# ======== ライブラリ ======== #
library(readxl)
## Warning: package 'readxl' was built under R version 4.3.3
library(dplyr)
##
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
##
## filter, lag
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
library(tidyr)
library(ggplot2)
library(lme4)
## Warning: package 'lme4' was built under R version 4.3.3
## Loading required package: Matrix
##
## Attaching package: 'Matrix'
## The following objects are masked from 'package:tidyr':
##
## expand, pack, unpack
library(lmerTest)
##
## Attaching package: 'lmerTest'
## The following object is masked from 'package:lme4':
##
## lmer
## The following object is masked from 'package:stats':
##
## step
library(emmeans)
## Warning: package 'emmeans' was built under R version 4.3.3
## Welcome to emmeans.
## Caution: You lose important information if you filter this package's results.
## See '? untidy'
library(emmeans)
library(effectsize)
## Warning: package 'effectsize' was built under R version 4.3.3
# ======== データ読み込み ======== #
df <- read_excel("~/Library/CloudStorage/Dropbox/土居_関口/2.Analysis_Doi/writing_results_scored_4criteria.xlsx")
# 条件を統一・順序付け
df <- df %>%
mutate(
Condition = factor(case_when(
条件 == "ai-wcf" ~ "AI",
条件 == "model text" ~ "Model",
条件 == "control" ~ "Control",
TRUE ~ 条件
), levels = c("AI", "Model", "Control")),
ID = as.factor(学籍番号)
)
# ======== LMEを回す関数を定義 ======== #
run_lme <- function(data, criterion_name, pre_col, post_col) {
# データを縦長化
df_long <- data %>%
select(ID, Condition, !!sym(pre_col), !!sym(post_col)) %>%
pivot_longer(
cols = c(!!sym(pre_col), !!sym(post_col)),
names_to = "Time",
values_to = "Score"
) %>%
mutate(
Time = ifelse(Time == pre_col, "Pre", "Post"),
Time = factor(Time, levels = c("Pre", "Post"))
)
# 混合効果モデル
model <- lmer(Score ~ Condition * Time + (1 | ID), data = df_long)
cat("\n\n===== LME結果:", criterion_name, "=====\n")
print(anova(model))
print(summary(model))
# EMM
emm <- emmeans(model, ~ Condition * Time)
# プロット
emm_df <- as.data.frame(emm)
p <- ggplot(emm_df, aes(x = Time, y = emmean, group = Condition, color = Condition)) +
geom_line(linewidth = 1.2) +
geom_point(size = 3) +
geom_errorbar(aes(ymin = emmean - SE, ymax = emmean + SE), width = 0.1, linewidth = 0.8) +
scale_color_manual(values = c("AI" = "#1f78b4", "Model" = "#33a02c", "Control" = "#e31a1c")) +
labs(
title = paste("Interaction Plot:", criterion_name, "(Condition × Time)"),
x = "Time (Pre / Post)",
y = "Estimated Mean Score",
color = "Condition"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(face = "bold", hjust = 0.5),
axis.title = element_text(face = "bold"),
axis.text = element_text(size = 12),
legend.position = "top",
legend.title = element_text(face = "bold")
) +
coord_cartesian(ylim = c(0, 9))
print(p)
}
# ======== 各観点＋合計スコアでLME実行 ======== #
# Task Achievement
run_lme(df, "Task Achievement", "Pre_TA", "Post_TA")
##
##
## ===== LME結果: Task Achievement =====
## Type III Analysis of Variance Table with Satterthwaite's method
## Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
## Condition 0.2619 0.1310 2 52 0.3465 0.7088
## Time 9.4410 9.4410 1 52 24.9805 6.935e-06 ***
## Condition:Time 1.0383 0.5191 2 52 1.3736 0.2622
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: Score ~ Condition * Time + (1 | ID)
## Data: df_long
##
## REML criterion at convergence: 279.2
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.84394 -0.55484 -0.04545 0.54726 2.06919
