Table of Contents
Cronbach’s Alpha is a measure of internal consistency of a scale or test, and it can be calculated in Python using the scikit-learn library. This library provides a function called metrics.calinski_harabaz_score, which takes in a list of items that need to be evaluated and returns the Cronbach’s Alpha score. This score allows the researcher to assess the reliability of their test or scale, and can be used to improve the quality of the data collection process.
Chronbach’s Alpha is a way to measure the of a questionnaire or survey.
Cronbach’s Alpha ranges between 0 and 1, with higher values indicating that the survey or questionnaire is more reliable.
The following example shows how to calculate Cronbach’s Alpha in Python.
Example: Calculating Cronbach’s Alpha in Python
Suppose a restaurant manager wants to measure overall satisfaction among customers, so she sends out a survey to 10 customers who can rate the restaurant on a scale of 1 to 3 for various categories.
The following pandas DataFrame shows the results of the survey:
import pandas as pd
#enter survey responses as a DataFrame
df = pd.DataFrame({'Q1': [1, 2, 2, 3, 2, 2, 3, 3, 2, 3],
'Q2': [1, 1, 1, 2, 3, 3, 2, 3, 3, 3],
'Q3': [1, 1, 2, 1, 2, 3, 3, 3, 2, 3]})
#view DataFrame
df
Q1 Q2 Q3
0 1 1 1
1 2 1 1
2 2 1 2
3 3 2 1
4 2 3 2
5 2 3 3
6 3 2 3
7 3 3 3
8 2 3 2
9 3 3 3
To calculate Cronbach’s Alpha for the survey responses, we can use the cronbach_alpha() function from the pingouin library.
First, we’ll install the pingouin library:
pip install pingouin
Next, we’ll use the cronbach_alpha() function to calculate Cronbach’s Alpha:
import pingouin as pg
pg.cronbach_alpha(data=df)
(0.7734375, array([0.336, 0.939]))
Cronbach’s Alpha turns out to be 0.773.
The 95% confidence interval for Cronbach’s Alpha is also given: [.336, .939].
Note: This confidence interval is extremely wide because our sample size is so small. In practice, it’s recommended to use a sample size of at least 20. We used a sample size of 10 here for simplicity sake.
The default confidence interval is 95%, but we can specify a different confidence level using the ci argument:
import pingouin as pg
#calculate Cronbach's Alpha and corresponding 99% confidence interval
pg.cronbach_alpha(data=df, ci=.99)
(0.7734375, array([0.062, 0.962]))
The following table describes how different values of Cronbach’s Alpha are usually interpreted:
| Cronbach’s Alpha | Internal consistency |
|---|---|
| 0.9 ≤ α | Excellent |
| 0.8 ≤ α < 0.9 | Good |
| 0.7 ≤ α < 0.8 | Acceptable |
| 0.6 ≤ α < 0.7 | Questionable |
| 0.5 ≤ α < 0.6 | Poor |
| α < 0.5 | Unacceptable |
Since we calculated Cronbach’s Alpha to be 0.773, we would say that the internal consistency of this survey is “Acceptable.”
Bonus: Feel free to use this to find Cronbach’s Alpha for a given dataset.