[Limdep Nlogit List] Interpretation of Mixed logit model

[Mikołaj Czajkowski]

Dear Fan,

in mixed logit a parameter is described using a parametric probability 
distribution. This distribution could have a mean that is close to zero 
(not significantly different) and some non-zero standard deviation 
(significantly different from 0). This is a fine result meaning that on 
average (mean) the effect is close to 0, but there is some heterogeneity 
around this. On the other hand, if standard deviation is not 
significantly different from 0 - you might as well model such a 
parameter as fixed, as heterogeneity around the mean is not significant.

I cannot say if it makes sense for your variables / model, but 
technically such a result is fine.

Best,
Mik



On 2017-08-17 10:49, FAN Zhou via Limdep wrote:
> Hi all,
>
>
> I have run a Mixed logit model with Nlogit 6. But I am confused by the results.
>
>
> I was wondering how to interpret the insignificant mean but significant standard distribution for a random parameter. For example, if I set "cost" as random parameter, the mean 'B_COST' is insignificant but its associated standard deviation "NsCOST" is significant. Does results like this make sense?
>
>
> I appreciate any help.
>
>
> Cheers,
>
> Fan
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> http://limdep.itls.usyd.edu.au
>