##
## Random effects:
## Groups Name Variance Std.Dev.
## ID (Intercept) 0.5077 0.7125
## Residual 0.3779 0.6148
## Number of obs: 110, groups: ID, 55
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 3.70000 0.21043 78.27649 17.583 < 2e-16 ***
## ConditionModel -0.20000 0.31565 78.27649 -0.634 0.528177
## ConditionControl -0.01579 0.30149 78.27649 -0.052 0.958365
## TimePost 0.70000 0.19441 52.00000 3.601 0.000708 ***
## ConditionModel:TimePost 0.05000 0.29161 52.00000 0.171 0.864525
## ConditionControl:TimePost -0.38421 0.27853 52.00000 -1.379 0.173661
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) CndtnM CndtnC TimPst CnM:TP
## ConditinMdl -0.667
## CondtnCntrl -0.698 0.465
## TimePost -0.462 0.308 0.322
## CndtnMdl:TP 0.308 -0.462 -0.215 -0.667
## CndtnCnt:TP 0.322 -0.215 -0.462 -0.698 0.465
# Coherence & Cohesion
run_lme(df, "Coherence & Cohesion", "Pre_CC", "Post_CC")
##
##
## ===== LME結果: Coherence & Cohesion =====
## Type III Analysis of Variance Table with Satterthwaite's method
## Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
## Condition 0.3378 0.1689 2 52 0.4507 0.6397
## Time 9.3918 9.3918 1 52 25.0613 6.741e-06 ***
## Condition:Time 0.2037 0.1019 2 52 0.2718 0.7631
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: Score ~ Condition * Time + (1 | ID)
## Data: df_long
##
## REML criterion at convergence: 278
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.16077 -0.46778 0.00935 0.51884 1.75971
##
## Random effects:
## Groups Name Variance Std.Dev.
## ID (Intercept) 0.4995 0.7067
## Residual 0.3748 0.6122
## Number of obs: 110, groups: ID, 55
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 3.6500 0.2091 78.4059 17.458 < 2e-16 ***
## ConditionModel -0.0875 0.3136 78.4059 -0.279 0.78097
## ConditionControl -0.1763 0.2995 78.4059 -0.589 0.55781
## TimePost 0.6000 0.1936 52.0000 3.099 0.00313 **
## ConditionModel:TimePost 0.0875 0.2904 52.0000 0.301 0.76436
## ConditionControl:TimePost -0.1263 0.2773 52.0000 -0.455 0.65069
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) CndtnM CndtnC TimPst CnM:TP
## ConditinMdl -0.667
## CondtnCntrl -0.698 0.465
## TimePost -0.463 0.309 0.323
## CndtnMdl:TP 0.309 -0.463 -0.215 -0.667
## CndtnCnt:TP 0.323 -0.215 -0.463 -0.698 0.465
# Lexical Resource
run_lme(df, "Lexical Resource", "Pre_LR", "Post_LR")
##
##
## ===== LME結果: Lexical Resource =====
## Type III Analysis of Variance Table with Satterthwaite's method
## Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
## Condition 0.5970 0.2985 2 52 0.7502 0.4773
## Time 8.7913 8.7913 1 52 22.0942 1.946e-05 ***
## Condition:Time 0.0728 0.0364 2 52 0.0915 0.9127
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: Score ~ Condition * Time + (1 | ID)
## Data: df_long
##
## REML criterion at convergence: 275
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.72272 -0.51164 0.04315 0.52693 1.64557
