1
$\begingroup$

My independent variable is interest rate, which is small positive variable(less than 1). When I log transform the values, they turn negative because the values are smaller than 1. What is the solution? Can I add 1 to the log equation in stata , gen loginterest = log(interest+1)? Should the constant be smaller than 1? My interest rate data ranges from 5% to 21%. Please advise.

I have tried running linear regression without transforming the values, but my errors are not normally distributed.

Additional question, if I run a regression with ONE main independent variable and several control variables, is it still called multiple linear regression? Or is such term used only if we have several independent variables? Thank you

$\endgroup$
5
  • 2
    $\begingroup$ Why are negative values an issue to you? $\endgroup$ Commented Oct 4, 2022 at 7:38
  • $\begingroup$ Logically speaking, the institutional data I am using would not ever have negative interest rate. So, it did not make sense to me, and thought it would mess with the relationship I am trying to determine with my independent variable (positive, non log value). $\endgroup$
    – Laiy
    Commented Oct 4, 2022 at 7:44
  • 4
    $\begingroup$ But a negative log interest rate corresponds to a positive interest rate. So there is no problem there. $\endgroup$ Commented Oct 4, 2022 at 7:56
  • $\begingroup$ @StephanKolassa thank you! So, does this change how I interpret the coefficient? If the my independent variable (also in percentage, but non log) coefficient is say, -0.15. Is it correct to interpret it as a unit change in X will decrease the interest rate by 0.15%? $\endgroup$
    – Laiy
    Commented Oct 4, 2022 at 8:00
  • 2
    $\begingroup$ You can model it using log transformed values, and after you convert the results back to the original scale for interpretation. $\endgroup$ Commented Oct 4, 2022 at 9:31

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.