Traj command (dropout model)

[Janelle Gravesande]

Hello everyone,

I am using the traj command to examined gait speed trajectories and I have a question about the dropout model option. When I used this option, some of my models had positive BICs and some had negative BICs. However, when I removed that option, all BICs were positive. What is the benefit of using this option? Can I omit this option?


Here’s my code and BIC values:

2 groups (with dropout model)

traj, var(walkspeed_*) indep(time_*) model(cnorm) order(0 0) min(0.07533158) max(1.9544182) dropout(0 0)

BIC= -873.03 (N=9525) BIC= -869.39 (N=2836) AIC= -851.54 ll= -845.54

2 groups (without dropout model)

traj, var(walkspeed_*) indep(time_*) model(cnorm) order(0 0) min(0.07533158) max(1.9544182)

BIC= 2004.89 (N=9525) BIC= 2007.31 (N=2836) AIC= 2019.21 ll= 2023.21

Thanks in advance.

[Andrew Lover]

It appears to expand the models to assess both the probability of study dropout, and of group membership.




Ref here:

http://10.1146/annurev.clinpsy.121208.131413

[Janelle Gravesande]

Thank you very much. I will read the reference you provided.