The pig sign data (the dependent variable) is repeatedly measured over 4 time periods at 17 separate sites. The habitat variables (the explanatory variables) are time-invariant, but have been measured for each of the 17 sites. I've appropriately reorganized my data to the person-period format. I'm using RStudio to run glmm and lmm analyses, and have included a random intercept of time nested within site and time, respectively. I see that the software will only run the pig sign data from the rows that correspond with the habitat data, which makes the other pig sign data irrelevant (cutting my sample size by 75%, and making my time factor irrelevant). Is it statistically sound to augment my data by repeating the same habitat data for all time periods and run that against the time-varying pig sign data?
I don't see any other way to deal with this in the literature and I sure could use a little assistance with this stats issue.
Philip D DeWitt I tried to respond to this last week and it looks like it might not have worked properly. Thank you so much for answering my question and for your recommendations! I really appreciate it! I will definitely be careful about which habitat variables I define as time-invariant.
I did try time as a fixed effect but there is so much variance in the pig sign and/or so little change over time that no significant time effect can be detected.
Thanks again! Your advice has been of great help to me!
It's fine to do that because there are no rules in statistics
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running