PS11-26

# Thisalsomakestheuseoftheconfidenceintervalsproblematic

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: e is small. (c) xqqmath(~ weight | feed, data = chickwts, + main = &quot;chickwt data&quot;, ylab = &quot;Weight at six weeks (gm)&quot;) The t­distribution can also be used with a normal distribution, but the QQ­plots show that the data do not fall on a normal distribution. This also makes the use of the confidence intervals problematic because that also requires a normal distribution. 4.45 (a) ?miaa05 iidsamples &lt;- replicate(20, sample(miaa05\$PTSG, size=15, replace=TRUE)) dim(iidsamples) for(i in 1:20) { print(t.test(iidsamples[,i], conf.level=0.90)\$conf.int) } [1] 1.294278 5.345722 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 3.343936 6.562731 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 2.898659 8.994674 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 2.150573 5.476093 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 4.943442 9.869891 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 2.742127 8.511206 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 3.409359 7.217308 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 2.271465 6.688535 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 4.065795 9.200872 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 2.437983 6.548684 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 3.566657 8.286676 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 3.26077 7.05923 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 3.853180 8.160153 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 3.188825 9.237841 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 2.804882 7.461784 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 4.073282 9.886718 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 3.199127 7.054206 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 2.388703 5.851297 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 4.359960 8.373373 attr(,&quot;conf.level&quot;) [1] 0.9 [1] 2.60393 6.75607 attr(,&quot;conf.level&quot;) [1] 0.9 (b) mean(miaa05\$PTSG) = 5.593284 18 out of 20 CIs contained the true mean. (c and d) I would have expected 90% (i.e. 18 out of the CIs) to contain the true mean, so I am not surprised by the answer in (b)....
View Full Document

## This note was uploaded on 01/25/2014 for the course STAT 23400 taught by Professor Collins during the Winter '09 term at UChicago.

Ask a homework question - tutors are online