##
## Random effects:
## Groups Name Variance Std.Dev.
## ID (Intercept) 0.4111 0.6411
## Residual 0.3979 0.6308
## Number of obs: 110, groups: ID, 55
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 3.90000 0.20112 82.65733 19.392 <2e-16 ***
## ConditionModel -0.08750 0.30168 82.65733 -0.290 0.7725
## ConditionControl -0.32105 0.28814 82.65733 -1.114 0.2684
## TimePost 0.50000 0.19947 52.00000 2.507 0.0154 *
## ConditionModel:TimePost 0.12500 0.29921 52.00000 0.418 0.6778
## ConditionControl:TimePost 0.07895 0.28579 52.00000 0.276 0.7835
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) CndtnM CndtnC TimPst CnM:TP
## ConditinMdl -0.667
## CondtnCntrl -0.698 0.465
## TimePost -0.496 0.331 0.346
## CndtnMdl:TP 0.331 -0.496 -0.231 -0.667
## CndtnCnt:TP 0.346 -0.231 -0.496 -0.698 0.465
# Grammatical Range & Accuracy
run_lme(df, "Grammatical Range & Accuracy", "Pre_GRA", "Post_GRA")
##
##
## ===== LME結果: Grammatical Range & Accuracy =====
## Type III Analysis of Variance Table with Satterthwaite's method
## Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
## Condition 0.3696 0.1848 2 52 0.4685 0.6285
## Time 8.3291 8.3291 1 52 21.1177 2.786e-05 ***
## Condition:Time 0.3086 0.1543 2 52 0.3913 0.6782
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: Score ~ Condition * Time + (1 | ID)
## Data: df_long
##
## REML criterion at convergence: 261.8
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.76504 -0.57180 0.07507 0.55286 1.73361
##
## Random effects:
## Groups Name Variance Std.Dev.
## ID (Intercept) 0.2807 0.5298
## Residual 0.3944 0.6280
## Number of obs: 110, groups: ID, 55
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 3.55000 0.18373 88.67004 19.322 < 2e-16 ***
## ConditionModel 0.01250 0.27559 88.67004 0.045 0.96393
## ConditionControl -0.07632 0.26323 88.67004 -0.290 0.77255
## TimePost 0.55000 0.19860 52.00000 2.769 0.00777 **
## ConditionModel:TimePost 0.13750 0.29790 52.00000 0.462 0.64632
## ConditionControl:TimePost -0.12895 0.28453 52.00000 -0.453 0.65230
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) CndtnM CndtnC TimPst CnM:TP
## ConditinMdl -0.667
## CondtnCntrl -0.698 0.465
## TimePost -0.540 0.360 0.377
## CndtnMdl:TP 0.360 -0.540 -0.251 -0.667
## CndtnCnt:TP 0.377 -0.251 -0.540 -0.698 0.465
# Total
run_lme(df, "Total Score", "Pre_Total", "Post_Total")
##
##
## ===== LME結果: Total Score =====
## Type III Analysis of Variance Table with Satterthwaite's method
## Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
## Condition 4.009 2.005 2 52 0.4506 0.6397
## Time 143.718 143.718 1 52 32.3027 6.057e-07 ***
## Condition:Time 4.101 2.050 2 52 0.4608 0.6333
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: Score ~ Condition * Time + (1 | ID)
## Data: df_long
##
## REML criterion at convergence: 540.7
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.51356 -0.41599 -0.02121 0.52255 1.85533
##
## Random effects:
## Groups Name Variance Std.Dev.
## ID (Intercept) 6.813 2.610
## Residual 4.449 2.109
## Number of obs: 110, groups: ID, 55
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 14.8000 0.7504 76.1363 19.723 < 2e-16 ***
## ConditionModel -0.3625 1.1256 76.1363 -0.322 0.748301
## ConditionControl -0.5895 1.0751 76.1363 -0.548 0.585097
## TimePost 2.3500 0.6670 52.0000 3.523 0.000898 ***
## ConditionModel:TimePost 0.4000 1.0005 52.0000 0.400 0.690949
## ConditionControl:TimePost -0.5605 0.9556 52.0000 -0.587 0.560045
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) CndtnM CndtnC TimPst CnM:TP
