# Manova or Linear mixed models

I’m a newbie in statistics and R, so I hope that my question isn’t too banal for you.

I’m going to carry out an experiment in order to verify whether cortisol and Hemoglobin levels are different when soccer players fail a penalty kick, compared with when they score a goal.

Each player pulls 3 penalty kicks. Here is an example for four subjects:

subj<-c(1,1,1,2,2,2,3,3,3,4,4,4)
err <-c("goal","fail","goal" ,"goal","goal","goal", "fail", "goal", "goal", "goal", "goal", "fail")
hemo<-c (.12,.23,.23,.43,.12,.13,.61,.23,.13,.34,.56,.11)
cort<-c (220,130,210,130,150,130,630,230,230,340,560,110)
data<-data.frame(subj, err,hemo, cort)


I don't know which statistics I have to use ... I'm a bit confused. Any advice? Thanks in advance

I would use linear mixed models:

library(lme4)
lmer (hemo ~ err + (1|subj))
lmer (cort ~ err + (1|subj))


Perhaps a repeated measures ANOVA would work as well, I'm not too familiar with that approach, but a linear mixed model is a good choice here.

• very thanks for your suggestion. If I wanted to extract a p value could I use a Likehood ratio test? a<-lmer (hemo ~ err + (1|subj)) b<-lmer(hemo ~ 1 + (1|subj)) anova(a,b) Sep 20 '15 at 20:22
• No. See my answer to this question for a quick review of when to use a likelihood ratio-test: stats.stackexchange.com/questions/173375/… . lmer() doesn't support p-values, but you can obtain them though some different methods: mindingthebrain.blogspot.se/2014/02/…
– JonB
Sep 21 '15 at 7:16