478 lines
17 KiB
R
478 lines
17 KiB
R
# mixed anova not working
|
||
# 12/09/2025
|
||
# add sum contrasts
|
||
|
||
# Mixed ANOVA Analysis for Domain Means
|
||
# EOHI Experiment Data Analysis - Domain Level Analysis
|
||
# Variables: NPast_mean_pref, NPast_mean_pers, NPast_mean_val, NPast_mean_life
|
||
# NFut_mean_pref, NFut_mean_pers, NFut_mean_val, NFut_mean_life
|
||
|
||
# Load required libraries
|
||
library(tidyverse)
|
||
library(ez)
|
||
library(car)
|
||
library(nortest) # For normality tests
|
||
library(ggplot2) # For plotting
|
||
library(emmeans) # For post-hoc comparisons
|
||
library(purrr) # For map functions
|
||
|
||
# Global options to remove scientific notation
|
||
options(scipen = 999)
|
||
|
||
# Set contrasts to sum for mixed ANOVA (necessary for proper interpretation)
|
||
options(contrasts = c("contr.sum", "contr.poly"))
|
||
|
||
setwd("C:/Users/irina/Documents/DND/EOHI/eohi1")
|
||
|
||
# Read the data
|
||
data <- read.csv("exp1.csv")
|
||
|
||
# Display basic information about the dataset
|
||
print(paste("Dataset dimensions:", paste(dim(data), collapse = " x")))
|
||
print(paste("Number of participants:", length(unique(data$pID))))
|
||
|
||
# Verify the specific variables we need
|
||
required_vars <- c("NPast_mean_pref", "NPast_mean_pers", "NPast_mean_val", "NPast_mean_life",
|
||
"NFut_mean_pref", "NFut_mean_pers", "NFut_mean_val", "NFut_mean_life")
|
||
|
||
missing_vars <- required_vars[!required_vars %in% colnames(data)]
|
||
if (length(missing_vars) > 0) {
|
||
print(paste("Warning: Missing variables:", paste(missing_vars, collapse = ", ")))
|
||
} else {
|
||
print("All required domain mean variables found!")
|
||
}
|
||
|
||
# Define domain mapping
|
||
domain_mapping <- data.frame(
|
||
variable = c("NPast_mean_pref", "NPast_mean_pers", "NPast_mean_val", "NPast_mean_life",
|
||
"NFut_mean_pref", "NFut_mean_pers", "NFut_mean_val", "NFut_mean_life"),
|
||
time = c(rep("Past", 4), rep("Future", 4)),
|
||
domain = rep(c("Preferences", "Personality", "Values", "Life"), 2),
|
||
stringsAsFactors = FALSE
|
||
)
|
||
|
||
# Domain mapping created
|
||
|
||
# More efficient data pivoting (avoiding pivot_longer issues)
|
||
pivot_domain_means <- function(data, domain_mapping) {
|
||
# Pre-allocate the result data frame for better performance
|
||
n_rows <- nrow(data) * nrow(domain_mapping)
|
||
long_data <- data.frame(
|
||
pID = character(n_rows),
|
||
ResponseId = character(n_rows),
|
||
TEMPORAL_DO = character(n_rows),
|
||
TIME = character(n_rows),
|
||
DOMAIN = character(n_rows),
|
||
MEAN_DIFFERENCE = numeric(n_rows),
|
||
stringsAsFactors = FALSE
|
||
)
|
||
|
||
row_idx <- 1
|
||
|
||
for (i in 1:nrow(domain_mapping)) {
|
||
var_name <- domain_mapping$variable[i]
|
||
time_level <- domain_mapping$time[i]
|
||
domain_level <- domain_mapping$domain[i]
|
||
|
||
# Check if variable exists
|
||
if (!var_name %in% colnames(data)) {
|
||
print(paste("Warning: Variable", var_name, "not found in data"))
|
||
next
|
||
}
|
||
|
||
# Get the number of rows for this variable
|
||
n_data_rows <- nrow(data)
|
||
|
||
# Fill in the data for this variable
|
||
long_data$pID[row_idx:(row_idx + n_data_rows - 1)] <- as.character(data$pID)