## ConditinMdl -0.667
## CondtnCntrl -0.698 0.465
## TimePost -0.444 0.296 0.310
## CndtnMdl:TP 0.296 -0.444 -0.207 -0.667
## CndtnCnt:TP 0.310 -0.207 -0.444 -0.698 0.465
# ==== 単純効果比較関数 ==== #
compare_prepost_by_condition <- function(data, criterion_name, pre_col, post_col) {
df_long <- data %>%
select(ID, Condition, !!sym(pre_col), !!sym(post_col)) %>%
pivot_longer(
cols = c(!!sym(pre_col), !!sym(post_col)),
names_to = "Time",
values_to = "Score"
) %>%
mutate(
Time = ifelse(Time == pre_col, "Pre", "Post"),
Time = factor(Time, levels = c("Pre", "Post"))
)
# LME
model <- lmer(Score ~ Condition * Time + (1 | ID), data = df_long)
emm <- emmeans(model, ~ Condition * Time)
cat("\n\n===== 単純効果比較: ", criterion_name, "=====\n")
# 各群内のPre vs Post
result <- pairs(emm, by = "Condition", adjust = "tukey")
print(result)
# 平均とSEを一緒に出す
cat("\n推定平均（参考）:\n")
print(summary(emm))
}
# ==== 各観点で実行 ==== #
compare_prepost_by_condition(df, "Task Achievement", "Pre_TA", "Post_TA")
##
##
## ===== 単純効果比較: Task Achievement =====
## Condition = AI:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.700 0.194 52 -3.601 0.0007
##
## Condition = Model:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.750 0.217 52 -3.451 0.0011
##
## Condition = Control:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.316 0.199 52 -1.583 0.1194
##
## Degrees-of-freedom method: kenward-roger
##
## 推定平均（参考）:
## Condition Time emmean SE df lower.CL upper.CL
## AI Pre 3.70 0.210 78.3 3.28 4.12
## Model Pre 3.50 0.235 78.3 3.03 3.97
## Control Pre 3.68 0.216 78.3 3.25 4.11
## AI Post 4.40 0.210 78.3 3.98 4.82
## Model Post 4.25 0.235 78.3 3.78 4.72
## Control Post 4.00 0.216 78.3 3.57 4.43
##
## Degrees-of-freedom method: kenward-roger
## Confidence level used: 0.95
compare_prepost_by_condition(df, "Coherence & Cohesion", "Pre_CC", "Post_CC")
##
##
## ===== 単純効果比較: Coherence & Cohesion =====
## Condition = AI:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.600 0.194 52 -3.099 0.0031
##
## Condition = Model:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.688 0.216 52 -3.176 0.0025
##
## Condition = Control:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.474 0.199 52 -2.385 0.0208
##
## Degrees-of-freedom method: kenward-roger
##
## 推定平均（参考）:
## Condition Time emmean SE df lower.CL upper.CL
## AI Pre 3.65 0.209 78.4 3.23 4.07
## Model Pre 3.56 0.234 78.4 3.10 4.03
## Control Pre 3.47 0.215 78.4 3.05 3.90
## AI Post 4.25 0.209 78.4 3.83 4.67
## Model Post 4.25 0.234 78.4 3.78 4.72
## Control Post 3.95 0.215 78.4 3.52 4.37
##
## Degrees-of-freedom method: kenward-roger
## Confidence level used: 0.95
compare_prepost_by_condition(df, "Lexical Resource", "Pre_LR", "Post_LR")
##
##
## ===== 単純効果比較: Lexical Resource =====
## Condition = AI:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.500 0.199 52 -2.507 0.0154
##
## Condition = Model:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.625 0.223 52 -2.802 0.0071
##
## Condition = Control:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.579 0.205 52 -2.829 0.0066
##
## Degrees-of-freedom method: kenward-roger
##
## 推定平均（参考）:
## Condition Time emmean SE df lower.CL upper.CL
## AI Pre 3.90 0.201 82.7 3.50 4.30
## Model Pre 3.81 0.225 82.7 3.37 4.26
## Control Pre 3.58 0.206 82.7 3.17 3.99
## AI Post 4.40 0.201 82.7 4.00 4.80
## Model Post 4.44 0.225 82.7 3.99 4.88
## Control Post 4.16 0.206 82.7 3.75 4.57
##
## Degrees-of-freedom method: kenward-roger
## Confidence level used: 0.95
compare_prepost_by_condition(df, "Grammatical Range & Accuracy", "Pre_GRA", "Post_GRA")
