The Relationship between Behaviors of Nonresident Fathers and Children’s Lives: A Critique of Amato and Gilbreth (1999)
In reviewing studies of nonresident fathers, Amato and Gilbreth (1999) noted that the one variable consistently related to enhancing the lives of children was payment of child support. Recognition of the importance of child support has had the positive effect of states making greater efforts to enforce legally imposed requirements. On the other hand, previous research had failed to provide evidence that amount of visitation was associated with any aspects of children’s lives. The researchers proposed that differences in behaviors during visitation were not analyzed but could be associated with children’s well-being. Because in previous studies enough information was available to create measures of two variables, use of authoritative parenting techniques and father/child emotional ties, the researchers used these studies to measure the association of these two variables, as well as child support and visitation, with three measures of children’s well-being, school achievement, externalizing problems (e.g., aggression) and internalizing problems (e.g., depression).
Based on 63 studies, they reported a statistically significant positive Pearson correlation between child-support payment and school success, a negative one with behavioral problems, and no relationship with internalizing behaviors. There was a statistically significant correlation between visitation and school success, a negative one with internalizing problems, and none with externalizing problems. For both authoritative parenting and closeness of parent/child relationship, there were significant positive correlations with school success and negative ones with internalizing and externalizing problems. Acknowledging some weaknesses in the study, for example, in forming the 2 new variables, in interpreting the direction of correlations (e.g., characteristics of children could determine whether fathers behaved authoritatively), as well as the need for further research, Amato and Gilbreth (1999) concluded that their results supported current efforts to change policies so that fathers are better able to be involved in their children’s lives.
The purpose of the study was to evaluate the relationship between each of 4 characteristics of fathers (payment of child support, visitation, closeness of relationship with child, and use of an authoritative parenting style) with 3 characteristics of children (school success, and externalizing and internalizing problems). The researchers reported that “they used the product-moment correlation (r) as the measure of effect size” (563). The particular effect size measured by the statistic r is the size or strength of the relationship between variables, a statistic that can range from -1 to +1, where sign does not indicate strength but whether the relationship is negative (as scores on one variable increase, scores on another decrease) or positive (as scores on one variable increase, scores on another also increase). The size of the absolute value of r measures strength, where values of at least .7 indicate strong relationships, those between .5 and .7 indicate moderate ones, those between .26 and .29 indicate weak ones, and those between 0 and .25 indicate little or no relationships (Morse, 1998). To clarify the meaning of r, consider that a number of variables can account for individual differences in another variable. For example, variables such as height, gender, caloric intake, and cultural trends are associated with, or help account for, the variability in weight. Squaring r provides a measure of the proportion (or percentage) of variability in one variable that can be accounted for by another variable, for example, if the correlation between height and weight were .6, .36 (the square of .6) or 36% of the variability in one of the variables can be accounted for by the other variable. This explanation was required to understand my disbelief when Amato and Gilbreth presented rs in Table 2, headed “Mean Effect Sizes…” (p. 564) but said nothing about the sizes of the rs, commenting only on statistical significance (which allows generalizing the results of a sample to the populations from which the sample was drawn, Morse, 1998). With large sample sizes, very small rs often are statistically significant but meaningless – as were the statistically significant rs reported by the researchers – ranging in absolute values from .03 (!) to .15. How does one respond to a description of a correlation coefficient of -.01 as “not statistically significant… [but consistent with] the overall pattern?” (p. 564).
The above were the only results of interest (other results related to the quality of the measures used, other variables controlled, etc.) and the only warranted interpretation would have been that, contrary to what the researchers reported (pp.568-569), the study did not provide evidence of even weak relationships, as described above, between father behaviors and characteristics of their children. If their measure of child support reflected differences in socioeconomic status, failure to find a relationship with measures of children’s well-being would be surprising. However, based on their description of the data, it seems possible that they correlated the mean percentages of fathers paying support reported in some studies and mean dollar amounts reported in other studies with reported means of well-being, so there would be no reason to expect correlations.
The researchers’ suggestions that policies should encourage enforcement of child support orders and, barring unusual circumstances, facilitate involvement of divorced parents in their children’s lives were certainly warranted. Their conclusion that research was needed on relationships between parent and child characteristics following divorce also was warranted. Indeed, there remains much to be learned about the relationships between characteristics of parents and children in general.
Amato, P.R., & Gilbreth, J.G. (1999). Nonresident fathers and children’s well-being: A meta-
analysis. Journal of Marriage and the Family, 61, 557-572.
Morse, G. (1998). Statistics tutorial: Correlation. Retrieved December 6, 2008, from