[R-sig-ME] including the same variable both as fixed and as part of a complex random variable?

[ONKELINX, Thierry]

Dear Hans,

I am assuming that labour.market.position (LMP) is continuous.

fit.1 and fit.2 are different. In fit 2, the fixed effect of LMP is the slope in an 'average' country. the random effect of LMP in then the difference in slope between a country and the 'average' country. In fit 1 you have no slope for an 'average' country (the slope = 0). So the random effect of LMP in fit 1 is the slope for each country.

in fit.1 and fit.2 you assume that there is no intercept, hence only a random slope along LMP per country.
in fit.3 you assume both a random slope along LMP and a random intercept AND you assume that the random slope and random intercept are independent.

Best regards,

Thierry

> -----Oorspronkelijk bericht-----
> Van: r-sig-mixed-models-bounces at r-project.org [mailto:r-sig-mixed-models-
> bounces at r-project.org] Namens Hans Ekbrand
> Verzonden: maandag 3 oktober 2011 16:40
> Aan: r-sig-mixed-models at r-project.org
> Onderwerp: [R-sig-ME] including the same variable both as fixed and as part of a
> complex random variable?
> 
> Dear list followers,
> 
> I am wondering wheter I should include some variables both as fixed and as
> parts of complex random terms, and I hope you can give some guidance.
> 
> Are the "real" differences between be models below? If so, what are the
> differences?
> 
> I have used the variables labour.market.position and country in the exampels
> below, but the question is of course a general one.
> 
> fit.1 <- glmer(poverty.third.year ~ 1 + (0 + labour.market.position | country) +
>                gender, family = binomial("logit"), data = poverty.risks)
> 
> fit.2 <- glmer(poverty.third.year ~ 1 + labour.market.position + (0 +
> labour.market.position | country) +
>                 gender, family = binomial("logit"), data = poverty.risks)
> 
> fit.3 <- glmer(poverty.third.year ~ 1 + labour.market.position + (1 | country) + (0
> + labour.market.position | country) +
>              gender, family = binomial("logit"), data = poverty.risks)
> 
> If there are no "real" differences between these, then which do you prefer, and
> why?
> 
> --
> Hans Ekbrand
> 
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models