使用 lmer() 进行重复测量方差分析的问题

Question

I am trying to use lmer() to perform a sort of repeated measures ANOVA. I sampled the same pools multiple times throughout the year, and am trying to see if there is a difference in a response variable (body size) based on the sampling date, with pool as a random effect. My issue in this case is that I am getting separate estimates and p-values for my fixed effects (sampling dates), instead of the single estimate and p-value I would normally get from this type of model.

require(lme4)
df <- data.frame(wing_length = c(5.1, 4.9, 4.7, 4.6, 5.1,2.4,4.3,4.4),
  date = c('Jan','Jan','Feb','Feb','Mar','Mar','Apr','Apr'),
  pool = c('1','2','1','2','1','2','1','2'))
mod <- lmer(wing_length ~ date + (1|pool),df)

summary(mod)

The summary output gives me various estimates and p-values for the different dates, but I was expecting one single estimate and p-value that tells me whether there is a difference in the mean wing length based on date.

I think my issue comes from my data format or some misunderstanding of what I am trying to do with this function.

Thank you for any advice you can offer.

Answer 1

You will need to convert date from a factor to an integer variable. Otherwise, lmer will automatically create dummmies.

library(lme4)
df <- data.frame(wing_length = c(5.1, 4.9, 4.7, 4.6, 5.1,2.4,4.3,4.4),
                 date = c('Jan','Jan','Feb','Feb','Mar','Mar','Apr','Apr'),
                 pool = c('1','2','1','2','1','2','1','2'))

df$date <- match(tolower(df$date), tolower(month.abb))
mod <- lmer(wing_length ~ date + (1|pool),df)

summary(mod)

Note that converting date to an integer will produce a different model--one where you are estimating the month trend instead of individual month effects (fixed effects).