multicollinearity Problem

[rob86]

Hi Guys

I’m trying to reproduce two papers for my dissertation, the papers ran individual regressions for each of the independent variables, then combined the all the individually significant independent variables in a final regression. Both paper then checked for multicollinearity using a correlation of .7 a cut of point, thought there was no correlation that high. When I reproduced this I got a correlation of .8 between one pair of variables, so basically my question is whether I should leave both variables in, just drop one or is there anything else I could do? If I should just drop one is there any way to decide which one or is it just a matter of judgement?

Thanks