##
##
## ===== 単純効果比較: Grammatical Range & Accuracy =====
## Condition = AI:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.550 0.199 52 -2.769 0.0078
##
## Condition = Model:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.688 0.222 52 -3.096 0.0032
##
## Condition = Control:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.421 0.204 52 -2.066 0.0438
##
## Degrees-of-freedom method: kenward-roger
##
## 推定平均（参考）:
## Condition Time emmean SE df lower.CL upper.CL
## AI Pre 3.55 0.184 88.7 3.18 3.92
## Model Pre 3.56 0.205 88.7 3.15 3.97
## Control Pre 3.47 0.189 88.7 3.10 3.85
## AI Post 4.10 0.184 88.7 3.73 4.47
## Model Post 4.25 0.205 88.7 3.84 4.66
## Control Post 3.89 0.189 88.7 3.52 4.27
##
## Degrees-of-freedom method: kenward-roger
## Confidence level used: 0.95
compare_prepost_by_condition(df, "Total Score", "Pre_Total", "Post_Total")
##
##
## ===== 単純効果比較: Total Score =====
## Condition = AI:
## contrast estimate SE df t.ratio p.value
## Pre - Post -2.35 0.667 52 -3.523 0.0009
##
## Condition = Model:
## contrast estimate SE df t.ratio p.value
## Pre - Post -2.75 0.746 52 -3.688 0.0005
##
## Condition = Control:
## contrast estimate SE df t.ratio p.value
## Pre - Post -1.79 0.684 52 -2.615 0.0117
##
## Degrees-of-freedom method: kenward-roger
##
## 推定平均（参考）:
## Condition Time emmean SE df lower.CL upper.CL
## AI Pre 14.8 0.750 76.1 13.3 16.3
## Model Pre 14.4 0.839 76.1 12.8 16.1
## Control Pre 14.2 0.770 76.1 12.7 15.7
## AI Post 17.1 0.750 76.1 15.7 18.6
## Model Post 17.2 0.839 76.1 15.5 18.9
## Control Post 16.0 0.770 76.1 14.5 17.5
##
## Degrees-of-freedom method: kenward-roger
## Confidence level used: 0.95
# --- データを縦長に変換（Total Score用） ---
df_long <- df %>%
select(ID, Condition, Pre_Total, Post_Total) %>%
pivot_longer(
cols = c(Pre_Total, Post_Total),
names_to = "Time",
values_to = "Score"
) %>%
mutate(
Time = factor(ifelse(Time == "Pre_Total", "Pre", "Post"),
levels = c("Pre", "Post")),
Condition = factor(Condition, levels = c("AI", "Model", "Control"))
)
# --- LMEモデル構築 ---
model_total <- lmer(Score ~ Condition * Time + (1 | ID), data = df_long)
# --- 効果量算出 ---
eta_sq <- eta_squared(model_total, partial = TRUE)
eta_sq
## # Effect Size for ANOVA (Type III)
##
## Parameter | Eta2 (partial) | 95% CI
## ----------------------------------------------
## Condition | 0.02 | [0.00, 1.00]
## Time | 0.38 | [0.22, 1.00]
## Condition:Time | 0.02 | [0.00, 1.00]
##
## - One-sided CIs: upper bound fixed at [1.00].
# ============================================
# 3要因混合効果モデル
# Criterion × Condition × Time + (1 | ID)
# ============================================
# --------------------------------------------
# 縦長に変換（Pre/Post × 4観点）
# --------------------------------------------
df_long <- df %>%
select(ID, Condition,
starts_with("Pre_TA"), starts_with("Pre_CC"), starts_with("Pre_LR"), starts_with("Pre_GRA"),
starts_with("Post_TA"), starts_with("Post_CC"), starts_with("Post_LR"), starts_with("Post_GRA")) %>%
pivot_longer(
cols = -c(ID, Condition),
names_to = c("Time", "Criterion"),
names_pattern = "(Pre|Post)_(.*)",
values_to = "Score"
) %>%
mutate(
Time = factor(Time, levels = c("Pre", "Post")),
Criterion = factor(Criterion, levels = c("TA", "CC", "LR", "GRA"))
)
# --------------------------------------------
# 3要因混合効果モデルの構築
# --------------------------------------------
model_3way <- lmer(Score ~ Criterion * Condition * Time + (1 | ID), data = df_long)
# --------------------------------------------
# 結果の出力
# --------------------------------------------
cat("\n===== Type III ANOVA: Criterion × Condition × Time =====\n")
##
## ===== Type III ANOVA: Criterion × Condition × Time =====
anova(model_3way, type = 3)
## Type III Analysis of Variance Table with Satterthwaite's method
## Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
