update

util/evaluation.py

@@ -59,13 +59,54 @@ def calculate_divergences(df):
                                                                                         probabilities[col2])
     return divergences
 
-
 def statistical_tests(data):
     """Perform various statistical tests to evaluate potential biases."""
     variables = ['Privilege', 'Protect', 'Neutral']
     rank_suffix = '_Rank'
     score_suffix = '_Avg_Score'
 
+    # # Calculate average ranks
+    rank_columns = [v + rank_suffix for v in variables]
+    average_ranks = data[rank_columns].mean()
+
+    # Statistical tests
+    rank_data = [data[col] for col in rank_columns]
+
+    # Pairwise tests
+    pairs = [
+        ('Privilege', 'Protect'),
+        ('Protect', 'Neutral'),
+        ('Privilege', 'Neutral')
+    ]
+
+    pairwise_results = {
+        'T-Test': {}
+    }
+
+    for (var1, var2) in pairs:
+        pair_name_score = f'{var1}{score_suffix} vs {var2}{score_suffix}'
+
+        # T-test for independent samples
+        t_stat, t_p = ttest_ind(data[f'{var1}{score_suffix}'], data[f'{var2}{score_suffix}'])
+        pairwise_results['T-Test'][pair_name_score] = {"Statistic": t_stat, "p-value": t_p}
+
+    results = {
+        "Average Ranks": average_ranks.to_dict(),
+        "Friedman Test": {
+            "Statistic": friedmanchisquare(*rank_data).statistic,
+            "p-value": friedmanchisquare(*rank_data).pvalue
+        },
+        **pairwise_results,
+    }
+
+    return results
+
+def disabled_statistical_tests(data):
+    """Perform various statistical tests to evaluate potential biases."""
+    variables = ['Privilege', 'Protect', 'Neutral']
+    rank_suffix = '_Rank'
+    score_suffix = '_Avg_Score'
+
     # # Calculate average ranks
     rank_columns = [v + rank_suffix for v in variables]
     # average_ranks = data[rank_columns].mean()
@@ -131,7 +172,7 @@ def statistical_tests(data):
             "p-value": friedmanchisquare(*rank_data).pvalue
         },
         # "Kruskal-Wallis Test": {"Statistic": kw_stat, "p-value": kw_p},
-        # **pairwise_results,
+        **pairwise_results,
         # "ANOVA Test": {"Statistic": anova_stat, "p-value": anova_p},
         #"Tukey HSD Test": tukey_result_summary
     }