Hypothesis Testing Flashcards
Hypothesis Testing
Test theory about population parameter
Two types of hypotheses
Null hypotheses and Alternative hypotheses
Chi-Square Tests
Goodness-of-fit Test and Contingency Tables
Goodness-of-fit Test
A way to determine if a given population has a specified theoretical distribution by comparing the frequencies of the original sample to the expected frequencies of a hypothesized distribution.
Null hypothesis for goodness-of-fit test
Hypothesized distribution is a good fit for the sample
k in goodness-of-fit test formula
of classes
df in goodness-of-fit test
df=k-1
For goodness-of-fit test, need each ei to do what? If this requirement isn’t met, what can you do?
Need each ei ≥ 5, else combine adjacent classes
If we use the sample to estimate the relevant parameters of the distribution, then
df = # of classes - 1 - # of parameters estimated
Contingency Tables
To test the independence of two variables of classification
Null hypothesis in Contingency Tables Test
Two factors are independent
If A and B were independent, we would expect P(A ∏ B) =
P(A)P(B)
Contingency tables test degrees of freedom =
(# rows - 1)(# columns - 1)
Statistical significance of association does not equal
a practical significance
A large chi squared value indicates what? What does it not indicate?
Indicates strong evidence of association, but not necessarily a strong association.