# 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

# Read the data
data <- read.csv("eohi1/exp1.csv")

# Display basic information about the dataset
cat("Dataset dimensions:", dim(data), "\n")
cat("Number of participants:", length(unique(data$pID)), "\n")

# Check experimental conditions
cat("\nExperimental conditions:\n")
print(table(data$GROUP, data$TEMPORAL_DO, data$ITEM_DO))

# Check what domain mean columns are available
cat("\nChecking available domain mean columns:\n")
domain_mean_cols <- colnames(data)[grepl("mean_(pref|pers|val|life)", colnames(data))]
print(domain_mean_cols)

# 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) {
  cat("Warning: Missing variables:", paste(missing_vars, collapse = ", "), "\n")
} else {
  cat("All required domain mean variables found!\n")
}

# =============================================================================
# STEP 1: DATA PIVOTING TO LONG FORMAT
# =============================================================================

cat("STEP 1: DATA PIVOTING TO LONG FORMAT\n")

# 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
)

cat("Domain mapping:\n")
print(domain_mapping)

# Function to pivot data to long format
pivot_domain_means <- function(data, domain_mapping) {
  long_data <- data.frame()
  
  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)) {
      cat("Warning: Variable", var_name, "not found in data\n")
      next
    }
    
    # Create subset for this variable
    subset_data <- data[, c("pID", "ResponseId", "GROUP", "TEMPORAL_DO", "ITEM_DO", var_name)]
    subset_data$TIME <- time_level
    subset_data$DOMAIN <- domain_level
    subset_data$MEAN_DIFFERENCE <- subset_data[[var_name]]
    subset_data[[var_name]] <- NULL  # Remove original column
    
    # Add to long data
    long_data <- rbind(long_data, subset_data)
  }
  
  # 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$GROUP <- as.factor(long_data$GROUP)
  long_data$TEMPORAL_DO <- as.factor(long_data$TEMPORAL_DO)
  long_data$ITEM_DO <- as.factor(long_data$ITEM_DO)
  
  return(long_data)
}

# Pivot data to long format
cat("\nPivoting data to long format...\n")
tryCatch({
  long_data <- pivot_domain_means(data, domain_mapping)
  cat("Data pivoting completed successfully.\n")
}, error = function(e) {
  cat("Error in data pivoting:", e$message, "\n")
  stop("Cannot proceed without proper data structure")
})

cat("Long format data dimensions:", dim(long_data), "\n")
cat("Unique participants:", length(unique(long_data$pID)), "\n")
cat("TIME levels:", paste(levels(long_data$TIME), collapse = ", "), "\n")
cat("DOMAIN levels:", paste(levels(long_data$DOMAIN), collapse = ", "), "\n")

# Check data types
cat("\nData types check:\n")
cat("TIME is factor:", is.factor(long_data$TIME), "\n")
cat("DOMAIN is factor:", is.factor(long_data$DOMAIN), "\n")
cat("pID is factor:", is.factor(long_data$pID), "\n")
cat("MEAN_DIFFERENCE is numeric:", is.numeric(long_data$MEAN_DIFFERENCE), "\n")

# Display structure and sample
cat("\nLong data structure:\n")
str(long_data)

cat("\nFirst 10 rows of long_data:\n")
print(utils::head(long_data, 10))

# Show example data for one participant
cat("\nExample: Participant 1 across all domains and times:\n")
participant_1_data <- long_data[long_data$pID == 1, c("pID", "GROUP", "TEMPORAL_DO", "ITEM_DO", "TIME", "DOMAIN", "MEAN_DIFFERENCE")]
print(participant_1_data)

# =============================================================================
# STEP 2: ASSUMPTION CHECKING
# =============================================================================


# 2.1 Check for missing values
cat("\n2.1 Missing Values Check:\n")
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'
  )

cat("Missing values by TIME and DOMAIN:\n")
print(missing_summary)

