Poisson regression

[Arina Pershina]

It's my first time using the poisson regression model, I was wondering if I interpreted the coefficients correctly:

The coefficients of the Poisson regression model represent the log level, in order to obtain the coefficients for direct interpretation this paper performs a transformation (eCoef.- 1). This model predicted the positive relationship between respondent’s age and the digitalisation level of their company, the one unit increase in the age variable is associated with the 0.098 unit increase in the digitalisation variable.

[Maxence Morlet]

In Poisson, coefficients on continuous regressors are directly interpretable as semi-elasticities, and do not need to be transformed.

[Arina Pershina]

Thanks a lot for help! Is it correct that the interpretation can be like this: the one percentage point increase in the age variable is associated with the 7.77% increase in the digitalisation variable? Also, what is the correct way to interpret it: as change in X in units or percentage points? And, is it the same for binary regressors?

[Maxence Morlet]

Semi elasticity: unit change in X --> % change in the dependent variable

Binary X: X "turning on" leads to a ..% rise in the dependent variable. However, you have to convert the coefficient on binary regressors: 100*(exp(Beta)-1) if my memory serves.

In your case, you specified age as quadratic, so this complicates things a bit.