Social Science Statistics Blog

« November 10, 2005 | Main | November 15, 2005 »

14 November 2005

Applied Statistics - You, Jong-Sung

This week, the Applied Statistics Workshop will present a talk by You, Jong-Sung, a PhD candidate in Public Policy at the Kennedy School of Government. Jong-Sung’s dissertation on “corruption, inequality, and social trust� explores how corruption and inequality reinforce each other and erode social trust. His dissertation chapter on cross-national study of causal effect of inequality on corruption was published in ASR (February 2005) as an article with S. Khagram. His research interests include comparative politics and political sociology of corruption and anti-corruption reform and political economy of inequality and social policy. Before coming to Harvard, he worked for an NGO in Korea, “Citizens’ Coalition for Economic Justice�, as Director of Policy Research and General Secretary. He spent more than two years in prison because of democratization movement under military regimes. He has a BA in social welfare from Seoul National University, and a MPA from KSG. He is also one of the authors of this blog.

The talk is entitled “A Multilevel Analysis of Correlates of Social Trust: Fairness Matters More Than Similarity,� and draws on Jong-Sung’s dissertation research. The abstract follows on the jump:

I argue that the fairness of a society affects its level of social trust more than does its homogeneity. Societies with fair procedural rules (democracy), fair administration of rules (freedom from corruption), and fair (relatively equal and unskewed) income distribution produce incentives for trustworthy behavior, develop norms of trustworthiness, and enhance interpersonal trust. Based on a multi-level analysis using the World Values Surveys data that cover 80 countries, I find that (1) freedom from corruption, income equality, and mature democracy are positively associated with trust, while ethnic diversity loses significance once these factors are accounted for; (2) corruption and inequality have an adverse impact on norms and perceptions of trustworthiness; (3) the negative effect of inequality on trust is due to the skewness of income rather than its simple heterogeneity; and (4) the negative effect of minority status is greater in more unequal and undemocratic countries, consistent with the fairness explanation.

Posted by Mike Kellermann at 11:55 AM

Considering Spatial Dependence in Lattice Data: Two Views

Drew Thomas

Last year during Prof. Rima Izem's Spatial Statistics course, I started to wonder about different analytical techniques for comparing lattice data (say voting results, epidemiological information, or the prevalence of basketball courts) on a map with distinct spatial units such as counties.

A set of techniques had been demonstrated to determine spatial autocorrelation through the use of a fixed-value neighbour matrix, with one parameter determining the strength of the autocorrelation. The use of the fixed neighbour matrix perturbed me somewhat, since the practice of geostatistics uses a tool called the empirical variogram - a functional estimate of variance between sample sites through a regression, based on taking each possible pair of points and computing the difference between squared values - which might give a more reasonable estimate of autocorrelation than a simpler model.

As it turned out, this same question was asked by Prof. Melanie Wall from the Biostatistics Department at the University of Minnesota about a year before I got around to it. In her paper "A close look at the spatial structure implied by the CAR and SAR models" (J. Stat. Planning and Inference, v121, no.2), Prof. Wall tests the idea of using a variogram approach to model spatial structure on SAT data against more common lattice models. And what do you know - the variogram approach holds up to scrutiny. In some cases it outperforms the lattice model, such as in the extreme case of Tennessee and Missouri, which have a bizarrely low correlation due to the fact that each state has eight neighbours.

As well as feeling relief that this difficulty with the model wasn't just in my imagination, I'm glad to see that this type of inference crosses so many borders.

Posted by Andrew C. Thomas at 3:37 AM