Correlation matrix significance test

[hansie]

Hello,

I use panel data to test for fixed effects. I also need a correlationmatrix. I know how to generate such a matrix (Quick -> Group Statistics -> Correlations). However, I would like to know how I can calculate in Eviews 7.2 if these correlation significantly differ from zero. Does anybody know how to deal with this?

Thank in advance.

Hans

[EViews Glenn]

You can produce individual tests for correlation if you use the covariance analysis tools (View/Covariance Analysis...) on a group. We don't produce sphericality tests there, but if you let us know which ones you want, we can probably figure out how to do that computation for you.

Incidentally, you said that you were using panel data. How are your data structured? I'm asking because for stacked data, the Quick method of computing the correlations may not be producing what you want.

Lastly, if you have EViews 8, you can use the CDTest add-in to compute the CD test. That might work for you directly, or could be modified to accommodate your particular application.

[hansie]

Dear Glenn,

Thanx for your response. I tried the CDTest add-in. However, I doesn't work because I use the student version of Eviews 8. I have a regular version of Eviews 7.2 so I guess I should try the covariance analysis tools.

After I set-up the correlation matrix, I opened Views -> Covariance Analysis. I found the desired test.

Thanks