Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: repeated (sic) longitudinal analysis


From   Hubert Roth <[email protected]>
To   [email protected]
Subject   st: repeated (sic) longitudinal analysis
Date   Wed, 15 Sep 2010 18:18:28 +0200

Thanks to Dave, I quit xtgee and had a try with xtmixed.
The problem was : 
14 subjects come to have a meal at noon where food is "ad libitum" and quantity swallowed (energy, glucids, lipids, proteins) is assessed every minute. Duration is variable according to the subject and the meal, from 15 to 31 min
They do this 3 times : once fasted since the day before (meal=0) and twice non fasted (meal = 1 or 2)
Initial energy (kcal) swallowed at t=0 is 0, so I chose "noconstant"
For me, meal (0, 1 or 2) is a fixed effect, time too and subject is a random one, so I chose the following model

xi: xtmixed kcal i.meal minute, noconstant || num_sujet: minute, cov(unstructured)

Am I right ?
Do I have to add a second random effect : "|| num_sujet: i.num_repas, cov(unstructured)" ?

Thanks for your help. Best regards,
Hubert



> I think -xtgee- only allows one level of repeated measures data, not more. If you have multilevel data, you need to use mixed models, such as -xtmixed-.
> 
> -Dave
> 
> > Dear all,
> > I am analysing the following study :
> > - healthy subjects are eating a meal ad libitum and the quantity of food (number of forks for example) is recorded every minute : energy, glucids, lipids and proteins
> > - the meal is repeated 3 times in 2-4 weeks : once fasted (meal 0) and twice not fasted (meals 1 & 2).
> > Usually I use xtgee to do longitudinal analyses when subjects are different, something like :
> > xtset id_subject minute
> > xi: xtgee kcal i.meal, family(gaussian) link(identity) corr(exchangeable)
> > 
> > But this time the same subject eats the three meals and I want to know if the three curves are different in their shape (besides the AUC, slopes...).
> > Do you have any idea how to include in the GEE model the fact that it is the same subject who is re-recorded to take into account this variability ?
> > It sounds to me like clusters but I didn't find how to put it in the GEE model. I had also a look at gllamm but I am not familiar with it.
> > Thanks for your help. Best regards,
> > Hubert
> 
> 
-------------------------------------------------------------------
Hubert ROTH, engineer, biostatistician
Centre de Recherche en Nutrition Humaine Rhone-Alpes
Rhone-Alpes Research Center for Human Nutrition
site: www.crnh-rhone-alpes.fr
BP217 - F38043 Grenoble cedex 9 - France




*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index