0

I've got a problem chossing the right model. I have a model with various variables (covariables and dummyvariables). I was trying to find the best seize of the model, so I first started in comapring the models with AIC. From this it followed, that the minimum AIC was reached when allowing all Variables to stay in the model (with the whole bunch to interact with all dumies). When I compute the summary of the model, all effects are absolutely not significant and its std. errors are very high. I was a bit confused, when comparing the "best" (on AIC) model with a smaller model with any Interaction. The smaller model had small standard errors and nice p-values... But the AIC is higher compared to the big model. What might be the problem? Overspecification?

I really need help in this, because I have absolutely no idea how to handle this!

Thanks alot

merv
  • 67,214
  • 13
  • 180
  • 245
user734124
  • 489
  • 8
  • 20

1 Answers1

0

I would recommend you also computing AICc and compare the results with AIC. This might be helpful : http://en.wikipedia.org/wiki/Akaike_information_criterion

Sincerely

Tu.2
  • 233
  • 1
  • 3
  • 5
  • Yes I did that before! But the model with the best minimal AIC has nasty signifikance within the p-values. – user734124 May 11 '11 at 07:37
  • Hi, if I were you I will try (1) use the stepwise variable selection to select the significant variables from the full model, then see what AIC value will be and (2) Use the "deviance" to compare the two model instead of AIC, AICc or BIC. If you use R, glm function can compute deviance score for you. – Tu.2 May 11 '11 at 18:20