|
||
long_data$ResponseId[row_idx:(row_idx + n_data_rows - 1)] <- as.character(data$ResponseId)
|
||
long_data$TEMPORAL_DO[row_idx:(row_idx + n_data_rows - 1)] <- as.character(data$TEMPORAL_DO)
|
||
long_data$TIME[row_idx:(row_idx + n_data_rows - 1)] <- time_level
|
||
long_data$DOMAIN[row_idx:(row_idx + n_data_rows - 1)] <- domain_level
|
||
long_data$MEAN_DIFFERENCE[row_idx:(row_idx + n_data_rows - 1)] <- data[[var_name]]
|
||
|
||
row_idx <- row_idx + n_data_rows
|
||
}
|
||
|
||
# Convert to factors with proper levels
|
||
long_data$TIME <- factor(long_data$TIME, levels = c("Past", "Future"))
|
||
long_data$DOMAIN <- factor(long_data$DOMAIN, levels = c("Preferences", "Personality", "Values", "Life"))
|
||
long_data$pID <- as.factor(long_data$pID)
|
||
long_data$TEMPORAL_DO <- as.factor(long_data$TEMPORAL_DO)
|
||
|
||
return(long_data)
|
||
}
|
||
|
||
# Pivot data to long format
|
||
tryCatch({
|
||
long_data <- pivot_domain_means(data, domain_mapping)
|
||
}, error = function(e) {
|
||
print(paste("Error in data pivoting:", e$message))
|
||
stop("Cannot proceed without proper data structure")
|
||
})
|
||
|
||
print(paste("Long data dimensions:", paste(dim(long_data), collapse = " x")))
|
||
print(paste("Number of participants:", length(unique(long_data$pID))))
|
||
|
||
# =============================================================================
|
||
# DESCRIPTIVE STATISTICS
|
||
# =============================================================================
|
||
|
||
# Overall descriptive statistics by TIME and DOMAIN
|
||
desc_stats <- long_data %>%
|
||
group_by(TIME, DOMAIN) %>%
|
||
summarise(
|
||
n = n(),
|
||
mean = round(mean(MEAN_DIFFERENCE, na.rm = TRUE), 5),
|
||
variance = round(var(MEAN_DIFFERENCE, na.rm = TRUE), 5),
|
||
sd = round(sd(MEAN_DIFFERENCE, na.rm = TRUE), 5),
|
||
median = round(median(MEAN_DIFFERENCE, na.rm = TRUE), 5),
|
||
q1 = round(quantile(MEAN_DIFFERENCE, 0.25, na.rm = TRUE), 5),
|
||
q3 = round(quantile(MEAN_DIFFERENCE, 0.75, na.rm = TRUE), 5),
|
||
min = round(min(MEAN_DIFFERENCE, na.rm = TRUE), 5),
|
||
max = round(max(MEAN_DIFFERENCE, na.rm = TRUE), 5),
|
||
.groups = 'drop'
|
||
)
|
||
|
||
print("Descriptive statistics by TIME and DOMAIN:")
|
||
print(desc_stats)
|
||
|
||
# Descriptive statistics by between-subjects factors
|
||
desc_stats_by_temporal <- long_data %>%
|
||
group_by(TEMPORAL_DO, TIME, DOMAIN) %>%
|
||
summarise(
|
||
n = n(),
|
||
mean = round(mean(MEAN_DIFFERENCE, na.rm = TRUE), 5),
|
||
variance = round(var(MEAN_DIFFERENCE, na.rm = TRUE), 5),
|
||
sd = round(sd(MEAN_DIFFERENCE, na.rm = TRUE), 5),
|
||
.groups = 'drop'
|
||
)
|
||
|
||
print("Descriptive statistics by TEMPORAL_DO, TIME, and DOMAIN:")
|
||
print(desc_stats_by_temporal)
|
||
|
||
|
||
# =============================================================================
|
||
# ASSUMPTION TESTING
|
||
# =============================================================================
|
||
|
||
# Remove missing values for assumption testing
|
||
long_data_clean <- long_data[!is.na(long_data$MEAN_DIFFERENCE), ]
|
||
print(paste("Data after removing missing values:", paste(dim(long_data_clean), collapse = " x")))
|
