Presentation on theme: "Chapter 9: Simple Regression Continued"— Presentation transcript:
1Chapter 9: Simple Regression Continued Hypothesis Testing and Confidence Intervals
2Inferences on regression coefficients To place confidence intervals or test hypotheses on a and b we need to know sa2, and sb2 which will be estimated by sa2 and sb2.Where s2 is estimated by:
3Inferences on regression coefficients. If the model is correct, then b/sb and a/sa are distributed as a t distribution with n-2 degrees of freedom.Confidence limits on a are:Confidence limits on b are:
4Hypothesis Testing: Test on a Ho: a = aoHa: a ≠ aoTest StatisticReject Ho if
5Hypothesis testing: Test on b Ho: b = boHa: b ≠ boTest StatisticReject Ho if
6Hypothesis Testing: Significance of regression equation Ho: b = 0 (equivalent to Ho: r = 0)Ha: b ≠ 0Test statistic and rejection region same as previous test on b.If this hypothesis is not rejected thenmay be estimated byIf r = 0 then s2 ≈ sy2 or the regression line does not explain a significant amount of the variation in Y.
7Confidence Intervals on the Regression Line Determined by first calculating the variance of,the predicted mean of for a given Xk.The standard error of can be estimated bycalculated as:
8Confidence Intervals on the regression line The variance of depends on the value of X at which the variance is being determined.Var ( ) is a minimum where Xk = and increases as Xk deviates from .Confidence limits on the regression line are:
9Confidence Intervals on the regression line Since increases as Xk - increases, the confidence intervals on Xk = are at their narrowest and widen as Xk deviates from .Confidence limits on an individual predicted value of Y would be wider than the confidence interval on the regression line, since for an individual Y, the Var (e) or s2 would have to be added to the Var( ).Thus the variance of an individual predicted value of Y would be Var( +s2).
10Confidence intervals on an individual predicted value of Y Can be calculated by the previous confidence interval equations wherewould be substituted for
11Confidence Intervals on the standard error Can be made by noting that (n-2)s2/s2 is distributed as a chi-squared distribution with n-2 degrees of freedom. Limits are given by:where