more on truncating to avoid violations of assumptions

questions concerning analysis/theory using software from the Patuxent Software Archive

more on truncating to avoid violations of assumptions

Postby shannonbarbermeyer » Thu Jun 25, 2009 3:40 pm

I should add to my post that if it isn't advisable to truncate to remove a source of model violation (for the obvious reasons like since our capture histories are pooled and such that the day 1s are different calendar days and because I don't know the biological or logistical reason why it would make sense to truncate) then can I stick with Mh and acknowledge the violations and indicate that when data were truncated to remove the violations the estimates were basically the same showing the robustness of that estimator to that level of violation? Or is it better to use the Mh results (even though they don't differ much) from the non-violation causing truncated data set? For example say the GOF for Mh fails but no other model is close in selection. Or in another datset case say time violation is an issue but not when taking off the last pooled day. Basically if you can show the truncated dataset with the violation corrected still gives a similar answer should you use the exact results from the full dataset even though a violation may have been present (if the researcher can't think of a fair reason why truncating is legit) or should you use the truncated case? I'm nervous about data manipulation to get things to 'work' and think it may be safer to stick with the full dataset and explain what other permutations showed in support of the model selected.

Thank you for your help! Shannon
Posts: 24
Joined: Tue Aug 05, 2008 1:32 pm
Location: Wailuku, HI

Return to analysis help

Who is online

Users browsing this forum: No registered users and 1 guest