作者sploo (正确的方向决定一切)
看板graduate
标题[请益] 统计考古题的问题
时间Wed Jan 30 21:27:52 2013
主要是问C小题的问题
其他的不知道有没有写对
请高手帮忙
A study was conducted in a large city following the dramatic increase in
coffee price during the final quarter of 2003. The objective was to compare
the mean supermarket prices of the "four" leading brands at the end of the
year.
"Eight" supermarkets in the city were selected, and the price per pound was
recorded for each brand. The mean square of prices for brands was .05 and
that for supermarkets was .03. The total sum of squares of prices was .45.
A. Are the average prices of coffee the same for the eight supermarket?
Use α = .05
B. Are the average prices of coffee the same for the four brands?
Use α = .05
C. Find a 90% confidence intervals for the difference in the mean price of
Brand A and Brand B if he sample mean price of Brand A was $2.43 and that
of Brand B was $2.49.
D. List the assumption required for the above tests.
====================================================
A和B是用ANOVA表去算出F值
F(brands)=11.628
F(supermarkets)=6.977
然後用H0和H1去做有没有全等的检定
算出A B 皆在拒绝域 AcH1
C.一开始先用Ho H1检定"变异数"有没有相等
拒绝域 CR:{Fo< F0.025(3,7)=0.1698,Fo<F0.975(3,7)=5.89}
接下来不知道该怎麽用ANOVA表的东西写出 Fo
D.常态性 同质性 独立性
感谢
--
※ 发信站: 批踢踢实业坊(ptt.cc)
◆ From: 220.132.123.76
※ 编辑: sploo 来自: 220.132.123.76 (01/30 21:34)
1F：→ sevres:假设brand a,brand b样本数分别为na,nb，然後scheffe's? 01/30 22:10
2F：→ sevres:题目好像没有给na,nb，所以好像也算不出来?个人浅见 01/30 22:11
3F：推 ethefalse:他只有要求单纯的CI吧?就直接用一般的CI 变异数用MSE代 01/30 22:24
4F：→ ethefalse:个人看法-.-" 01/30 22:24
5F：推 robinlee:不用检定变异数 ANOVA已经假定变异数相同了 01/30 22:30
6F：→ robinlee:直接求Scheffe's区间 na=nb=8 01/30 22:31
7F：→ robinlee:如果用Scheffe's法求算 答案应该是(-0.1473, 0.0273) 01/30 22:43
8F：→ robinlee:有错请指教 谢谢 01/30 22:44
9F：→ sploo:也是算一样的答案 我把他想的太复杂 这题应该是一因子集区 01/30 22:52
※ 编辑: sploo 来自: 220.132.123.76 (01/30 23:05)
10F：推 ethefalse:可以请教一下为什麽要用Scheffe吗@@? 01/30 23:23
11F：→ sploo:单一是事前检定 sch是事後 第二题我检定是接受H1 为了找出 01/30 23:47
12F：→ sploo:不全等 所以用sch 有错请指教 01/30 23:48
13F：推 goshfju:C.题目没有要你检定变异数 (请看D. 你已经假设相等了) 01/31 00:47
14F：→ goshfju:他只是要普通的信赖区间 不是要两两比较用的联合信赖区间 01/31 00:48
15F：→ goshfju:所以就算一般的信赖区间就好了 01/31 00:49
16F：推 kobebryant08:第4题的假设应该有饱含变异数相等 01/31 01:22