This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Chapter 10.4 Testing the Equality of Variances Consider two random samples. Assume: x independent samples, and x normally distributed populations. The samples have and observations and the population variances are and. x n y n 2 X V 2 Y V Estimators of the population variances are the random variables and. 2 X s 2 Y s Introduce the random variable: 2 Y 2 Y 2 X 2 X s s F V V F is the ratio of two independently distributed chisquare random variables. A result from statistical theory is that F has an Fdistribution with numerator degrees of freedom and denominator degrees of freedom )1 n( x )1 n( y . That is, )1 n,1 n( y x F~F n is distributed as Econ 325 Chapter 10.4 1 Like the chisquare distribution, the F distribution is defined only for nonnegative values and the skewed shape of the probability density function depends on the degrees of freedom. PDF of the F distribution with (10, 20) and (20, 20) degrees of freedom. 0.5 1 1.5 2 2.5 3 3.5 F(10,20) F(20,20) Appendix Table 9, pages 8457 of the textbook, lists cutoff points or critical values that give an upper tail area of either 0.05 or 0.01 . Econ 325 Chapter 10.4 2 The application of interest is to test the null hypothesis: population variances are equal 2 Y 2 X :H V V against the twosided alternative: population variances are not equal 2 Y 2 X 1 :H VzV When the null hypothesis is true, , and the random variable: 2 Y 2 X V V 2 Y 2 X s s F has an distribution. )1 n,1 n( y x F From the numeric data set calculate the sample variances and. Arrange the two samples so that. 2 x s 2 y s 2 y 2 x s s " A test statistic is calculated as the variance ratio: 2 y 2 x s s The test statistic exceeds one since is bigger than. 2 x s 2 y s Econ 325 Chapter 10.4 3 For a chosen significance level , the decision rule is to reject the null hypothesis of equal variances if: c 2 y 2 x F s s " where is the critical value from the Fdistribution that satisfies: c F 5 " )F F(P c )1 n,1 n( y x Appendix Table 9 lists critical values for upper tail probabilities of...
View
Full
Document
 Spring '10
 WHISTLER
 Economics

Click to edit the document details