Dear all,
I have a question of correlation and my coefficient.

I use pwcorr to test the correlation of my variables, whereas the sign of the correlation shows different with my regression. I know it may be a statistical issue. But before moving there, I am wondering can I know whether it might be influenced due to I use reghafe, absorb(sic_2 fiscal_year analys) vce(cluster analys) in my regression? I use the same control variables with previous literature, so it should not be an multicollinearity issue.

For example, the correlation between forecast error and high_tech is negative and significant, whereas the coefficient in the regression is positive and significant.

Thank you so much.