sem with categorical data
Dear Frank, Did you take a look at the data you generated? The contingency tables are extremely sparse. Try, e.g., with(dat, table(q1, q2)). (I assume, BTW, that you're using the cut2() function in the Hmisc package.) Beyond that, I'm not sure what the point is of your example. You're generating three independent lognormal random variables while the model assumed by polychoric correlation is of multinormal random variables; of course, these *could* be independent, but generally would not. I hope this helps, John On Wed, 20 May 2009 12:55:49 -0400
"Frank Lawrence" <frl2 at psu.edu> wrote:
I am trying to run a confirmatory factor analysis using the SEM package. My data are ordinal. I have read http://socserv.mcmaster.ca/jfox/Misc/sem/SEM-paper.pdf. When I apply the hetcor function, I receive the following error: Error in checkmvArgs(lower = lower, upper = upper, mean = mean, corr = corr, : at least one element of 'lower' is larger than 'upper' Example: set.seed(781) q1 <- cut2(x = rlnorm(n = 1e2, m = 2), cuts = seq(1,6)) q2 <- cut2(x = rlnorm(n = 1e2, m = 0), cuts = seq(1,6)) q3 <- cut2(x = rlnorm(n = 1e2, m = 1), cuts = seq(1,6)) dat <- data.frame(q1,q2,q3) hetcor(data = dat, ML = T, pd = T, use = 'p') Any help in understanding and correcting the error would be appreciated. unlist(R.Version()) platform arch "i386-pc-mingw32" "i386" os system "mingw32" "i386, mingw32" status major "" "2" minor year "9.0" "2009" month day "04" "17" svn rev language "48333" "R" version.string "R version 2.9.0 (2009-04-17)" Respectfully, Frank Lawrence
-------------------------------- John Fox, Professor Department of Sociology McMaster University Hamilton, Ontario, Canada http://socserv.mcmaster.ca/jfox/