Lecture 16: Chi-Squared Test Flashcards
What does a chi-squared test do
Determines if there is a association between 2 categorical variables
T/F: a chi-squared test can be one-sided and two-sided
False, it can only be two-sided, even though it has a one-sided probability
What does it mean when the x^2 is very high
That the test is highly significant
What are two test that deal with a lack of continuity in small samples
- Fisher’s exact test
- Jates’s correlation
What does Fisher’s exact test do
It calculates the exact x^2 for small samples, when x^2-distribution does not yet suffice (cell size < 5)
What’s a fallacy of Jate’s correlation
It may tend to overcorrecte which can lead to overly conservative results
What are three other ways to look at a chi-squared test
- Likelihood ratio
- Standardized residuals
- Effect size
What does the effect size look at
The odds ratio based on the observed values
What is the odds ratio
The quantification of the magnitude of the effect, how much more likely is one than the other
What are the degrees of freedom under the chi-distribution
Df = (r - 1)(c - 1), where r is number of rows and c is amount of columns
What is the chi-distribution
The chi-distribution describes the test statistic under the assumption of H0, given the degrees of freedom
What are the two most important metrics
- Chi-squared test
- Odds ratio
What is a limitation of the chi-squared test
Obtaining significance here doesn’t say very much because it’s very influenced by the sample size; if the sample is big than it can lead to significance even though it shouldn’t