||
|
||
# 1. Missing values check
|
||
missing_summary <- long_data %>%
|
||
group_by(TIME, DOMAIN) %>%
|
||
summarise(
|
||
n_total = n(),
|
||
n_missing = sum(is.na(MEAN_DIFFERENCE)),
|
||
pct_missing = round(100 * n_missing / n_total, 2),
|
||
.groups = 'drop'
|
||
)
|
||
|
||
print("Missing values by TIME and DOMAIN:")
|
||
print(missing_summary)
|
||
|
||
# 2. Outlier detection
|
||
outlier_summary <- long_data_clean %>%
|
||
group_by(TIME, DOMAIN) %>%
|
||
summarise(
|
||
n = n(),
|
||
mean = mean(MEAN_DIFFERENCE),
|
||
sd = sd(MEAN_DIFFERENCE),
|
||
q1 = quantile(MEAN_DIFFERENCE, 0.25),
|
||
q3 = quantile(MEAN_DIFFERENCE, 0.75),
|
||
iqr = q3 - q1,
|
||
lower_bound = q1 - 1.5 * iqr,
|
||
upper_bound = q3 + 1.5 * iqr,
|
||
n_outliers = sum(MEAN_DIFFERENCE < lower_bound | MEAN_DIFFERENCE > upper_bound),
|
||
.groups = 'drop'
|
||
)
|
||
|
||
print("Outlier summary (IQR method):")
|
||
print(outlier_summary)
|
||
|
||
# 3. Anderson-Darling normality test (streamlined)
|
||
normality_results <- long_data_clean %>%
|
||
group_by(TIME, DOMAIN) %>%
|
||
summarise(
|
||
n = n(),
|
||
ad_statistic = ad.test(.data$MEAN_DIFFERENCE)$statistic,
|
||
ad_p_value = ad.test(.data$MEAN_DIFFERENCE)$p.value,
|
||
.groups = 'drop'
|
||
)
|
||
|
||
print("Anderson-Darling normality test results:")
|
||
# Round only the numeric columns
|
||
normality_results_rounded <- normality_results %>%
|
||
mutate(across(where(is.numeric), ~ round(.x, 5)))
|
||
print(normality_results_rounded)
|
||
|
||
# 4. Homogeneity of variance (Levene's test)
|
||
# Test homogeneity across TIME within each DOMAIN
|
||
homogeneity_time <- long_data_clean %>%
|
||
group_by(DOMAIN) %>%
|
||
summarise(
|
||
levene_F = leveneTest(MEAN_DIFFERENCE ~ TIME)$`F value`[1],
|
||
levene_p = leveneTest(MEAN_DIFFERENCE ~ TIME)$`Pr(>F)`[1],
|
||
.groups = 'drop'
|
||
)
|
||
|
||
print("Homogeneity of variance across TIME within each DOMAIN:")
|
||
print(homogeneity_time)
|
||
|
||
# Test homogeneity across DOMAIN within each TIME
|
||
homogeneity_domain <- long_data_clean %>%
|
||
group_by(TIME) %>%
|
||
summarise(
|
||
levene_F = leveneTest(MEAN_DIFFERENCE ~ DOMAIN)$`F value`[1],
|
||
levene_p = leveneTest(MEAN_DIFFERENCE ~ DOMAIN)$`Pr(>F)`[1],
|
||
.groups = 'drop'
|
||
)
|
||
|
||
print("Homogeneity of variance across DOMAIN within each TIME:")
|
||
print(homogeneity_domain)
|
||
|
||
# =============================================================================
|
||
# HARTLEY'S F-MAX TEST WITH BOOTSTRAP CRITICAL VALUES
|
||
# =============================================================================
|
||
|
||
# Function to calculate Hartley's F-max ratio
|
||
calculate_hartley_ratio <- function(variances) {
|
||
max(variances, na.rm = TRUE) / min(variances, na.rm = TRUE)
|
||
}
|
||
|
||
# =============================================================================
|
||
# CALCULATE OBSERVED F-MAX RATIOS FOR MIXED ANOVA
|
||
# =============================================================================
|
||
|
||
# For mixed ANOVA: Test homogeneity across BETWEEN-SUBJECTS factor (TEMPORAL_DO)
|
||
# within each combination of within-subjects factors (TIME × DOMAIN)
|
||
|
||
