In many epidemiologic longitudinal and rehabilitation studies, the outcome variable has floor or ceiling effects such as the Barthel Index and Stroke Impact Scale (SIS). Although not correct these variables are treated as normally distributed continuous and analyzed accordingly by using i.e. lineair mixed models.
Longitudinal Tobit analyses show better fit as compared to LMM thus being more suited for analyses of these outcome variables (J. Twisk, F.Rijmen, Journal of Clinical Epidemiology 62 (2009) 953-958).
|Who would benefit from this IDEA?||All researchers working with outcome variables that are either upper or lowered censored.|
How should it work?
In the article of Twisk a tobit model for longitudinal data is formulated and analysis is performed with Stata using the GLLAMM procedure. Results show a much better performance of the tobit model than LMM.
|Priority Justification||This new statistical technique offers a nice solution for longitudinal analysis of outcome variables with floor or ceiling effects.|
|Customer Name||Gerard Volker|
NOTICE TO EU RESIDENTS: per EU Data Protection Policy, if you wish to remove your personal information from the IBM ideas portal, please login to the ideas portal using your previously registered information then change your email to "email@example.com" and first name to "anonymous" and last name to "anonymous". This will ensure that IBM will not send any emails to you about all idea submissions