##TEMPLATE R CODE MULTILEVEL MODEL
##Example with dependent variable "relation_qua_repli"


#load packages
library(nlme) #for main multilevel analyses
library(performance) #for icc
library(sjPlot) #for export
library(foreign) #for import datafile

#set working directory
setwd("put file path")

#load data - all groups
dataquest<-read.spss("put name spss file",use.value.labels=FALSE, to.data.frame=TRUE)

#select dyads only (for triads, replace 2 by 3). Skip for Study 2
dataquest <- subset(dataquest,group_size=="2")

#Grand-mean centering age and positive affect
age_c <- dataquest$age - mean(dataquest$age)
affect_pos_c <- dataquest$affect_pos_m - mean(dataquest$affect_pos_m)

#1. Assess need for multilevel model (ICC)
#1.a. Random intercept model (variation across groups, no IV)
randomInterceptOnly <-lme(relation_qua_repli ~ 1, data = dataquest, random = ~1|group_number, method = "ML")
summary(randomInterceptOnly)
#1.b. Compute ICC (based on randamInterceptOnly model)
icc(randomInterceptOnly)

#2. Multilevel model: random intercept phone + covariates gender age positive affect (no random slope)
randomInterceptcov <-lme(relation_qua_repli ~ condition +gender+age_c+affect_pos_c, data = dataquest, random = ~1|group_number, method = "ML")
summary(randomInterceptcov)
intervals(randomInterceptcov)
#export results
table <- tab_model(randomInterceptcov, show.se=TRUE, show.std=TRUE, show.stat=TRUE,show.re.var=TRUE,show.df=TRUE)
table
#see results in viewer panel on bottom right

#3. Multilevel model: no covariates (random intercept phone)
randomInterceptphone <-lme(relation_qua_repli ~ condition, data = dataquest, random = ~1|group_number, method = "ML")
summary(randomInterceptphone)