# First, let's check what values TEMPORAL_DO actually has
|
||
print("=== CHECKING TEMPORAL_DO VALUES ===")
|
||
print("Unique TEMPORAL_DO values:")
|
||
print(unique(long_data_clean$TEMPORAL_DO))
|
||
print("TEMPORAL_DO value counts:")
|
||
print(table(long_data_clean$TEMPORAL_DO))
|
||
|
||
print("\n=== OBSERVED F-MAX RATIOS: TEMPORAL_DO within each TIME × DOMAIN combination ===")
|
||
|
||
observed_temporal_ratios <- long_data_clean %>%
|
||
group_by(TIME, DOMAIN) %>%
|
||
summarise(
|
||
# Calculate variances for each TEMPORAL_DO level within this TIME × DOMAIN combination
|
||
past_var = var(MEAN_DIFFERENCE[TEMPORAL_DO == "01PAST"], na.rm = TRUE),
|
||
fut_var = var(MEAN_DIFFERENCE[TEMPORAL_DO == "02FUT"], na.rm = TRUE),
|
||
# Calculate F-max ratio
|
||
f_max_ratio = max(past_var, fut_var) / min(past_var, fut_var),
|
||
.groups = 'drop'
|
||
) %>%
|
||
select(TIME, DOMAIN, past_var, fut_var, f_max_ratio)
|
||
|
||
print(observed_temporal_ratios)
|
||
|
||
# More efficient bootstrap function for Hartley's F-max test
|
||
bootstrap_hartley_critical <- function(data, group_var, response_var, n_iter = 1000) {
|
||
# Get unique groups and their sample sizes
|
||
groups <- unique(data[[group_var]])
|
||
|
||
# Calculate observed variances for each group
|
||
observed_vars <- data %>%
|
||
group_by(!!sym(group_var)) %>%
|
||
summarise(var = var(!!sym(response_var), na.rm = TRUE), .groups = 'drop') %>%
|
||
pull(var)
|
||
|
||
# Handle invalid variances
|
||
if(any(observed_vars <= 0 | is.na(observed_vars))) {
|
||
observed_vars[observed_vars <= 0 | is.na(observed_vars)] <- 1e-10
|
||
}
|
||
|
||
# Calculate observed F-max ratio
|
||
observed_ratio <- max(observed_vars) / min(observed_vars)
|
||
|
||
# Pre-allocate storage for bootstrap ratios
|
||
bootstrap_ratios <- numeric(n_iter)
|
||
|
||
# Get group data once
|
||
group_data_list <- map(groups, ~ {
|
||
group_data <- data[data[[group_var]] == .x, response_var]
|
||
group_data[!is.na(group_data)]
|
||
})
|
||
|
||
# Bootstrap with pre-allocated storage
|
||
for(i in 1:n_iter) {
|
||
# Bootstrap sample from each group independently
|
||
sample_vars <- map_dbl(group_data_list, ~ {
|
||
bootstrap_sample <- sample(.x, size = length(.x), replace = TRUE)
|
||
var(bootstrap_sample, na.rm = TRUE)
|
||
})
|
||
bootstrap_ratios[i] <- max(sample_vars) / min(sample_vars)
|
||
}
|
||
|
||
# Remove invalid ratios
|
||
valid_ratios <- bootstrap_ratios[is.finite(bootstrap_ratios) & !is.na(bootstrap_ratios)]
|
||
|
||
if(length(valid_ratios) == 0) {
|
||
stop("No valid bootstrap ratios generated")
|
||
}
|
||
|
||
# Calculate critical value (95th percentile)
|
||
critical_95 <- quantile(valid_ratios, 0.95, na.rm = TRUE)
|
||
|
||
# Return only essential information
|
||
return(list(
|
||
observed_ratio = observed_ratio,
|
||
critical_95 = critical_95,
|
||
n_valid_iterations = length(valid_ratios)
|
||
))
|
||
}
|
||
|
||
# Hartley's F-max test across TEMPORAL_DO within each TIME × DOMAIN combination
|
||
print("\n=== HARTLEY'S F-MAX TEST RESULTS ===")
|
||
set.seed(123) # For reproducibility
|
||
|
||
hartley_temporal_results <- long_data_clean %>%
|
||
group_by(TIME, DOMAIN) %>%
|
||
summarise(
|
||