# Remove missing values
long_data_clean <- long_data[!is.na(long_data$MEAN_DIFFERENCE), ]
cat("\nData after removing missing values:", dim(long_data_clean), "\n")

# 2.2 Outlier detection
cat("\n2.2 Outlier Detection:\n")
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'
  )

cat("Outlier summary (IQR method):\n")
print(outlier_summary)

# 2.3 Normality tests
cat("\n2.3 Normality Tests:\n")
normality_results <- long_data_clean %>%
  group_by(TIME, DOMAIN) %>%
  summarise(
    n = n(),
    shapiro_p = ifelse(n >= 3 & n <= 5000, 
                      shapiro.test(MEAN_DIFFERENCE)$p.value, 
                      NA),
    anderson_p = ifelse(n >= 7, 
                       ad.test(MEAN_DIFFERENCE)$p.value, 
                       NA),
    .groups = 'drop'
  ) %>%
  mutate(
    shapiro_normal = shapiro_p > 0.05,
    anderson_normal = anderson_p > 0.05,
    overall_normal = case_when(
      !is.na(shapiro_p) & !is.na(anderson_p) ~ shapiro_normal & anderson_normal,
      !is.na(shapiro_p) ~ shapiro_normal,
      !is.na(anderson_p) ~ anderson_normal,
      TRUE ~ NA
    )
  )

cat("Normality test results:\n")
print(normality_results)

# 2.4 Homogeneity of variance (Levene's test)
cat("\n2.4 Homogeneity of Variance Tests:\n")

# Test homogeneity across TIME within each DOMAIN
homogeneity_time <- long_data_clean %>%
  group_by(DOMAIN) %>%
  summarise(
    levene_p = leveneTest(MEAN_DIFFERENCE ~ TIME)$`Pr(>F)`[1],
    homogeneous = levene_p > 0.05,
    .groups = 'drop'
  )

cat("Homogeneity of variance across TIME within each DOMAIN:\n")
print(homogeneity_time)

# Test homogeneity across DOMAIN within each TIME
homogeneity_domain <- long_data_clean %>%
  group_by(TIME) %>%
  summarise(
    levene_p = leveneTest(MEAN_DIFFERENCE ~ DOMAIN)$`Pr(>F)`[1],
    homogeneous = levene_p > 0.05,
    .groups = 'drop'
  )

cat("Homogeneity of variance across DOMAIN within each TIME:\n")
print(homogeneity_domain)

# =============================================================================
# STEP 3: DESCRIPTIVE STATISTICS
# =============================================================================

cat("\n", paste(rep("=", 80), collapse = ""), "\n")
cat("STEP 3: DESCRIPTIVE STATISTICS\n")
cat(paste(rep("=", 80), collapse = ""), "\n")

# Overall descriptive statistics
desc_stats <- long_data_clean %>%
  group_by(TIME, DOMAIN) %>%
  summarise(
    n = n(),
    mean = round(mean(MEAN_DIFFERENCE), 5),
    sd = round(sd(MEAN_DIFFERENCE), 5),
    median = round(median(MEAN_DIFFERENCE), 5),
    q1 = round(quantile(MEAN_DIFFERENCE, 0.25), 5),
    q3 = round(quantile(MEAN_DIFFERENCE, 0.75), 5),
    min = round(min(MEAN_DIFFERENCE), 5),
    max = round(max(MEAN_DIFFERENCE), 5),
    .groups = 'drop'
  )

cat("Descriptive statistics by TIME and DOMAIN:\n")
print(desc_stats)

# Descriptive statistics by between-subjects factors
desc_stats_by_group <- long_data_clean %>%
  group_by(GROUP, TIME, DOMAIN) %>%
  summarise(
    n = n(),
    mean = round(mean(MEAN_DIFFERENCE), 5),
    sd = round(sd(MEAN_DIFFERENCE), 5),
    .groups = 'drop'
  )

cat("\nDescriptive statistics by GROUP, TIME, and DOMAIN:\n")
print(desc_stats_by_group)

