Independent samples t-test, multiple populations

Here we expand on the example with independent samples t-test and incorporate itertool.combinations()to test several populations against eachother in an effective way. We use the same dataset as in the previous example on independent samples t-test but slice the dataset by gaming platform instead of gender.

In [ ]:
import pandas as pd
from itertools import combinations
from scipy.stats import ttest_ind

# Import data
df = pd.read_csv('Gaming and Mental Health.csv')

# Extract unique entries in relevant column
populations = df['gaming_platform'].unique()

# Create a list of tuples with all possible combinations of 2 elements
possible_combinations = list(combinations(populations, 2))

# Create empty lists, we append results from each iteration here in order to be able to create a summary in the end
pop_1 = []
pop_2 = []
t_stat = []
p_value = []

# Iterate over the list of combinations
for combination in possible_combinations:

    # Perform t-test
    t, p = ttest_ind(df[df['gaming_platform']==combination[0]]['daily_gaming_hours'], df[df['gaming_platform']==combination[1]]['daily_gaming_hours'], equal_var=False)
    
    # Append data
    pop_1.append(combination[0])
    pop_2.append(combination[1])
    t_stat.append(t)
    p_value.append(p)

# Create a df from the lists with the results
df_summary = pd.DataFrame({'Population 1': pop_1,
                           'Population 2': pop_2,
                           't-stat': t_stat,
                           'p-value': p_value})

# Take a peek a the result, round to two digits for readability
print(df_summary.round(2))

     Population 1    Population 2  t-stat  p-value
0              PC  Multi-platform   -0.60     0.55
1              PC         Console   -0.85     0.40
2              PC          Mobile    0.84     0.40
3  Multi-platform         Console   -0.24     0.81
4  Multi-platform          Mobile    1.41     0.16
5         Console          Mobile    1.66     0.10

Some more reading: https://www.geeksforgeeks.org/maths/t-test-in-statistics/

This tutorial uses the “Gaming and Mental Health” licensed under the Apache License 2.0. © Shaista Shahid. License: http://www.apache.org/licenses/LICENSE-2.0 source: https://www.kaggle.com/datasets/shaistashahid/gaming-and-mental-health?resource=download