Christopher T. Moore
2010-Feb-25 22:59 UTC
[R] Heterogeneous Correlation Matrix with Survey Weights
Hello, I have a data set containing categorical and ordinal factors, as well as sampling weights (i.e., survey weights reflecting unequal probabilities of selection). I want to fit a structural equation model with sem(). I have run sem() on weighted covariance matrices using advice from John Fox (see <http://tolstoy.newcastle.edu.au/R/e5/help/08/12/8773.html> and <http://blog.lib.umn.edu/moor0554/canoemoore/2009/09/sem_complex_samples_r_update.html>). However, since I have categorical/ordinal variables, I would like to compute a weighted heterogeneous correlation matrix with hetcor(). Is there a way to do this in R? I couldn't find any guidance in the r-help archives or in the polycor help files. Should I truncate the sampling weights to integers and then populate the data set with redundant rows/cases so that the number of rows equals the population size (N>700,000)? Or is there a better way to compute a weighted heterogeneous correlation matrix? Thanks, Chris -- Christopher T. Moore, M.P.P. Doctoral Student Quantitative Methods in Education University of Minnesota 44.9785?N, 93.2396?W moor0554 at umn.edu http://umn.edu/~moor0554
Dear Chris, As you surmised, the functions in the polycor package make no provision for survey weights. You could get polychoric correlations by using weighted contingency tables as input, but that approach won't work for polyserial correlations. Regards, John -------------------------------- John Fox Senator William McMaster Professor of Social Statistics Department of Sociology McMaster University Hamilton, Ontario, Canada web: socserv.mcmaster.ca/jfox> -----Original Message----- > From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] On > Behalf Of Christopher T. Moore > Sent: February-25-10 6:00 PM > To: r-help at r-project.org > Subject: [R] Heterogeneous Correlation Matrix with Survey Weights > > Hello, > > I have a data set containing categorical and ordinal factors, as well as > sampling weights (i.e., survey weights reflecting unequal probabilities of > selection). I want to fit a structural equation model with sem(). I have > run sem() on weighted covariance matrices using advice from John Fox (see > <http://tolstoy.newcastle.edu.au/R/e5/help/08/12/8773.html> and > <http://blog.lib.umn.edu/moor0554/canoemoore/2009/09/sem_complex_samples_r_up > date.html>). > However, since I have categorical/ordinal variables, I would like to > compute a weighted heterogeneous correlation matrix with hetcor(). > > Is there a way to do this in R? I couldn't find any guidance in the r-help > archives or in the polycor help files. Should I truncate the sampling > weights to integers and then populate the data set with redundant > rows/cases so that the number of rows equals the population size > (N>700,000)? Or is there a better way to compute a weighted heterogeneous > correlation matrix? > > Thanks, > Chris > > -- > Christopher T. Moore, M.P.P. > Doctoral Student > Quantitative Methods in Education > University of Minnesota > 44.9785?N, 93.2396?W > moor0554 at umn.edu > http://umn.edu/~moor0554 > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code.