# =============================================================================
# STEP 4: MIXED ANOVA ANALYSES
# =============================================================================

cat("\n", paste(rep("=", 80), collapse = ""), "\n")
cat("STEP 4: MIXED ANOVA ANALYSES\n")
cat(paste(rep("=", 80), collapse = ""), "\n")

# 4.1 Main Mixed ANOVA
cat("\n4.1 Main Mixed ANOVA:\n")
cat("Within-subjects factors: TIME, DOMAIN\n")
cat("Between-subjects factors: GROUP, TEMPORAL_DO, ITEM_DO\n")
cat(paste(rep("-", 50), collapse = ""), "\n")

tryCatch({
  main_anova <- ezANOVA(
    data = long_data_clean,
    dv = MEAN_DIFFERENCE,
    wid = pID,
    within = c(TIME, DOMAIN),
    between = c(GROUP, TEMPORAL_DO, ITEM_DO),
    type = 3,
    detailed = TRUE,
    return_aov = TRUE
  )
  
  cat("Main ANOVA Results:\n")
  print(main_anova)
  
  # Check sphericity
  if (!is.null(main_anova$`Mauchly's Test for Sphericity`)) {
    cat("\nSphericity test results:\n")
    print(main_anova$`Mauchly's Test for Sphericity`)
  }
  
}, error = function(e) {
  cat("Error in main ANOVA:", e$message, "\n")
  
  # Try simpler model without all between-subjects factors
  cat("Attempting simpler model with only GROUP as between-subjects factor...\n")
  tryCatch({
    simple_anova <- ezANOVA(
      data = long_data_clean,
      dv = MEAN_DIFFERENCE,
      wid = pID,
      within = c(TIME, DOMAIN),
      between = GROUP,
      type = 3,
      detailed = TRUE,
      return_aov = TRUE
    )
    
    cat("Simplified ANOVA Results:\n")
    print(simple_anova)
    
    main_anova <<- simple_anova
    
  }, error = function(e2) {
    cat("Simplified ANOVA also failed:", e2$message, "\n")
  })
})

# 4.2 Domain-specific analyses
cat("\n4.2 Domain-specific Mixed ANOVAs:\n")
cat(paste(rep("-", 50), collapse = ""), "\n")

domain_results <- list()

for (domain in levels(long_data_clean$DOMAIN)) {
  cat("\nAnalyzing domain:", domain, "\n")
  
  domain_data <- long_data_clean[long_data_clean$DOMAIN == domain, ]
  
  tryCatch({
    domain_anova <- ezANOVA(
      data = domain_data,
      dv = MEAN_DIFFERENCE,
      wid = pID,
      within = TIME,
      between = c(GROUP, TEMPORAL_DO, ITEM_DO),
      type = 3,
      detailed = TRUE
    )
    
    cat("ANOVA results for", domain, ":\n")
    print(domain_anova)
    
    domain_results[[domain]] <- domain_anova
    
  }, error = function(e) {
    cat("Error in ANOVA for", domain, ":", e$message, "\n")
    
    # Fallback to simpler analysis
    cat("Attempting simpler repeated measures ANOVA...\n")
    tryCatch({
      simple_anova <- ezANOVA(
        data = domain_data,
        dv = MEAN_DIFFERENCE,
        wid = pID,
        within = TIME,
        between = GROUP,
        type = 3,
        detailed = TRUE
      )
      print(simple_anova)
      domain_results[[domain]] <- simple_anova
    }, error = function(e2) {
      cat("Simple ANOVA also failed:", e2$message, "\n")
    })
  })
}

