Chi-Square Test
Chi-square statistic (
- used to measure how the model compares to the actual data
Chi-square test is a test used to determine if a difference exists between groups.
- produces a chi-square statistic
- used to compare all sorts of variables:
- scores on a test
- responses to a medical test
- answers to a survey question
- and more
- reasons to conduct chi-square analysis:
- A chi-square test is used to compare actual results to what we expected the results would be
- A chi-square test also allows us to rule out that the observations happened by chance
- Chi-square testing identifies how confident we are that the results are (or are not) different from what we expected and that there is a relationship between the variables
- Chi-square testing is useful when we are analyzing data from a random sample and working with a categorical variable, like education, race, or gender
Types of Chi-Square Tests
- 2 common types of chi-square tests:
- test of independence
- goodness of fit
- both tests compare actual results against an expectation
Test of independence is a chi-square test that tests against multiple variables to analyze the distribution of the data.
- Used for the relationship between the variables
- tests against multiple variables
- evaluates whether there is a significant association between two categorical variables
Goodness of fit is a chi-square test that tests against a single variable to analyze the relationship between variables.
- Used for the distribution of the data
- tests against a single variable
- tests the hypothesized distribution and then provides the significance value to make a conclusion about whether or not it met what we expected
Example Output of a Chi-Square Test
Student Preparedness Fail Pass Total Very Prepared 9 (8.32) [0.06] 17 (17.68) [0.03] 26 Somewhat Prepared 11 (16.32) [1.73] 40 (34.68) [0.82] 51 Not Prepared 12 (7.36) [2.93] 11 (15.64) [1.38] 23 Total 32 68 100 The chi-square statistic is 6.9338 The p-value is 0.031213 This result is significant p < .05
- Reading the table:
- First cell under Fail shows:
- 9 is the number of students
- number in parentheses (8.32) is the calculated expected results
- the number in brackets [0.06] is the chi-square point for this data
- the established p-value is < 0.05
- means we can feel confident in our conclusions