Chi-square test

Analyzing Survey Data in Python

EbunOluwa Andrew

Data Scientist

Chi-square test

  • Inferences about categorical variable distribution
    • Compares observed observations to expected observations

Dice photo by Edge2Edge Media on Unsplash

Analyzing Survey Data in Python

Chi-square test in survey analysis

  • Decide relationship between two categorical variables of a population
  • $H_{o}$ = no relationship between variables
  • $H_{a}$ = relationship between variables
  • P-value
    • if significant (<0.05), reject null hypothesis
    • if insignificant (>0.05), accept null hypothesis
Analyzing Survey Data in Python

Why use chi-square testing in survey analysis

  • Input variables relevant to output variable

  • Understand impact of different variables on population

  • Check if differences are by chance or statistically significant

Survey results photo by Firmbee.com on Unsplash

1 Photo by Firmbee.com on Unsplash
Analyzing Survey Data in Python

Assumptions of chi-square test on survey analysis

  • Both variables = categorical
  • Sample randomly selected from population
  • Sample size > 100
  • Expected frequencies >=5
Analyzing Survey Data in Python

Survey data for chi-square analysis

pet_type current_pets time_spent reduces_stress
dog 1 420 yes
dog 1 180 yes
dog 4 30 yes
dog 1 30 yes
dog 1 60 yes
Analyzing Survey Data in Python

Survey data for chi-square analysis

  • Sample size >100
  • Two categorical variables:
    • pet_type
    • reduces_stress
  • $$H_{o} $$ NO relationship between the type of pet owned by pet owners and their perceived reduced stress
  • $$H_{a}$$ relationship between the type of pet owned by pet owners and their perceived reduced stress
Analyzing Survey Data in Python

Steps of chi-square analysis on pet_survey in python

import pandas as pd
import scipy.stats as st
data = pd.read_csv('pet_survey.csv') 

cross_table = pd.crosstab(data.reduces_stress, data.pet_type)

chi_analysis = st.chi2_contingency(cross_table)

print(chi_analysis)

|--------------------------|
| (67.7,                   | 
| 1.9e-16,                 |
| 1,                       |
| array([[1767.0, 1825.0], |
| [2251.0, 2325.0]]))      |
Analyzing Survey Data in Python

Result and interpretation of pet_survey

  • Frequencies >= 5

    • Valid results

  • p-value < 0.05

    • reject null hypothesis
    • pet_owned and reduces_stress are related

  • Type of pet owned has an effect on whether pet owners perceive stress reduction

chi-squared test results

Analyzing Survey Data in Python

Let's practice!

Analyzing Survey Data in Python

Preparing Video For Download...