# 4.3 Time-specific analyses
cat("\n4.3 Time-specific Mixed ANOVAs:\n")
cat(paste(rep("-", 50), collapse = ""), "\n")

time_results <- list()

for (time in levels(long_data_clean$TIME)) {
  cat("\nAnalyzing time:", time, "\n")
  
  time_data <- long_data_clean[long_data_clean$TIME == time, ]
  
  tryCatch({
    time_anova <- ezANOVA(
      data = time_data,
      dv = MEAN_DIFFERENCE,
      wid = pID,
      within = DOMAIN,
      between = c(GROUP, TEMPORAL_DO, ITEM_DO),
      type = 3,
      detailed = TRUE
    )
    
    cat("ANOVA results for", time, ":\n")
    print(time_anova)
    
    time_results[[time]] <- time_anova
    
  }, error = function(e) {
    cat("Error in ANOVA for", time, ":", e$message, "\n")
    
    # Fallback to simpler analysis
    cat("Attempting simpler repeated measures ANOVA...\n")
    tryCatch({
      simple_anova <- ezANOVA(
        data = time_data,
        dv = MEAN_DIFFERENCE,
        wid = pID,
        within = DOMAIN,
        between = GROUP,
        type = 3,
        detailed = TRUE
      )
      print(simple_anova)
      time_results[[time]] <- simple_anova
    }, error = function(e2) {
      cat("Simple ANOVA also failed:", e2$message, "\n")
    })
  })
}

# =============================================================================
# STEP 5: POST-HOC ANALYSES
# =============================================================================

cat("\n", paste(rep("=", 80), collapse = ""), "\n")
cat("STEP 5: POST-HOC ANALYSES\n")
cat(paste(rep("=", 80), collapse = ""), "\n")

# 5.1 Pairwise comparisons for significant effects
if (exists("main_anova") && !is.null(main_anova)) {
  cat("\n5.1 Post-hoc comparisons for main effects:\n")
  
  # Check for significant main effects and interactions
  anova_table <- main_anova$ANOVA
  
  if ("TIME" %in% anova_table$Effect && anova_table$p[anova_table$Effect == "TIME"] < 0.05) {
    cat("Significant TIME main effect found. Computing pairwise comparisons...\n")
    
    # Simple paired t-tests for TIME effect
    past_means <- long_data_clean$MEAN_DIFFERENCE[long_data_clean$TIME == "Past"]
    future_means <- long_data_clean$MEAN_DIFFERENCE[long_data_clean$TIME == "Future"]
    
    if (length(past_means) == length(future_means)) {
      time_t_test <- t.test(past_means, future_means, paired = TRUE)
      cat("Paired t-test for TIME effect:\n")
      cat("t =", round(time_t_test$statistic, 5), 
          ", df =", time_t_test$parameter, 
          ", p =", round(time_t_test$p.value, 5), "\n")
      cat("Mean difference (Past - Future):", round(time_t_test$estimate, 5), "\n")
    }
  }
  
  if ("DOMAIN" %in% anova_table$Effect && anova_table$p[anova_table$Effect == "DOMAIN"] < 0.05) {
    cat("Significant DOMAIN main effect found.\n")
    
    # Pairwise comparisons between domains
    domain_means <- long_data_clean %>%
      group_by(DOMAIN) %>%
      summarise(mean_diff = mean(MEAN_DIFFERENCE), .groups = 'drop')
    
    cat("Domain means:\n")
    print(domain_means)
  }
  
  if ("TIME:DOMAIN" %in% anova_table$Effect && anova_table$p[anova_table$Effect == "TIME:DOMAIN"] < 0.05) {
    cat("Significant TIME × DOMAIN interaction found.\n")
    
    # Simple effects analysis
    interaction_means <- long_data_clean %>%
      group_by(TIME, DOMAIN) %>%
      summarise(mean_diff = mean(MEAN_DIFFERENCE), .groups = 'drop')
    
    cat("TIME × DOMAIN interaction means:\n")
    print(interaction_means)
  }
}

# =============================================================================
# STEP 6: EFFECT SIZES
# =============================================================================

cat("\n", paste(rep("=", 80), collapse = ""), "\n")
cat("STEP 6: EFFECT SIZES\n")
cat(paste(rep("=", 80), collapse = ""), "\n")

if (exists("main_anova") && !is.null(main_anova)) {
  anova_table <- main_anova$ANOVA
  
