Download data and study materials from OSF
University of Konstanz
Sample size: 2100
Field period: 06/17/2010-11/15/2010
Both sociological and sociobiological research suggests a link between parental investment and social status. Sociological research shows that parents from higher social strata follow a parenting style that aims at reducing preexisting differences in children, but parents from lower strata one that is based on an idea of natural development and one that may reinforce differences (Hsin, 2009; Lareau, 2003). Sociobiological research on the Trivers-Willard effect suggests that parents of higher socioeconomic status bias their investment towards sons and parents of lower status towards daughters (Kolk & Schnettler, 2011; Trivers & Willard, 1973). To elucidate the underlying causal pathways, it is important to study behavior and attitudes about parental investment. In sociology, norms have played an important explanatory role in making sense of human behavior, but they have rarely been measured empirically and with regard to the specific conditions under which they apply (Bicchieri, 2006). The Trivers-Willard hypothesis is derived on the basis of evolutionary reasoning in which the proximate mechanisms remain elusive. Following the emphasis of evolutionary psychologists on the novelty of environmental cues in contemporary societies (Barkow, Cosmides, & Tooby, 1992), we may find situations in which predispositions don't manifest themselves in behavior, but are nevertheless visible in normative beliefs, attitudes, or emotions. I examine how respondents' favorability and fairness ratings of certain forms of parental investment are conditional on the same child characteristics that are assumed to trigger differential investment behavior by parents. Using a factorial survey design I can decrease social desirability biases.
H1a: With increasing socioeconoimc status, respondents are expected to judge differential parental investment as fairer when the benefiting child is a son rather than a daughter in a mixed-gender sibling dyad.
H1b: With decreasing socioeconoimc status, respondents are expected to judge differential parental investment as fairer when the benefiting child is a daughter rather than a son in a mixed-gender sibling dyad.
H2a: Parents of lower socioeconomic status are more likely to approve of differential parental investment when the investment difference is according to or even reinforcing pre-existing differences.
H2b: Parents of higher socioeconomic status are more likely to approve of differential parental investment when the investment difference alleviates differences between the siblings.
The experimental design consists of two vignettes with a combination of eight conditions of child characteristics each. In the first vignette, a fictitious sibling pair is described in which one sibling is breastfed and the other one formula fed. Respondents are asked to rate how fair they evaluate the mother's decision for breastfeeding only one child. Child characteristics that are randomly assigned are birth order of the breastfed child, sex of the breastfed child, and sex of the formula-fed child. In the second vignette, parental investment is operationalized as time spent with children. Here, a fictitious situation is described in which one child was born with low (LBW) and one with normal birth weight (NBW). In order to further reduce social desirability bias, respondents are introduced to three equally legitimate time investment strategies: compensation by more time spent with the LBW child, reinforcement by more time spent with the NBW child, and equal treatment. The respondents then learn that the fictitious parents chose the equal treatment strategy and are asked to rate how fair they find the decision of the fictitious parents. Child characteristics that are randomly varied are birth order of the LBW child and the respective sex of the LBW and NBW child.
The outcome variable is based on respondents' fairness ratings of the vignettes presented to them. I used two measures to evaluate the propriety of fairness norms and, in response to a recommendation of one anonymous reviewer, one measure for the validity of the fairness norm. Cronbach's alpha for combining the three ratings is high for both vignettes (.83 for vignette 1; .79 for vignette 2). Therefore, for each of the two vignettes I combined the three respective fairness ratings into one index that is built as the arithmetic mean of the three separate ratings that each range between -5 and +5 on a 11-point scale.
Based on a number of t-tests we don't find statistically significant differences in mean fairness ratings between conditions (that is, between different combinations of child characteristics). Two separate regression analyses reveal that strong main effects of socioeconomic status (education, income) exist but no interaction effect between the sex of the (dis-)advantaged child and parental status. That is, whereas hypotheses H1a and H1b must be rejected on statistical terms, hypotheses H2a and H2b cannot be falsified. The effects of education and income are both statistically significant if entered as sole independent variable in linear regression models, but only education remains significant when both education and income are entered into the respective models. In both examples of parental investment (breastfeeding, time investment), parents with higher education are predicted to rate equal treatment as fairer on average than parents with lower education. In multivariate regression analysis, when child characteristics are controlled for, the status effects remain stable and significant, but child characteristics don't play a role in explaining fairness evaluations. The strength of the status effect differs and the overall percentage of variation in fairness ratings explained varies between parental investment indicators. In conclusion, the findings on differences in parental fairness norms are consistent with findings on actual parental investment differences. Therefore, the current findings confirm the role of parental norms developed by Lareau (2003) on the basis of qualititive research. The factorial survey design seems appropriate to study in more detail for what types of child differences and on what dimensions of investment the reinforcement and reduction norms apply. The current study could show that time investment and breastfeeding are salient dimensions of parental investment where these norms seem to apply.