# What is best for comparing variability for environmental time series data: repeated-measures ANOVA or linear mixed models?

I've collected environmental data on the dust level using direct-reading instruments. The outputs from the instruments are concentrations recorded at each minute over 1 hour. The data collected at many locations and locations are grouped into Groups and Department. The data looks like this in wide format

Depart1  Group1  Location1:  3, 3, 4, 3.5, 4, 6, etc...
Depart1  Group1  Location2:  2, 7,5, 6, 3, 5, 5, etc..
Depart2  Group2  Location3:  8, 4, 3, 6, 6, 3, 6, etc..


I want to find the within and between variability among the Groups. But for some groups, I have data at 3 locations per group and sometimes I only have data at 1 location per group. Thus, I have uneven datasets for Groups.

I am wondering what is the best statistical model to answer the variability question. I've looked up ANOVA for repeated measurements but how do you deal with the uneven grouping?

Would a linear mixed model with time being considered as a fixed effect and Group and Location being random effects work here? At the end, I am only interested in within and between variability for the Group not the Location.

-
That should work. –  Nikhil Bellarykar Jan 3 '12 at 6:45
what should work? –  Tran Jan 4 '12 at 0:53
what you have said i.e. linear model with time as fixed effect. –  Nikhil Bellarykar Jan 4 '12 at 5:10