Taylor (Bambas) Rosecrans and Dr. Christopher Karpowitz, Department of Political Science

The purpose of my study is to find a link between government welfare programs and the political participation of recipients. This kind of study is called a policy feedback analysis. It examines a public policy to see what effect it had on human behavior. There is basis for this sort of research in political science. One highly respected work that was one of the first policy feedback studies was Suzanne Mettler’s book “Soldiers to Citizens” (2005). Mettler found that after WWII, the G.I. bill helped former soldiers to feel like highly valued citizens, funding their education and inspiring them to be involved in their communities. This generation became very politically and socially involved.

The policy I am studying is the government program Temporary Assistance for Needy Families (TANF), which took effect in 1997. This welfare program provides financial assistance to poverty-stricken families. My theory is that TANF has an opposite effect from the G.I. bill. Welfare programs are believed to have a paternalistic effect on recipients (Bruch et. al.; Mead 1997), and so I theorize that TANF depresses political participation because it decreases their sense of self-efficacy.

My methods were quantitative. When TANF was instituted, the program allowed states to set up their individual programs how they saw fit, as long as they met some basic guidelines. Because different states could have different policies, I decided to look for variation between how “strict” a state’s eligibility and activity guidelines were for recipients to see if “stricter” states had less political participation. This does take the assumption that stricter guidelines constitute a stronger paternalistic effect.

To measure the “strictness” of a state’s guidelines, I used the Welfare Rules Database (WRD) created and managed by the Urban Institute. The database keeps track of all the welfare rules in each state and sorts them into more than 500 variables. I chose the July 2008 dataset because it corresponded with the dataset I’ll explain in the next paragraph. I examined each variable and collected 20 variables for the 50 states and District of Columbia that I determined were reasonable measures of “strictness.” I then coded these for use in the statistical software program STATA.

To measure the political participation of these recipients, I used the 2008 Cooperative Congressional Election Study (CCES). The CCES is a large, nation-wide survey of more than 30,000 people conducted before and after national elections. This survey contained information about respondents and their voting and political activities. The survey did not have a question asking whether or not a person was receiving welfare, but it did have enough demographic information to approximate whether a respondent was eligible for welfare. In an estimation that did not take into account the different states’ many specific guidelines on eligibility, I went through and narrowed the respondents to those who would likely be eligible for welfare according to marital status, number of children under 18, family size, their state’s cost of living, and their income compared to federal poverty level guidelines. This brought me down to about 1,000 respondents.

A simple, though imperfect, measure of political participation is voter turnout. I took the welfare-eligible respondents, sorted them by state, and looked at the CCES variables measuring who reported voting in the general and primary elections, and whose votes were validated by public voter records. Taking these numbers, I then analyzed them against the WRD data.

This is where I ran into some problems with validity. Because the 1,000 welfare-eligible respondents were split between 50 states, I had some states with 40+ respondents, while others had less than 10. Rhode Island had no welfare-eligible respondents. I could hardly compare a state that had 4 respondents who all voted, creating a 100% voter turnout rate, to a state that had 30 respondents and a 57% turnout rate. The results of my analysis reflected this problem, showing skewed results not at all reflecting a state’s strictness in TANF guidelines.

I’m continuing to try to work with the data to possibly group states together that have similar guidelines or look at individual WRD variables instead of their sum. I plan on presenting my results to the Mary Lou Fulton poster conference at BYU on April 11, 2013. Though my results are not significant as of yet, I appreciate the opportunity to work on this project and experience original research. I hope that my finished results can be of some benefit to the social science community.