Insights on Southern Poverty The newsletter of the University of Kentucky Center for Poverty Research An equal opportunity institution Vol. 6 No. 2, Spring 2009 In this issue... Full length discussion papers of reports appearing in Insights on Southern Poverty are available at www.ukcpr.org. Choose the “Publications” link for “Discussion Papers.” Inluence of race and ethnicity on TANF sanctioning.............. 1 Welfare reform and juvenile violent crime arrests............................................ 3 SCHIP programs and private insurance crowd-out .............................. 4 Impact of Kentucky’s Relocation Assistance Program ............................... 6 Impact of Food Stamp program on employment ............................................. 8 Sanford Schram, Joe Soss, Richard Fording, and Linda Houser This research focuses on how de- cisions to sanction TANF clients are shaped by the interplay of racial sta- tus and character markers. Our analysis falls into a long tradition of research on the ways that major societal institutions produce racial disparities. Despite the broad scope of this research, it is unclear whether such findings might generalize to welfare implementation. To theorize the influence of race, we use the Racial Classification Model of policy choice (RCM) (Soss et al, 2008). Drawing on models of implicit racism (Quillian, 2008), the RCM specifies con- ditions that can produce policy-based ra- cial disparities even in the absence of ra- cial ill will or discriminatory intent. We test hypotheses derived from the RCM’s assertion that racial disparities become more likely when policy targets possess discrediting traits that are consistent with negative group reputations. Extending the work of Pager (2007), we show how stereotype-consistent markers can pro- vide “expectancy confirmation” (Cor- rell et al, 2007), and, hence, strengthen the link between racial status and policy treatment. To meet these goals, we test propositions from the RCM using both experimental and administrative data. Survey experimental design Our experimental data come from a Web-based survey of case managers from Florida’s welfare-to-work program, Welfare Transition (WT). To ensure an- onymity for respondents, the Agency for Workforce Innovation distributed the survey link through e-mail to its 24 Regional Workforce Boards for subse- quent distribution to caseworkers via e- mail. Case managers completed the sur- veys over a two-week period at the end of 2006. We received 144 responses from TANF case managers, for a response rate of about 65%. Our analysis is based on two 2x2 ex- periments, each of which presented case managers with a vignette and asked them to decide whether to impose a sanction. The 2x2 design includes variation on cli- ent race and a discrediting social mark- er. Each vignette portrays a hypothetical TANF participant who has arguably fall- en out of compliance with program re- quirements. For the racial dimension of the 2x2 design, we randomly assigned the client described in Vignette 1 either a Hispanic-sounding name or a white- sounding name. Similarly, we assigned the client described in Vignette 2 either a black-sounding name or a white-sound- ing name. 1 The second dimension of the 2x2 experiments manipulates client mark- ers that are commonly associated with (a) images of undeserving welfare clients and (b) negative stereotypes of minor- ity racial/ethnic groups. Our selection of these client traits is also based on our field interviews, which revealed substan- tial caseworker attention to these two cli- ent types: the young mother of multiple children and the repeat recipient who has been sanctioned in the past. Survey-experimental analysis results To account for possible differences between managers assigned to each con- dition of the experiment, we estimated multivariate models of the sanction de- cision, which we code as a dichotomous variable (1=sanction, 0=no sanction). Our multivariate models include dichot- omous variables representing three of the four experimental conditions (the Experimental evidence on the influence of race and ethnicity on TANF sanctioning (Continued on page 10) 1. To guard against confounding effects that might arise from the use of a specific name, we randomly assigned one of three names for each group in each vignette. White-sounding names included Sarah Walsh, Emily O’Brien, and Meredith McCarthy; African American-sounding names included Lakisha Williams, Aisha Jackson, and Tanisha Johnson; and Hispanic-sounding names included Sonya Perez, Maria Rodriguez, and Luisa Alvarez. To test for name-specific effects, we analyzed responses within each “race condition” to search for significant differences associated with each name. We found no significant differences and, hence, treat all racial name cues as equivalent.