Term
| Conditions for inference in regression |
|
Definition
Straight Enough Condition for linearity
Independence Assumption
Does the Plot Thicken? Condition for constant variance
Nearly Normal Condition for Normality of the residuals |
|
|
Term
| Residual standard deviation |
|
Definition
| Se = sqrt(sum(y-y-hat)^2/(n-2)) = sqrt(sum(e^2)/(n-2)) |
|
|
Term
| t-test for the regression slope |
|
Definition
| when assumptions are satisfied, we can perform a test for the slope coefficient; we usually test the null hypothesis that the true value of the slope is zero against the alternative that it is not; a zero slope would indicate a complete absence of linear relationship between y and x, H0:B1=0, t=(b1-0)/SE(b1), P-value is found from student's t-model with n-2 degrees of freedom |
|
|
Term
| Confidence interval for the regression slope (B) |
|
Definition
| when the assumptions are satisfied, we can find a confidence interval for the slope parameter from b1 +/- t*n-2 X SE(b1). The critical value t*n-2 depends on the confidence level specified and on Student's t-model with n-2 degrees of freedom |
|
|
