dear. How are you. There is long since I have been trying to solve a doubt about diference between nested and crossed random effects within my study. So I am here to ask and seek for some information. I analysed in a study whether built-up areas density and open waters features quantity variation influence on breeding spot’s density of A. aegypti in municipality of Campo Grande urban area. I have applied Generalized Linear Mixed Models to test this effect from explanatory variables (NDBI and MNDWI) on features density of A. aegypti. At the same time I have sorted neighborhoods and months/year as the random factors, and in this context we can always find the same set of Campo Grande's neighborhoods from where data were collected along different months and years (although inside a few years, or more often, a few months are found repetead). So, according to this experimental design, I need to know, how can I sort these randon effects?
0
Add a comment |
1 Answer
$\begingroup$ $\endgroup$
It sounds as though neighbourhoods and years are crossed, but months are nested in years.
That is, it sounds as though:
- neighbourhood 1 in year 1 is the same neighbourhood as neighbourhood 1 in year 2 (has the same neighbourhood random effect: crossed)
- year 1 in neighbourhood 1 is the same year as year 1 in neighbourhood 2 (has the same year random effect: crossed)
- January in year 1 is not the same month as January in year 2 (has a different month random effect: nested)
If you expect the month effects to be the same from year to year you might regard month as crossed with year -- but in that setting you also might regard month as a fixed effect and you might want to model seasonal variation more explicitly
