Dear friends,
If we can not control the levels of a factor which is usually defined a priori such as "time" (e.g. time 1, 2 ,3 and 4), should we consider it a random factor?
Example: We've got an intervention which the number of levels of the factor "time" depends upon the subject's skills. An experienced subject may reach 12 levels while an unexperienced likely stops at the level number four. In such case, "time"should be consider random factor? And how to deal with the missing values?
Best regards,
Paulo
Hi
If you want to establish relationship between skill and time taken to perform a job and want to estimate parameters associated to time, you may consider taking a distribution of the parameter associated to time. In this way the parameter will not be fixed. It will have some distribution and the model can account for the heterogeneity.
Regards.
- Badges
- Science topic
- Similar topics
- Political Science
- International History and Politics
- Globalization