Chi-Squared Test Study Notes
Chi-Squared Test Study Notes
Null and Alternate Hypothesis
Understanding Statistical Significance:
A method to determine if there is a statistically significant difference between two treatments: use the Standard Error of the Mean.
Error Bars Overlap:
If error bars for two sets of data overlap, it indicates no statistical difference.
Observed differences are likely due to random chance.
Error Bars Do Not Overlap:
If they do not overlap, there is a statistically significant difference.
Direct Testing of Hypothesis:
In some cases, hypotheses can be directly tested using statistical tests.
The Chi-Squared Test is emphasized as the only statistical test to be used in this course.
Formulating Hypotheses
Begin by crafting a hypothesis, which you will test for statistical significance.
Two types of hypotheses for a given phenomenon:
Null Hypothesis (H0):
States there is no relationship between two variables.
Example: "Cheese kept at room temperature has the same amount of mold as cheese in a refrigerator for a week."
Alternative Hypothesis:
States there is a relationship between two variables and the finding did not occur by chance.
Example: "Cheese kept at room temperature has a different amount of mold than cheese kept in a refrigerator for a week."
Chi-Squared Test Overview
Definition: The Chi-Squared Test assesses whether the differences between observed and expected results are statistically significant.
Function: It helps determine if the observed differences arise from random chance or another factor.
Method:
Propose a hypothesis.
Determine the probability that the hypothesis is wrong, providing a probability that the alternative hypothesis is correct.
Choosing a Hypothesis
The hypothesis should generate expected results for comparison against actual observed results.
Example: Hypothesis about a die roll states each number is equally likely (1/6 probability).
A comparison of actual frequencies against expected frequencies of 1/6 can lend or reduce support for the hypothesis.
If you assume unequal frequencies, it becomes untestable without specific expected values.
Chi-Squared Formula
Formula: χ^2 = Σ \frac{(o - e)^2}{e}
Where:
χ^2 = Chi-Squared statistic
Σ =