  # Calculate partial eta squared for each effect
  anova_table$partial_eta_squared <- round(anova_table$SSn / (anova_table$SSn + anova_table$SSd), 5)
  
  cat("Effect sizes (partial eta squared):\n")
  effect_sizes <- anova_table[, c("Effect", "partial_eta_squared")]
  print(effect_sizes)
}

# =============================================================================
# STEP 7: SUMMARY AND INTERPRETATION
# =============================================================================

cat("\n", paste(rep("=", 80), collapse = ""), "\n")
cat("STEP 7: SUMMARY AND INTERPRETATION\n")
cat(paste(rep("=", 80), collapse = ""), "\n")

cat("Analysis Summary:\n")
cat("- Total participants:", length(unique(long_data_clean$pID)), "\n")
cat("- Total observations:", nrow(long_data_clean), "\n")
cat("- Within-subjects factors: TIME (Past vs Future), DOMAIN (Preferences, Personality, Values, Life)\n")
cat("- Between-subjects factors: GROUP, TEMPORAL_DO, ITEM_DO\n")
cat("- Dependent variable: Mean absolute differences in domain ratings\n")

cat("\nResearch Question:\n")
cat("Do participants rate changes in domains differently from past to now vs past to future?\n")

if (exists("main_anova") && !is.null(main_anova)) {
  anova_table <- main_anova$ANOVA
  
  cat("\nKey Findings:\n")
  
  # Check for significant effects
  significant_effects <- anova_table$Effect[anova_table$p < 0.05]
  
  if (length(significant_effects) > 0) {
    cat("Significant effects found:\n")
    for (effect in significant_effects) {
      p_val <- anova_table$p[anova_table$Effect == effect]
      cat("-", effect, "(p =", round(p_val, 5), ")\n")
    }
  } else {
    cat("No significant effects found at α = 0.05\n")
  }
  
  # Interpret TIME effect
  if ("TIME" %in% anova_table$Effect) {
    time_p <- anova_table$p[anova_table$Effect == "TIME"]
    if (time_p < 0.05) {
      cat("\nTIME Effect: Participants show different levels of change when comparing\n")
      cat("past-to-now vs past-to-future perspectives (p =", round(time_p, 5), ")\n")
    } else {
      cat("\nTIME Effect: No significant difference between past-to-now and past-to-future\n")
      cat("perspectives (p =", round(time_p, 5), ")\n")
    }
  }
  
  # Interpret DOMAIN effect
  if ("DOMAIN" %in% anova_table$Effect) {
    domain_p <- anova_table$p[anova_table$Effect == "DOMAIN"]
    if (domain_p < 0.05) {
      cat("\nDOMAIN Effect: Different domains show different levels of perceived change\n")
      cat("(p =", round(domain_p, 5), ")\n")
    } else {
      cat("\nDOMAIN Effect: No significant differences between domains in perceived change\n")
      cat("(p =", round(domain_p, 5), ")\n")
    }
  }
  
  # Interpret interaction
  if ("TIME:DOMAIN" %in% anova_table$Effect) {
    interaction_p <- anova_table$p[anova_table$Effect == "TIME:DOMAIN"]
    if (interaction_p < 0.05) {
      cat("\nTIME × DOMAIN Interaction: The effect of time perspective on perceived change\n")
      cat("varies across domains (p =", round(interaction_p, 5), ")\n")
    } else {
      cat("\nTIME × DOMAIN Interaction: No significant interaction between time perspective\n")
      cat("and domain (p =", round(interaction_p, 5), ")\n")
    }
  }
}

cat("\n", paste(rep("=", 80), collapse = ""), "\n")
cat("ANALYSIS COMPLETE!\n")
cat(paste(rep("=", 80), collapse = ""), "\n")