hartley_result = list(bootstrap_hartley_critical(pick(TEMPORAL_DO, MEAN_DIFFERENCE), "TEMPORAL_DO", "MEAN_DIFFERENCE")),
|
||
.groups = 'drop'
|
||
) %>%
|
||
mutate(
|
||
observed_ratio = map_dbl(hartley_result, ~ .x$observed_ratio),
|
||
critical_95 = map_dbl(hartley_result, ~ .x$critical_95),
|
||
significant = observed_ratio > critical_95
|
||
) %>%
|
||
select(TIME, DOMAIN, observed_ratio, critical_95, significant)
|
||
|
||
print(hartley_temporal_results)
|
||
|
||
# =============================================================================
|
||
# MIXED ANOVA ANALYSIS
|
||
# =============================================================================
|
||
|
||
# Check for missing data patterns
|
||
table(long_data_clean$pID, long_data_clean$TIME, long_data_clean$DOMAIN, useNA = "ifany")
|
||
|
||
# Check data balance
|
||
xtabs(~ pID + TIME + DOMAIN, data = long_data_clean)
|
||
|
||
# Check data dimensions and structure
|
||
print(paste("Data size for ANOVA:", nrow(long_data_clean), "rows"))
|
||
print(paste("Number of participants:", length(unique(long_data_clean$pID))))
|
||
print(paste("Number of TIME levels:", length(levels(long_data_clean$TIME))))
|
||
print(paste("Number of DOMAIN levels:", length(levels(long_data_clean$DOMAIN))))
|
||
print(paste("Number of TEMPORAL_DO levels:", length(levels(long_data_clean$TEMPORAL_DO))))
|
||
|
||
# Check for complete cases
|
||
complete_cases <- long_data_clean[complete.cases(long_data_clean), ]
|
||
print(paste("Complete cases:", nrow(complete_cases), "out of", nrow(long_data_clean)))
|
||
|
||
# Check if design is balanced
|
||
design_balance <- table(long_data_clean$pID, long_data_clean$TIME, long_data_clean$DOMAIN)
|
||
|
||
print(summary(as.vector(design_balance)))
|
||
|
||
# Check for any participants with missing combinations
|
||
missing_combos <- long_data_clean %>%
|
||
group_by(pID) %>%
|
||
summarise(
|
||
n_combinations = n(),
|
||
expected_combinations = 8, # 2 TIME × 4 DOMAIN = 8
|
||
missing_combinations = 8 - n_combinations,
|
||
.groups = 'drop'
|
||
)
|
||
|
||
print("Missing combinations per participant:")
|
||
print(missing_combos[missing_combos$missing_combinations > 0, ])
|
||
|
||
# Mixed ANOVA using aov()
|
||
# Between-subjects: TEMPORAL_DO (2 levels: 01PAST, 02FUT)
|
||
# Within-subjects: TIME (2 levels: Past, Future) × DOMAIN (4 levels: Preferences, Personality, Values, Life)
|
||
|
||
mixed_anova_model <- aov(MEAN_DIFFERENCE ~ TEMPORAL_DO * TIME * DOMAIN + Error(pID/(TIME * DOMAIN)),
|
||
data = long_data_clean)
|
||
|
||
print("Mixed ANOVA Results:")
|
||
print(summary(mixed_anova_model))
|
||
|
||
# Extract effect sizes (generalized eta squared)
|
||
# For aov() objects, we need to extract from the summary
|
||
anova_summary <- summary(mixed_anova_model)
|
||
|
||
# Effect sizes will be calculated separately
|
||
|
||
# =============================================================================
|
||
# ALTERNATIVE: MIXED MODEL USING LMER (FASTER)
|
||
# =============================================================================
|
||
|
||
# Load required libraries for mixed models
|
||
library(lme4)
|
||
library(lmerTest)
|
||
|
||
print("Running alternative mixed model using lmer()...")
