Scaled independent variables

[Anuradha Saikia]

Hello everyone

I have an unbalanced panel and I am running a panel cox regression survival model. My independent variable for the study is not firm-specific but rather state-specific i.e. data corresponding to the subsidiary belonging to a particular state in India. Can I scale it by the total assets of the firm as that scaled variable is showing significance and if not scaled it is not significant even though the hazard ratio is more than 1 .
Please help.

[Eric Makela]

Sounds like the scaled vs. unscaled results would be a subsection of whatever you are writing. Do you disagree or is this distinction to be avoided in your hypothesis?

[Anuradha Saikia]

Hello Eric Makela . I am just trying to get significant results. I thought scaling would be a good idea, and the variable result is significant. But for scaling, a province or state variable cannot be scaled by a firm variable I think.

[Eric Makela]

Either sounds correct to me, as long as you are interpreting the coefficient correctly. Not sure what those results would mean... the coefficient on a state-level IV transformed to also reflect the state's share of assets of a particular firm or industry.

[Anuradha Saikia]

Thank you Eric Makela