Joe Soss, Sanford Schram, Thomas Vartanian, and Erin O'Brien took the TANF policy decisions that have been made by forty-nine states in the post-1996 period and compared them to various measures of the states' sociopolitical environment and social policy conditions. They suggest that only two of the independent variables that they examined actually have a strong predictive relationship where a state's decision to adopt a family cap is concerned. All other factors being equal, the data suggest that the states with a relatively high proportion of Latinos or African Americans on their TANF rolls were the ones that were most likely to have a family cap. As the percentage of black TANF recipients rises, the probability that a state will adopt a family cap increases from 0.09 to 0.75. The probability of a family cap is similarly lifted, although with a slightly less powerful effect, from 0.19 to 0.63, as the proportion of Latinos on the rolls increases. By comparison, Soss et al. concluded that the correlations between a state's decision to impose a family cap and its unmarried birthrate, case-load-to-population ratio, government ideology, degree of interparty competition, turnout rates among low-income voters, unemployment rate, incarceration rate, and welfare innovation record were rather weak.

The racial and ethnic makeup of the state's TANF beneficiary population appeared to influence other state-level policy decisions, but different factors seemed to be more salient.