calculating Lambda confidence intervals

[danikaschube]

I am working on a Pradlambda analysis on data set that is too large to run the model averaging functions in Mark or Rmark. So,I have pulled the beta parameters from models I wish to average into another program and I can replicate the calculations and results for the model averaged parameters for Phi and p, their associated SEs and 95% CIs. I can also replicate the calculations and results for Lambda and its SE. However, I can't calculate the CIs correctly.

I know that the CIs are not bounded by 0 and 1 like Phi and p so the confidence interval calculations used for Phi and p are not appropriate. However, the 95% CIs are also not simply Lambda+-1.96*SE. Can someone point me in the direction of the formula I should be using?

Thanks.

[cooch]

I am working on a Pradlambda analysis on data set that is too large to run the model averaging functions in Mark or Rmark. So,I have pulled the beta parameters from models I wish to average into another program and I can replicate the calculations and results for the model averaged parameters for Phi and p, their associated SEs and 95% CIs. I can also replicate the calculations and results for Lambda and its SE. However, I can't calculate the CIs correctly.

I know that the CIs are not bounded by 0 and 1 like Phi and p so the confidence interval calculations used for Phi and p are not appropriate. However, the 95% CIs are also not simply Lambda+-1.96*SE. Can someone point me in the direction of the formula I should be using?

Thanks.

Try the non-interactive model averaging approach. Its 'in the book'. This is often useful for large problems which blow up the GUI-based front end to averaging.