Hello, I have conducted an SEM in which the resultant standardized path coefficients are much higher than would be expected from the raw correlation matrix. To explore further, I stripped the model down to a simple bivariate relationship between two variables (NDVI, and species richness), where it's my understanding that the SEM's standardized path coefficient should equal the correlation coefficient.
The datafile can be downloaded at http://labs.bio.unc.edu/Hurlbert/pubs/NDVI_lep_data.csv The model specification file can be downloaded at http://labs.bio.unc.edu/Hurlbert/pubs/SEM_NDVI_AllS_model.txt and is simply: NDVI -> All.S, n2S, NA All.S -> All.S, S2S, NA NDVI <-> NDVI, n2n, 1 df = read.csv('NDVI_lep_data.csv',header=T) cor(df[,c('NDVI','All.S')]) yields: NDVI All.S NDVI 1.0000000 0.4156191 All.S 0.4156191 1.0000000 But, conducting an SEM yields: sem.mod = specifyModel('SEM_NDVI_AllS_model.txt') sem.mod.cov = rawMoments(~ NDVI + All.S, data = df) sem.mod.cov = sem.mod.cov[-1,-1] sem.mod.cov NDVI All.S NDVI 0.7820657 13.53573 All.S 13.5357259 245.71360 sem1 = sem(sem.mod, sem.mod.cov, N=29) stdCoef(sem1) n2S n2S 0.97643950 All.S <--- NDVI n2n n2n 1.00000000 NDVI <--> NDVI s2S s2S 0.04656591 All.S <--> All.S I am using version 3.0 of sem on R-14.0. Thanks for any insight anyone can provide. Allen Hurlbert, Ph.D. Department of Biology University of North Carolina Chapel Hill, NC 27599-3280 Tel: 919.843.9930 Wilson Hall 331 [[alternative HTML version deleted]] ______________________________________________ R-help@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.