Correlation Matrix

[Fawad Latif]

Hi

How do i get Correlation Matrix for Latent Variables in Lavaan. Or i shall use

Std.lv in the Covariances as Correlation between latent variables. 

Thanks

[Mauricio Garnier-Villarreal]

the std.all column are the factor correlations

If you want to extract the factor correlations matrix you can with

lavinspect(fit, "cor.lv")

[Harmanjit Singh]

Hi

Is there a way to get p-values in the matrix output along with correlation after using lavInspect(fit, "cor.lv")?

[balal izanloo]

Greetings

I think this link is appropriate for what you want. 

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/8e724d31-9717-485f-a3c5-f5188e0b9189n%40googlegroups.com.

[Mauricio Garnier-Villarreal]

not with the lavinspect, but you can subselet the needed information from the parameterestimates output. Like this

HS.model <- ' visual  =~ x1 + x2 + x3
              textual =~ x4 + x5 + x6
              speed   =~ x7 + x8 + x9 '

fit <- cfa(HS.model, data = HolzingerSwineford1939, std.lv=T)
summary(fit, fit.measures = TRUE)


lavInspect(fit, "cor.lv")
pe <- parameterestimates(fit, standardized = T)

subset(pe, lhs != rhs & op == "~~")[,c("lhs","op","rhs","std.all","pvalue")]

[Harmanjit Singh]

Thanks Mauricio!
This was easy!

You received this message because you are subscribed to a topic in the Google Groups "lavaan" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/lavaan/B2aCdAQJhWg/unsubscribe.
To unsubscribe from this group and all its topics, send an email to lavaan+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/d469b806-5fcf-4052-9dff-3f77f009b49bn%40googlegroups.com.

[Mauricio Garnier-Villarreal]

just one clarification. The p-value is evaluating the unstandardized estimates, so its not the p-value of the correlaton but the pvalue of the covariances. Which is what should be, have to test the unstandardized parameter, and use the standardized as measure of effect size

[Harmanjit Singh]

Thanks for the clarification!

To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/22159e6c-904e-420a-9de3-bec1fdaf6c41n%40googlegroups.com.

[Harmanjit Singh]

In that case, are these the same p-values that we have to report for the correlation matrix, since correlations pertain to std.all column but as you mentioned these p-values pertain to unstandardized estimates. Just got confused over this.

[Pat Malone]

The covariance is the estimated statistic, and the p-values are for it (specifically, the test of the covariance against the null that it is zero). The correlation is a number derived from three estimated statistics (the covariance and the two variances), so, if you're able to test the covariance--as in SEM--a p for the correlation would have no real added value. (We learn in first-year stats how to generate a p-value via a t-test of a correlation coefficient, but that's usually from a much simpler "model," and it takes into account the sample size in a more direct way.)

Correlations are just an easily interpretable effect size, as Mauricio said. I don't personally know many people who can draw meaning from glancing at a variance/covariance matrix.

Pat

To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/5c6faf69-41a3-4b15-bbc9-2b69ad5a90e4n%40googlegroups.com.

--

Patrick S. Malone, PhD

Sr Research Statistician, FAR HARBΦR

+1 803.553.4181  |  pat@  |  farharbor.com

This message may contain confidential information; if you are not the intended recipient please notify the sender and delete the message.

[Mauricio Garnier-Villarreal]

Seconding Pat, in short, we report the p-value and test the unstandardized parameter, but can use the correlation for easier interpretation.