Hello R list,
I am hoping to conduct a logistic regression with repeated measures,
and would love an actual "code run through" for such an analysis. I
found only one related post on this list, but a full answer was never
provided. I understand that the routine lmer (or lmer2) in the lme4
package is often recommended in such a case, but actually implementing
it is where I've hit a wall.
In a nutshell, the experiment involved presenting females from two
groups (treatment, control) with an opportunity to mate with a virgin
male every 6 hours for 48 hours. Every female was presented this
opportunity at every time step (i.e., whether or not she mated at 6
hr, she was again presented with a male at 12 hr, and so on). In
addition to which group a female belongs to, we have an a priori
reason to want to test the effect of her initial body mass as a
covariate. A subset of the data looks like this:
female group mass time mate
1 control 5.7 0 1
1 control 5.7 6 1
1 control 5.7 12 0
1 control 5.7 18 0
1 control 5.7 24 0
1 control 5.7 30 1
1 control 5.7 36 0
1 control 5.7 42 1
1 control 5.7 48 0
2 treatm 5.3 0 1
2 treatm 5.3 6 0
2 treatm 5.3 12 0
2 treatm 5.3 18 0
2 treatm 5.3 24 0
2 treatm 5.3 30 1
2 treatm 5.3 36 0
2 treatm 5.3 42 0
2 treatm 5.3 48 0
3 control 6.1 0 1
3 control 6.1 6 0
3 control 6.1 12 0
3 control 6.1 18 0
3 control 6.1 24 1
3 control 6.1 30 1
3 control 6.1 36 0
3 control 6.1 42 1
3 control 6.1 48 0
...
How, then, to determine whether treatment females display different
mating patterns over time than control females? Here's my crack at it:
foo1 <- lmer2(mate ~ group * mass * time + (time | female), family=binomial)
Thanks in advance,
Steve