F-Test & Model Significance Flashcards
what is the F-Test?
It is used in multiple regression to test whether the regression model as a whole is statistically significant
The F-test checks if multiple coefficients are jointly significant
What is the Null hypothesis (H0) in an F test?
all the regression coefficients (except the intercept) are zero
(e.g the independent variables do not explain Y)
What is the Alternative Hypothesis in an F-Test?
At least one of the regression coefficients is not zero, making the model useful
It means that the regression model explains some variation in Y and is statistically significant
What is the formula for the F-statistic?
F= (SSE/k) / (SSR/n-k-1)
SSE = sum of squared explained variation
SSR = sum of squared residuals
k = Number of independent variables
n = sample size
The higher the F-statistic means the model explains a lot of variation in Y
What is the decision rule for the F test?
1) compare the F-statistic to a critical value from the F-table
2) if F is greater than the critical value = reject H0 (model is significant)
3) if F is smaller = fail to reject H0 (model is not significant)
work through an example of the F Test:
Suppose we estimate the following multiple regression model:
wage=𝛽0+𝛽1(education)+𝛽2(experience)+𝛽3(gender)+𝑢
We get:
SSE=450
SSR=150
k=3 (3 independent variables: education, experience, gender)
n=50 (sample size)
1) Compute the F-Statistic
𝐹=(450/3) / (150/50−3−1)
𝐹=150/3.19
= 47.02
Step 2: Compare to Critical Value
From the F-table (at α=0.05), the critical value for k=3, n−k−1=46 is approximately 2.8.
Since 47.02 > 2.8, we reject
𝐻0 → the regression model is statistically significant.