## Criterion 3.605 1.202 3 364 4.2724 0.00555 **
## Condition 0.253 0.127 2 52 0.4506 0.63972
## Time 35.930 35.930 1 364 127.7290 < 2e-16 ***
## Criterion:Condition 0.858 0.143 6 364 0.5082 0.80213
## Criterion:Time 0.024 0.008 3 364 0.0280 0.99367
## Condition:Time 1.025 0.513 2 364 1.8222 0.16314
## Criterion:Condition:Time 0.598 0.100 6 364 0.3545 0.90711
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# --------------------------------------------
# 効果量（η²）を算出
# --------------------------------------------
eta_sq <- eta_squared(model_3way, partial = TRUE)
cat("\n===== 効果量 (partial η²) =====\n")
##
## ===== 効果量 (partial η²) =====
print(eta_sq)
## # Effect Size for ANOVA (Type III)
##
## Parameter | Eta2 (partial) | 95% CI
## --------------------------------------------------------
## Criterion | 0.03 | [0.01, 1.00]
## Condition | 0.02 | [0.00, 1.00]
## Time | 0.26 | [0.20, 1.00]
## Criterion:Condition | 8.31e-03 | [0.00, 1.00]
## Criterion:Time | 2.31e-04 | [0.00, 1.00]
## Condition:Time | 9.91e-03 | [0.00, 1.00]
## Criterion:Condition:Time | 5.81e-03 | [0.00, 1.00]
##
## - One-sided CIs: upper bound fixed at [1.00].
# --------------------------------------------
# 交互作用の推定平均（EMM）と単純効果比較
# --------------------------------------------
emm_3way <- emmeans(model_3way, ~ Criterion * Condition * Time)
cat("\n===== 単純効果比較: 各Criterion内のPre/Post差 =====\n")
##
## ===== 単純効果比較: 各Criterion内のPre/Post差 =====
pairs(emm_3way, by = c("Criterion", "Condition"))
## Criterion = TA, Condition = AI:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.700 0.168 364 -4.174 <.0001
##
## Criterion = CC, Condition = AI:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.600 0.168 364 -3.577 0.0004
##
## Criterion = LR, Condition = AI:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.500 0.168 364 -2.981 0.0031
##
## Criterion = GRA, Condition = AI:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.550 0.168 364 -3.279 0.0011
##
## Criterion = TA, Condition = Model:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.750 0.188 364 -4.000 0.0001
##
## Criterion = CC, Condition = Model:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.688 0.188 364 -3.666 0.0003
##
## Criterion = LR, Condition = Model:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.625 0.188 364 -3.333 0.0009
##
## Criterion = GRA, Condition = Model:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.688 0.188 364 -3.666 0.0003
##
## Criterion = TA, Condition = Control:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.316 0.172 364 -1.835 0.0673
##
## Criterion = CC, Condition = Control:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.474 0.172 364 -2.753 0.0062
##
## Criterion = LR, Condition = Control:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.579 0.172 364 -3.364 0.0008
##
## Criterion = GRA, Condition = Control:
## contrast estimate SE df t.ratio p.value
## Pre - Post -0.421 0.172 364 -2.447 0.0149
##
## Degrees-of-freedom method: kenward-roger
# --------------------------------------------
# 推定平均を確認
# --------------------------------------------
emm_df <- as.data.frame(emm_3way)
head(emm_df)
## Criterion Condition Time emmean SE df lower.CL upper.CL
## TA AI Pre 3.7000 0.2013696 104.36 3.300693 4.099307
## CC AI Pre 3.6500 0.2013696 104.36 3.250693 4.049307
## LR AI Pre 3.9000 0.2013696 104.36 3.500693 4.299307
## GRA AI Pre 3.5500 0.2013696 104.36 3.150693 3.949307
## TA Model Pre 3.5000 0.2251381 104.36 3.053561 3.946439
## CC Model Pre 3.5625 0.2251381 104.36 3.116061 4.008939
##
## Degrees-of-freedom method: kenward-roger
## Confidence level used: 0.95