analysis_mmob.py

import pandas as pd
from scipy.stats import ttest_ind
import numpy as np
from scipy.stats import chi2_contingency

# Replace 'your_file.csv' with the path to your CSV file
df = pd.read_csv('results.csv')

initial = [15] # Interaction order (SELECT WHICH ONE WAS FIRST)
tilt_vs_swipe = [42, 43] # How likely would you like to use to use a system using tilt vs swipe
overall = [41] # Overall preference for tilt vs swipe - 1 = tilt, 2 = swipe, 3 = no pref // Which interaction did you prefer overall?
swipe_confidence = [20]
swipe_informed = [21]
tilt_confidence = [31]
tilt_informed = [32]

pstart = 2
pend = 17

initial_group_series = df.iloc[pstart:pend, initial]
# Indices where initial_group == 1 (swipe first) and == 2 (tilt first)
swipe_first_idx = initial_group_series[initial_group_series['Q32'] == '1'].index
tilt_first_idx = initial_group_series[initial_group_series['Q32'] == '2'].index

# Separate data based on initial_group
swipe_first_data = df.loc[swipe_first_idx]
tilt_first_data = df.loc[tilt_first_idx]

def ttest(column):
    # Get the column name for tilt confidence
    tilt_conf_col = df.columns[column[0]]
    # Set the first row as the header with question texts
    print("T-test for question", tilt_conf_col , ":",df[tilt_conf_col][0])


    # Convert to numeric, coerce errors to NaN
    swipe_first_tilt_conf = pd.to_numeric(swipe_first_data[tilt_conf_col], errors='coerce')
    tilt_first_tilt_conf = pd.to_numeric(tilt_first_data[tilt_conf_col], errors='coerce')

    # Drop NaN values for the t-test
    swipe_first_tilt_conf_clean = swipe_first_tilt_conf.dropna()
    tilt_first_tilt_conf_clean = tilt_first_tilt_conf.dropna()

    # Calculate means
    mean_swipe_first = swipe_first_tilt_conf_clean.mean()
    mean_tilt_first = tilt_first_tilt_conf_clean.mean()

    # Perform independent t-test
    t_stat, p_val = ttest_ind(swipe_first_tilt_conf_clean, tilt_first_tilt_conf_clean, equal_var=True)

    print(f"Mean (Started with Swiping): {mean_swipe_first:.2f}")
    print(f"Mean (Started with Tilting): {mean_tilt_first:.2f}")
    print(f"T-test: t={t_stat:.3f}, p={p_val:.3f}")
    print("-" * 40)

ttest(tilt_confidence)
ttest(swipe_confidence)
ttest(tilt_informed)
ttest(swipe_informed)


# Chi-Square test: Is Initial_Group associated with Overall_Preference?
# Get the column names
initial_group_col = df.columns[initial[0]]
overall_pref_col = df.columns[overall[0]]
print(f"\nChi-Square Test for {initial_group_col} vs {overall_pref_col}")

# Only use valid rows (within pstart:pend)
subset = df.iloc[pstart:pend]

# Build contingency table
contingency_table = pd.crosstab(subset[initial_group_col], subset[overall_pref_col])

# Run chi-square test
chi2, p, dof, expected = chi2_contingency(contingency_table)

print("Contingency Table:")
print(contingency_table)
print(f"\nChi-square test: x²={chi2:.3f}, p={p:.3f}, dof={dof}")