対応あり t 検定：介入前後の比較

library(tidyverse)

# 付属のサンプルデータを読み込む
dat <- read.csv("sample-data/sample_paired_t_scores.csv")

# 差分を作る：対応あり t 検定ではここが主役
dat <- dat %>%
  mutate(diff = after - before)

# 要約統計
dat %>%
  summarise(
    n = n(),
    mean_before = mean(before),
    sd_before = sd(before),
    mean_after = mean(after),
    sd_after = sd(after),
    mean_diff = mean(diff),
    sd_diff = sd(diff)
  )

# 図用に long 形式へ
dat_long <- dat %>%
  pivot_longer(
    cols = c(before, after),
    names_to = "time",
    values_to = "score"
  )

# 対応あり t 検定
tt <- t.test(dat$after, dat$before, paired = TRUE)
tt

# 差分の確認：ヒストグラムや QQ プロットと合わせて見る
shapiro.test(dat$diff)

# 効果量の導入として Cohen's d_z を手計算
d_z <- mean(dat$diff) / sd(dat$diff)
d_z

# 参考：effectsize パッケージを使う場合
# install.packages("effectsize")
# library(effectsize)
# cohens_d(dat$after, dat$before, paired = TRUE)

library(ggplot2)

# 1) 各対象の変化を見せる折れ線プロット
ggplot(dat_long, aes(x = time, y = score, group = id)) +
  geom_line(color = "#b8c4d8", linewidth = 0.7) +
  geom_point(aes(color = time), size = 2.5) +
  stat_summary(aes(group = 1), fun = mean, geom = "line", linewidth = 1.1, color = "black") +
  stat_summary(aes(group = 1), fun = mean, geom = "point", size = 3, color = "black") +
  labs(x = NULL, y = "Test score", title = "Before-after change by participant") +
  theme_minimal(base_size = 13) +
  theme(legend.position = "none")

# 2) 分布を見せる箱ひげ図
ggplot(dat_long, aes(x = time, y = score, fill = time)) +
  geom_boxplot(width = 0.55, alpha = 0.65, outlier.shape = 21) +
  geom_jitter(width = 0.08, alpha = 0.65, size = 2) +
  labs(x = NULL, y = "Test score", title = "Distribution before and after the intervention") +
  theme_minimal(base_size = 13) +
  theme(legend.position = "none")

# 主な出力のイメージ

# Paired t-test
# data:  dat$after and dat$before
# t = 5.71, df = 19, p-value = 1.68e-05
# alternative hypothesis: true mean difference is not equal to 0
# 95 percent confidence interval:
#   3.13  6.75
# sample estimates:
# mean difference
#           4.94

# d_z
# [1] 1.28