25.11.2014 Views

Biostatistics

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

6.9 CONFIDENCE INTERVAL FOR THE VARIANCE OF A NORMALLY DISTRIBUTED POPULATION 195<br />

0.4<br />

d.f. = 1<br />

0.3<br />

d.f. = 2<br />

0.2<br />

d.f. = 4<br />

d.f. = 10<br />

0.1<br />

0.0<br />

0 2 4 6 8 10 12 14<br />

FIGURE 6.9.1 Chi-square distributions.<br />

(Source: Gerald van Belle, Lloyd D. Fisher, Patrick J. Heagerty, and Thomas Lumley, <strong>Biostatistics</strong>: A<br />

Methodology for the Health Sciences, 2nd Ed., # 2004 John Wiley & Sons, Inc. This material is reproduced<br />

with permission of John Wiley & Sons, Inc.)<br />

The Chi-Square Distribution Confidence intervals for s 2 are usually based on<br />

the sampling distribution of ðn 1Þs 2 =s 2 . If samples of size n are drawn from a normally<br />

distributed population, this quantity has a distribution known as the chi-square ðx 2 Þ<br />

distribution with n 1 degrees of freedom. As we will say more about this distribution in<br />

chapter 12, we only say here that it is the distribution that the quantity ðn 1Þs 2 =s 2 follows<br />

and that it is useful in finding confidence intervals for s 2 when the assumption that the<br />

population is normally distributed holds true.<br />

Figure 6.9.1 shows chi-square distributions for several values of degrees of freedom.<br />

Percentiles of the chi-square distribution are given in Appendix Table F. The column<br />

headings give the values of x 2 to the left of which lies a proportion of the total area under<br />

the curve equal to the subscript of x 2 . The row labels are the degrees of freedom.<br />

To obtain a 100ð1<br />

aÞ percent confidence interval for s 2 , we first obtain the<br />

100ð1 aÞpercent confidence interval for ðn 1Þs 2 =s 2 . To do this, we select the values of<br />

x 2 from Appendix Table F in such a way that a=2 is to the left of the smaller value and a=2<br />

is to the right of the larger value. In other words, the two values of x 2 are selected in such a<br />

way that a is divided equally between the two tails of the distribution. We may designate<br />

these two values of x 2 as x 2 a=2 and x2 1 ða=2Þ , respectively. The 100 ð 1 a Þ percent<br />

confidence interval for ðn 1Þs 2 =s 2 , then, is given by<br />

x 2 a=2 < ðn<br />

1Þs2<br />

s 2 < x 2 1 ða=2<br />

Þ

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!