1

I am running a quasi-poisson regression predicting a count outcome from 4 variables of interest.

enter image description here

I understand that these estimates need to be exponentiated to correctly interpret the results. I used the tab_model() command in sjPlot to gather these and noted that the p-value and associated 95% CI (displayed below) are different from the p-value provided in the glm() output (above).

enter image description here

Could someone please provide any clarification on the discrepancy between these two? Is it typical practice to report the IRR and 95% CI (and possibly whatever this associated p-value is in the tab_model() command)?

User1865345
  • 8,202
stilesb
  • 31
  • 2
  • After doing a little bit of digging around here, it looks like the p-values from the glm() output is derived using Wald tests (https://stats.stackexchange.com/questions/237073/how-does-r-calculate-the-p-value-for-this-binomial-regression). I read that Wald tests may be "valid" for "large samples" but less so in smaller ones, which is arguably not the case in mine (N = 21). Is it reasonable then to report the exponentiated coefficients/CIs and the associated p-values generated via tab_model()? – stilesb Jan 16 '23 at 16:31

0 Answers0