(Need Help Please) Trend and Multilevel Model (intercept and slope and graph) - graph

Question

I am just a newbie to R and Statistics. I desperately needs help. Here are my following questions:
I attempted to solve Q1, but I am not sure whether it is correct. But for Q2, I attempted to try too...but I truly have no clue for that.
**Q1. What is the trend of CES-D score accorss four waves? Whether the trends differ by gender and by age group. Plot a graph indicating this?
**Q2. Conduct a multilevel model to examine the basline factors associated with the trend (intercept and slope) of CES-D score. **
These are the column names Baseline Factors
Q1 Attempts
# subset data
DataCleanedSelected <- H_CHARLS_D_DataCleaned[c("id", "ragender", "r1agey", "raeduc_c", "r1shlt",
"r1dressa", "r1batha", "r1eata", "r1beda",
"r1toilta", "r1urina", "r1cesd10", "r2cesd10",
"r3cesd10", "r4cesd10", "h1rural")]
CESDSelected <- H_CHARLS_D_DataCleaned[c("id","ragender", "r1agey", "r1cesd10", "r2cesd10",
"r3cesd10", "r4cesd10")]
#Check dataset
head(CESDSelected)
names(CESDSelected)
#Attach dataset
attach(CESDSelected)
#Data preparation - Wide format
head(CESDSelected)
# Gather columns x4 into long format
# Convert id and time into factor variables
CHARLS.format <- gather(CESDSelected, time, cesd,
r1cesd10:r4cesd10, factor_key = TRUE)
CHARLS.format
# | Summary statistics
CHARLS.format %>%
group_by(time) %>%
get_summary_stats(cesd, type = "mean_sd")
ind.anova <- aov(cesd ~ time, data=CHARLS.format)
summary(ind.anova)
rm.anova <- aov(cesd ~ time + Error(id/time),
data = CHARLS.format)
summary(rm.anova)
# Visualise the Result#
time <- c("W1","W2","W3","W4")
mean.CESD <- c(mean(r1cesd10, na.rm = TRUE),mean(r2cesd10, na.rm = TRUE),
mean(r3cesd10, na.rm = TRUE),mean(r4cesd10, na.rm = TRUE))
se.CESD <- c(sd(r1cesd10, na.rm = TRUE), sd(r2cesd10, na.rm = TRUE),
sd(r3cesd10, na.rm = TRUE),sd(r4cesd10,na.rm = TRUE))/sqrt(length(id))
CESD.means.frame <- data.frame(time,mean.CESD,se.CESD)
CESD.means.frame
dev.off()
ggplot(CESD.means.frame,aes(x=time,y=mean.CESD)) +
geom_point(size=3)+
geom_errorbar(aes(ymin=mean.CESD-se.CESD,
ymax=mean.CESD+se.CESD),width=.2)
contrasts(CHARLS.format$time) <- matrix(c(-3,-1,1,3,1,-1,
-1,1,-1,3,-3,1), 4, 3)
rm.anova <- aov(cesd ~ time + Error(id/time),
data=CHARLS.format,
contrasts = contrasts(CHARLS.format$time))
summary(rm.anova,split=list(Time=list("Linear" =1,
"Quadratic"=2,"Cubic" =3)))
# Visualisation#
CHARLS.format %>%
group_by(time) %>%
identify_outliers(cesd)
# Normality Assumption#
CHARLS.format%>%
group_by(time) %>%
shapiro_test(cesd)
res.aov <- anova_test(data = CHARLS.format, dv = cesd, wid = ID, within = time)
get_anova_table(res.aov)
##Post-hoc test - pairwise comparisons##
pwc <- CHARLS.format %>%
pairwise_t_test(
cesd ~ time, paired = TRUE,
p.adjust.method = "bonferroni"
)
pwc
##Visualization: box plots with p-values##
pwc <- pwc %>% add_xy_position(x = "time")
bxp +
stat_pvalue_manual(pwc) +
labs(
subtitle = get_test_label(res.aov, detailed = TRUE),
caption = get_pwc_label(pwc)
)
Q2 Attempts
library(lme4)
library(lattice)
library(Matrix)
##Select those who are female
Female <- CHARLS.format[ragender =2, na.rm =False]
str(Female)
summary(Female)
names(CHARLS.format)
##Null hypothesis##
lmer(cesd ~ 1 + (1 | id), data=CHARLS.format)
##Fixed predictors - Gender - Random intercept##
lmer(cesd ~ ragender + (1 | id), data=CHARLS.format)
model.GenderIntercept <- lmer(cesd ~ ragender + (1 | id), data=CHARLS.format)
plot(model.GenderIntercept, main="Residual Plot of Gender Intercept")
##Fixed predictors - Age - Random intercept##
lmer(cesd ~ r1agey + (1 | id), data=CHARLS.format)
model.AgeIntercept <- lmer(cesd ~ r1agey + (1 | id), data=CHARLS.format)
plot(model.AgeIntercept, main="Residual Plot of Age Intercept")
##Random intercept, Random slope - Gender##
lmer(cesd ~ ragender + (ragender | id), data=CHARLS.format)
model.GenderSlope <- lmer(cesd ~ ragender + (ragender | id), data=CHARLS.format)
plot(model.GenderSlope, main="Residual Plot of Gender Slope")
##Random intercept, Random slope - Age##
lmer(cesd ~ r1agey + (r1agey | id), data=CHARLS.format)
model.AgeSlope <- lmer(cesd ~ r1agey + (r1agey | id), data=CHARLS.format)
plot(model.AgeSlope, main="Residual Plot of Age Slope")
##Random intercept, individual and group level predictors##
lmer(cesd ~ ragender + r1agey + (1 + ragender | id), data=CHARLS.format)
Model.IndiGrp <- lmer(cesd ~ ragender + r1agey + (1 + ragender | id), data=CHARLS.format)
plot(Model.IndiGrp, main="Random intercept, individual and group level predictors")
##Random intercept, cross-level interaction##
lmer(cesd ~ ragender * r1agey + (1 + ragender | id), data=CHARLS.format)
Model.RandomCross <- lmer(cesd ~ ragender + r1agey + (1 + ragender | id), data=CHARLS.format)
plot(Model.RandomCross, main="Random intercept, cross-level interaction")