In Hmisc package in R or with cor.test, we can get the pvalues for pearson or any other correlation. I wanted to calculate biserial and polyserial correlation for data with numeric and ordinal values.

I am using functions biserial.cor from ltm package and polyserial from polycor package. I am wondering how can I calculate, pvalues to test if the correlations are statistically significant in case of biserial and polyserial correlations.

Example from the help function:

biserial.cor(rowSums(LSAT), LSAT[[1]])
  • $\begingroup$ You can use the standard error from polyserial. $\endgroup$ – mdewey May 11 '16 at 12:42
  • $\begingroup$ Could you please show me an example? Thanks! How do I calculate std.err (preferably pvalues) from values returned polyserial function when std.error is set to TRUE? $\endgroup$ – discipulus May 11 '16 at 13:00
  • 1
    $\begingroup$ ?polyserial tells you what it returns and you can then extract the standard error. You get a p-value by dividing the estimate by its standard error. Note you are assuming the estimate is normally distributed which for absolute values close to unity is almost certainly false. $\endgroup$ – mdewey May 11 '16 at 13:19
  • $\begingroup$ Sorry for my stupidity but I get ... variance (which is a matrix), chisquare, degrees of freedom, in addition to type, rho and cuts, what would be the formula for standard error. Thanks! $\endgroup$ – discipulus May 11 '16 at 16:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.