Dear all,
I am running a logistic regression and wanted to use the boxtid command to check whether my variables are correctly specified. However, for one variable, the population size of a city, the boxtid results keep indicating that the relationship of the logit of that variable and the dependent variable (does a given city apply a policy innovation) is not linear. If I try just the variable population it suggest a P1 of -0,4. So, I tried the same model with the natural log of population and another one with the log of the basis 2 (as the variable is heavily skewed to the right). Also for these, the boxtid test tells me that my logits are not linear.
However, the lowess graph suggest that the log of the basis2 of population shows the required s-shape function. Also the odd ratios seem fine and are significant and the linktest command suggest that the model is correctly specified.
Does anyone have an idea what could be the problem with my variable? Or does anybody no further specification tests I could run?
Thank you for your help!
I am running a logistic regression and wanted to use the boxtid command to check whether my variables are correctly specified. However, for one variable, the population size of a city, the boxtid results keep indicating that the relationship of the logit of that variable and the dependent variable (does a given city apply a policy innovation) is not linear. If I try just the variable population it suggest a P1 of -0,4. So, I tried the same model with the natural log of population and another one with the log of the basis 2 (as the variable is heavily skewed to the right). Also for these, the boxtid test tells me that my logits are not linear.
However, the lowess graph suggest that the log of the basis2 of population shows the required s-shape function. Also the odd ratios seem fine and are significant and the linktest command suggest that the model is correctly specified.
Does anyone have an idea what could be the problem with my variable? Or does anybody no further specification tests I could run?
Thank you for your help!