|
||
start_time_lmer <- Sys.time()
|
||
|
||
# Mixed model approach (much faster than aov with complex error structures)
|
||
mixed_lmer_model <- lmer(MEAN_DIFFERENCE ~ TEMPORAL_DO * TIME * DOMAIN +
|
||
(1|pID) + (1|pID:TIME) + (1|pID:DOMAIN),
|
||
data = long_data_clean)
|
||
|
||
end_time_lmer <- Sys.time()
|
||
print(paste("lmer model completed in:", round(end_time_lmer - start_time_lmer, 2), "seconds"))
|
||
|
||
print("Mixed Model Results (lmer):")
|
||
print(summary(mixed_lmer_model))
|
||
|
||
# ANOVA table for lmer model
|
||
print("ANOVA Table for Mixed Model:")
|
||
print(anova(mixed_lmer_model))
|
||
|
||
# =============================================================================
|
||
# POST-HOC COMPARISONS
|
||
# =============================================================================
|
||
|
||
# Post-hoc comparisons using emmeans
|
||
print("\n=== POST-HOC COMPARISONS ===")
|
||
|
||
# Main effect of TIME
|
||
print("Main Effect of TIME:")
|
||
time_emmeans <- emmeans(mixed_anova_model, ~ TIME)
|
||
time_contrasts <- pairs(time_emmeans, adjust = "bonferroni")
|
||
print(time_contrasts)
|
||
|
||
# Main effect of DOMAIN
|
||
print("\nMain Effect of DOMAIN:")
|
||
domain_emmeans <- emmeans(mixed_anova_model, ~ DOMAIN)
|
||
domain_contrasts <- pairs(domain_emmeans, adjust = "bonferroni")
|
||
print(domain_contrasts)
|
||
|
||
# Main effect of TEMPORAL_DO
|
||
print("\nMain Effect of TEMPORAL_DO:")
|
||
temporal_emmeans <- emmeans(mixed_anova_model, ~ TEMPORAL_DO)
|
||
temporal_contrasts <- pairs(temporal_emmeans, adjust = "bonferroni")
|
||
print(temporal_contrasts)
|
||
|
||
# Two-way interactions
|
||
print("\nTIME × DOMAIN Interaction:")
|
||
time_domain_emmeans <- emmeans(mixed_anova_model, ~ TIME * DOMAIN)
|
||
time_domain_contrasts <- pairs(time_domain_emmeans, by = "TIME", adjust = "bonferroni")
|
||
print(time_domain_contrasts)
|
||
|
||
print("\nTEMPORAL_DO × TIME Interaction:")
|
||
temporal_time_emmeans <- emmeans(mixed_anova_model, ~ TEMPORAL_DO * TIME)
|
||
temporal_time_contrasts <- pairs(temporal_time_emmeans, by = "TEMPORAL_DO", adjust = "bonferroni")
|
||
print(temporal_time_contrasts)
|
||
|
||
print("\nTEMPORAL_DO × DOMAIN Interaction:")
|
||
temporal_domain_emmeans <- emmeans(mixed_anova_model, ~ TEMPORAL_DO * DOMAIN)
|
||
temporal_domain_contrasts <- pairs(temporal_domain_emmeans, by = "TEMPORAL_DO", adjust = "bonferroni")
|
||
print(temporal_domain_contrasts)
|
||
|
||
# Three-way interaction
|
||
print("\nTEMPORAL_DO × TIME × DOMAIN Interaction:")
|
||
three_way_emmeans <- emmeans(mixed_anova_model, ~ TEMPORAL_DO * TIME * DOMAIN)
|
||
three_way_contrasts <- pairs(three_way_emmeans, by = c("TEMPORAL_DO", "TIME"), adjust = "bonferroni")
|
||
print(three_way_contrasts)
|
||
|