I'm just wondering what is the difference between empirical = TRUE or FALSE when simulating normally distributed multivariate. Is it that when empirical = TRUE every simulated data will indicate the identical preset covariance matrix?

  • 2
    $\begingroup$ You could generate a tiny dataset and check in less time that you will have to wait for an answer... $\endgroup$
    – whuber
    Jun 2 at 15:13
  • 1
    $\begingroup$ Hi Whuber, thanks for your suggestion! I just forgot this very simple option ... I tried both argument settings. When empirical = TRUE, the generated dataset will exactly replicate the specified covariance matrix. When empirical = FALSE, the generated dataset would contain more or less sampling error as noted by Billy. But many thanks for all of the answers! $\endgroup$
    – Bo WANG
    Jun 3 at 15:20

1 Answer 1


"If true, mu and Sigma specify the empirical not population mean and covariance matrix." (from mvrnorm documentation)

Meaning that if TRUE, the covariance matrix would be as specified

  • $\begingroup$ A covariance must always be specified (this is the Sigma argument). The documentation just specifies that when empirical = TRUE then the provided mean and covariance matrix are sample-based estimates (i.e., derived empirically) whereas when empirical = FALSE the mean and covariance matrix are treated as known/true population parameters estimated without sampling error $\endgroup$
    – Billy
    Jun 2 at 22:39
  • $\begingroup$ Would you want to make that an answer and I can delete mine copied from the documentation? $\endgroup$
    – pep
    Jun 3 at 1:14
  • $\begingroup$ These answers are all valuable! I would rather retain all of them for those who may have similar questions. Many thanks to both of you! $\endgroup$
    – Bo WANG
    Jun 3 at 15:24

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.