Full Citation

Title: Clustering, Spatial Correlations and Randomization Inference

Citation Type: Working Paper

Publication Year: 2010

ISBN:

ISSN:

DOI:

NSFID:

PMCID:

PMID:

Abstract: It is standard practice in empirical work to allow for clustering in the error covariance matrix if theexplanatory variables of interest vary at a more aggregate level than the units of observation. Often,however, the structure of the error covariance matrix is more complex, with correlations varying inmagnitude within clusters, and not vanishing between clusters. Here we explore the implications ofsuch correlations for the actual and estimated precision of least squares estimators. We show thatwith equal sized clusters, if the covariate of interest is randomly assigned at the cluster level, onlyaccounting for non-zero covariances at the cluster level, and ignoring correlations between clusters,leads to valid standard errors and confidence intervals. However, in many cases this may not suffice.For example, state policies exhibit substantial spatial correlations. As a result, ignoring spatial correlationsin outcomes beyond that accounted for by the clustering at the state level, may well bias standard errors.We illustrate our findings using the 5% public use census data. Based on these results we recommendresearchers assess the extent of spatial correlations in explanatory variables beyond state level clustering,and if such correlations are present, take into account spatial correlations beyond the clustering correlationstypically accounted for.

User Submitted?: No

Authors: Barrios, Thomas; Imbens, Guido W.; Kolesar, Michal; Diamond, Rebecca

Series Title:

Publication Number: 15760

Institution: National Bureau of Economic Research

Pages:

Publisher Location: Cambridge, MA

Data Collections: IPUMS USA

Topics: Methodology and Data Collection, Other

Countries: