My response variable is a combining of two variables which I would like to rescale with different weights. ratio:0.5/0.5; 0.25/0.75; 0.1/0.9. Now I would like to test which is the best fit. The three models have overdispersion so I used quasipoisson regression.

Following former instructions in this matter, I used the F test, but it didn't work because of the different response variable. I got this massage:

    anova(glm0.5_0.5,glm0.75_0.25,glm0.9_0.1, test = "F")        

Model: quasipoisson, link: log

Response: data$dep0.5_0.5

Terms added sequentially (first to last)

             Df Deviance Resid. Df Resid. Dev       F    Pr(>F)    
NULL                           575    11386.0                      
academic95    1   3720.1       574     7666.0 231.505 < 2.2e-16 ***
israeli_ac95  1   1239.1       573     6426.8  77.113 < 2.2e-16 ***
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Warning message:
In anova.glmlist(c(list(object), dotargs), dispersion = dispersion,  :
  models with response ‘c("data$dep0.75_0.25", "data$dep0.9_0.1")’ removed because response differs from model 1

what should I do?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.