3

I have following data

Model 1: Y ∼ B1 + B2 + B3 + B4: log L = −213.4
Model 2: Y ∼ B1 + B2 + B3 + B5: log L = −567.1

The question that i have is "Calculate AIC for these 2 models". It doesn't say much. I know the formula for AIC is

$$ AIC = -2log(L) + 2p $$

Based on this, is the solution for Model 1 is literally -2 * -213.4 = 2 * 4 ? I don't think that it would be that simple. Also, how would i know which one to pick based on the answer.

Em Ae
  • 351

1 Answers1

2

Assuming one of your B's is the intercept, then your p is correct. The p captures the number of parameters in the model, including the intercept. It looks like you have a typo in your formula though. The = should be +.

To choose between the models, compare the AIC scores. Choose the model with the lower AIC score.

Anavir
  • 97
  • 4
  • Error variance is another parameter that is being estimated. I think it should be included in the parameter count. For these two models, ithat would not change which AIC is lower, though. – Richard Hardy Nov 29 '20 at 08:47