sociability, the coefficient of the Rotter score is 0.033 (0.027), and the coefficient of the Rosenberg Self-esteem Scale is 0.186 (0.028).

The estimate controls for the same covariates as in Table 5.

Sociability in early adulthood is standardized to have a mean of zero and a variance of one. The regression controls for education, a quadratic in the AFQT score and age, marital status, place of residence (region and urban area), and year dummies.