Statistics 5

STEVIE-G™

2009-11-19 4:04 am

As follows:

http://i707.photobucket.com/albums/ww74/stevieg90/13-9.jpg

更新1:

在(b)中,若sampling error是6個人而非6%,計法會有不同嗎?

更新2:

(b) part主要唔明如果個單位係%,那如何辦?

更新3:

回答 (1)

myisland8132

2009-11-19 5:19 am

✔ 最佳答案

(a) 1-[15C0(0.13)^0(0.87)^15+15C1(0.13)(0.87)^14]=0.5987

(b) p=0.13,1-p=0.87
Let √p(1-p)/n<=0.06
=>n>=31.42. At least 32 samples are needed.

(c) p=1-28/350=0.92,
Test-statistic
=(0.92-0.87)/√0.87(1-0.87)/350
=2.7815

Since 2.7815>1.96, reject H0 and conclude that there is an envidence that the response rate has been dropped at 5% level of significance.

(d) (0.13-√0.08(1-0.08)/350<p<0.13-√0.08(1-0.08)/350)
0.1155<p<0.1445

2009-11-19 12:36:50 補充：
I agree with you, it shoud be 1-tailed test and use 1.645 instead of 1.96

👍 0 👎 0