Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

Deciding if a covariate is useful or not


I am analyzing CBC HB data and I am trying a combination of 3 covariates to improve the model. for each of the 8 models ( 1 without covariates & 7 of the different combinations) I have calculated the aggregate RLH and percentage of levels that are 90% of the times positive or negative during the used iterations. the results are as follows and I am not sure if (Comorb+Calm) would be the best choice or not. Note that the S.E. increases as the number of covariate levels increase. Am I missing more information to decide?.

                             RLH*100    covariate levels    Max. # affected levels/covariate levels (affected intercepts)
No covariates    72.99%      Ref.                            Ref.
Comorbid            73.6%             1                                    4/1 (-ve & +ve equal) (13)
Importance    73.7%             1                                    4/1 (-ve & +ve equal) (15)
Calm death    76.3%             4                                    26/4 (-ve) (9)
Comorb+Impt    74.24%             2                                    9/2 (-ve) (13)
Comorb+Calm    79.3%             5                                    31/5 (-ve) (8)
Import+Calm    78.8%             5                                    30/5 (+ve) (8)
All 3                    80.2%             6                                    25/6 (-ve) (5)
asked May 22 by AMYN Bronze (2,980 points)

Your solution to the original question

Please only use this to answer the original question. Otherwise please use comments.
Your name to display (optional):
Privacy: Your email address will only be used for sending these notifications.
Anti-spam verification:

To avoid this verification in future, please log in or register.