te_metaanalysis.Rmd

---
title: "TE Meta-analysis"
author: "Alvin Tan"
date: "2022-03-09"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(tidyverse)
library(meta)
```

## Import data
```{r import, include=FALSE}
data_df <- read_tsv("data.tsv") |>
  select(!(`Publication Year`:Url)) |>
  mutate(`%TE` = as.double(`%TE`),
         `%TE(either)` = as.double(`%TE(either)`),
         Measure = fct_recode(`C/P`, "Comprehension" = "C", "Production" = "P"),
         `Study name` = sub("–", "-", `Study name`))
```

```{r descriptives}
num_te_discuss <- sum(data_df$`Discusses TEs?`, na.rm = TRUE)

langs_list <- list()
for (i in data_df$Langs) {
  langs_list <- c(langs_list, str_split(i, ","))
}
langs_list <- langs_list |> unlist()
langs_hist <- langs_list |> table() |> as_tibble()

studies_con <- data_df |> 
  filter(!is.na(`SD(%TE)`))
num_studies_con <- studies_con |> nrow()

studies_lib <- data_df |>
  filter(!is.na(`SD(%TE(either))`))
num_studies_lib <- studies_lib |> nrow()
```

```{r lang-hist}
lh <- ggplot(data = langs_hist, aes(x = reorder(langs_list, -n), y = n)) +
  geom_col() +
  xlab("Language")
lh
```

```{r meta-analysis}
ma_con <- metamean(n = N,
                   mean = `%TE(either)`,
                   sd = `SD(%TE(either))`,
                   studlab = `Study name`,
                   data = studies_con,
                   sm = "MRAW",
                   random = TRUE,
                   method.tau = "REML",
                   title = "Proportion of TEs",
                   subgroup = Measure)
summary(ma_con)

ma_lib <- metamean(n = N,
                   mean = `%TE(either)`,
                   sd = `SD(%TE(either))`,
                   studlab = `Study name`,
                   data = studies_lib,
                   sm = "MRAW",
                   random = TRUE,
                   method.tau = "REML",
                   title = "Proportion of TEs",
                   subgroup = Measure)
summary(ma_lib)
```

```{r forest}
forest.meta(ma_con, sortvar = `%TE(either)`)
forest.meta(ma_lib, sortvar = `%TE(either)`)
```