Evaluating evolutionary changes in state TANF policies

Abstract

Over the past decade narrowly focused studies have evaluated the effectiveness of state-level welfare policies. In general, they evaluate reforms within a particular state, focus on a small number of outcome variables (usually caseload levels) and/or use a very narrowly defined time period. This narrow and partial analysis is perplexing, from an institutional perspective, as Temporary Assistance for Needy Families (TANF) forces states into a zero-sum funding game, where shares depend on differential relative success in achieving policy objectives metrics. This institutional structure incentivizes states to mimic and improve upon more successful counterparts to recapture a larger share of TANF block grants. Given this dynamic institutional structure, an evolutionary evaluation of state TANF programmes is warranted. This article uses cluster analysis to explore evolutionary changes in state TANF policies (as characterized by a comprehensive set of outcome variables) immediately following the imposition of TANF (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005). We identify or benchmark clusters of 'successful' and 'less successful' TANF programmes. The results allow us to track which states in which year fall into the 'successful' and 'less successful' clusters over the 9-year period. The results support the notion that initially unsuccessful states mimic other successful state programmes over time.

Examining the Early Impact of COVID-19 on Single-Parent TANF Caseloads: a Brief Analysis of New Jersey

“…Most of the existing research on safety net programs use traditional cluster analysis methods, such as a k -means cluster analysis and hierarchical cluster analysis (Chambers & Potter, 2008 ; Slack et al, 2014 ; Snarr et al, 2012 ). The limitation of the traditional cluster analysis methods is that it does not allow for use of both categorial and continuous variables in the same model.…”

Mapping the interrelationships between community assets

“…Of particular interest are the linkages between residents' ratings of the importance of each of these community assets. The cluster analysis is applied to the data to group each of these assets into different clusters, or groups, based on residents' perceptions of the importance of each of these assets (Snarr et al , 2012; Cricelli et al , 2018). In doing so, it is possible to gain a deeper understanding of how residents assess the importance of individual community assets and also how those assets are linked in the minds of the community's residents.…”

“…A nonhierarchical cluster analysis presumes a set number of clusters or groups and attempts to classify respondents into clusters based on how similarly or dissimilarly they rate each of the 20 community assets. We employed both types of cluster analyses, each of which employed Ward’s linkages and squared Euclidean distance (Hair et al , 2006; Snarr et al , 2012; Friesner et al , 2018). In addition, our nonhierarchical cluster analysis used k-means clustering with seven clusters.…”

“…One empirical technique that may be useful in appropriately identifying the linkages between the perceived importance of various community assets is cluster analysis (Snarr et al , 2012; Cricelli et al , 2018; Friesner et al , 2018). The technique is purely exploratory in nature.…”