Descriptive statistics

Gender breakdown

Code

The following packages were used to generate this table:

library(papaja)
library(tidyverse)
library(knitr)
library(kableExtra)
library(here)

The files needed for this table are available at osf.io/mzfu9 in the Individual Study Output folder.

First we load the individual study analysis objects.

study.names = c("EAS", "HRS", "LBC", "LBLS", "MAP", "MAS", "MIDUS", "NAS", "OATS", "ROS", "SLS", "WLS")

lapply(here(paste0("mortality/study output/", study.names, "_survival_output.Rdata")), load, .GlobalEnv)

First we extract the Cronbach’s alpha values from the data objects. These are stored in a dataframe, with each reliability coefficient from each perosnality meausre from each study comprising a single row.

alpha.list <- data.frame()
n = 0
for(i in study.names){
  n = n+1
  x = get(paste0(i,"_survival_output")) # get output object
  if(!is.null(x$alpha)){
    y = as.data.frame(unlist(x$alpha))
    y$study = i
    alpha.list = rbind(alpha.list, y)
  }
}

alpha.list <- alpha.list %>%
  mutate(var = rownames(.),
         var = gsub("[0-9]", "", var),
         var = gsub("alpha_", "", var)) %>%
  rename(alpha = `unlist(x$alpha)`)

We extract and wrangle the descriptive statistics (a data frame created using the describe() function in the psych package).

We identify the rows corresponding to each data set. We then select the minimum and maximum rows as starting and end points for grouping.

rows.EAS = which(describe.df$study == "EAS"); rows.EAS = c(min(rows.EAS), max(rows.EAS))
rows.LBC = which(describe.df$study == "LBC"); rows.LBC = c(min(rows.LBC), max(rows.LBC))
rows.LBLS = which(describe.df$study == "LBLS"); rows.LBLS = c(min(rows.LBLS), max(rows.LBLS))
rows.HRS = which(describe.df$study == "HRS"); rows.HRS = c(min(rows.HRS), max(rows.HRS))
rows.MAP = which(describe.df$study == "MAP"); rows.MAP = c(min(rows.MAP), max(rows.MAP))
rows.MAS = which(describe.df$study == "MAS"); rows.MAS = c(min(rows.MAS), max(rows.MAS))
rows.MIDUS = which(describe.df$study == "MIDUS"); rows.MIDUS = c(min(rows.MIDUS), max(rows.MIDUS))
rows.NAS = which(describe.df$study == "NAS"); rows.NAS = c(min(rows.NAS), max(rows.NAS))
rows.OATS = which(describe.df$study == "OATS"); rows.OATS = c(min(rows.OATS), max(rows.OATS))
rows.ROS = which(describe.df$study == "ROS"); rows.ROS = c(min(rows.ROS), max(rows.ROS))
rows.SLS = which(describe.df$study == "SLS"); rows.SLS = c(min(rows.SLS), max(rows.SLS))
rows.WLS = which(describe.df$study == "WLS"); rows.WLS = c(min(rows.WLS), max(rows.WLS))

Finally, we pipe the data frame into the kable() function and additional formatting through the kableExtra package. We remove the study column, as this becomes redundant with the grouping headers.

describe.df %>%
  select(-study) %>%
  kable(.,  booktabs = T, escape = F, digits = 2, format = "html", 
        col.names = c("Variable", "N Valid", "Mean", "SD", "Min", "Max", "$\\alpha$")) %>%
  kable_styling(full_width = T, latex_options = c("repeat_header")) %>%
  group_rows("EAS", rows.EAS[1], rows.EAS[2]) %>%
  group_rows("HRS", rows.HRS[1], rows.HRS[2]) %>%
  group_rows("LBC1936", rows.LBC[1], rows.LBC[2]) %>%
  group_rows("LBLS", rows.LBLS[1], rows.LBLS[2]) %>%
  group_rows("MAP", rows.MAP[1], rows.MAP[2]) %>%
  group_rows("MAS", rows.MAS[1], rows.MAS[2]) %>%
  group_rows("MIDUS", rows.MIDUS[1], rows.MIDUS[2]) %>%
  group_rows("NAS", rows.NAS[1], rows.NAS[2]) %>%
  group_rows("OATS", rows.OATS[1], rows.OATS[2]) %>%
  group_rows("ROS", rows.ROS[1], rows.ROS[2]) %>%
  group_rows("SLS", rows.SLS[1], rows.SLS[2]) %>%
  group_rows("WLS", rows.WLS[1], rows.WLS[2]) 

The code to calculate gender breakdown is:

gender.tab <- data.frame()
n <- 0
for(i in study.names){
  x <- get(paste0(i, "_survival_output"))
  n <- n+1
  gender.tab[n, "study"] <- i
  if(!is.null(x$descriptives$gender.tab)){
    table <- x$descriptives$gender.tab
    gender.tab[n, "male"] <- table[which(names(table) == x$descriptives$gender.lab$male)]
    gender.tab[n, "female"] <- table[which(names(table) == x$descriptives$gender.lab$female)]
    gender.tab[n, "percent female"] <- gender.tab[n, "female"]/(gender.tab[n, "female"] + 
                                                                gender.tab[n, "male"])*100}
}

kable(gender.tab, booktabs = T, escape = F, digits = 2, format = "html") %>%
  kable_styling()