Message-ID: <CAJ0DVYySnrK2-Kz6cD=GxTiPnKZTYVVhzAk5z_TapUd3u5o6Fg@mail.gmail.com>
Date: 2014-09-28T15:41:57Z
From: Etn
Subject: individual Linear mixed models
Hi all,
I would be very grateful if you could please give any advice...
I am running a linear mixed model on a complete randomised block experiment
blocks = 11
treatments = 5 (A,B,C,D,E)
data recorded on days 1, 2, 11 and 18
I'm interested in looking at the difference between treatments on a
particular day
e.g. difference between treatments on day 1
(I am not interested in the difference between treatments on day 1 versus
day 11 etc.)
My question is: Is it OK to run linear mixed regression models on each day
individually? as I'm not concerned with the changes over time.....
(Results for individual days show a difference between treatments on day 1
and day 2, and no difference between treatments on day 11 and 18 - this
follows with what would be expected in reality)
fit <- lme(y ~ Treatment, random = ~1|block)
Following this I run a tukey HSD to find comparison between treatments
Is this the correct approach ? any advice is greatly appreciated
Aside: I did try a more complicated model:
fit <- lme(y ~ Treatment*day, random = ~1|block/day)
however none of the treatments were significant and it only showed that day
11 and day 8 were significantly different)
Many thanks
Etn
[[alternative HTML version deleted]]