Seeking Guidance for Group-Based Trajectory Analysis with Imputed Datasets

[YYK273]

Hi SAS Community,

I’m currently working on a project that involves 20 imputed datasets, and I need to perform group-based trajectory modeling using PROC TRAJ (MODEL zip) . I’ve found limited reference publications on using PROC TRAJ with multiple imputations and would greatly appreciate any advice or guidelines.

Here are my specific questions:

1. Should I analyze each imputed dataset separately using PROC TRAJ? 
2. Once the trajectories are estimated for each dataset, what is the best way to combine the results? Are there specific examples for pooling results?
3. Are there any best practices, resources, or examples you can point me to for guidance on this?

Thank you so much in advance for your help! Looking forward to hearing your insights.

[Ksharp]

If PROC TRAJ support BY statement, you could use it by 

proc traj data=have;

by _imputed_;

.....
run;

Or you can split this dataset into many sub-dataset and run it under proc traj:

proc traj data=have(where=(_imputed_=1));

.....

run;

proc traj data=have(where=(_imputed_=2));

.....

run;

1. Should I analyze each imputed dataset separately using PROC TRAJ? 

Yes. You should as I showd above.

1. Once the trajectories are estimated for each dataset, what is the best way to combine the results? Are there specific examples for pooling results?

Once you could get StdErr of Parameters , you could get POOLED parameters.

proc mianalyze data=Binomial;
  modeleffects _BIN_;
  stderr E_BIN;
run;

1. Are there any best practices, resources, or examples you can point me to for guidance on this?

Check UCLA website:

https://stats.oarc.ucla.edu/sas/seminars/multiple-imputation-in-sas/mi_new_1/