又一個statistic 問題

我首先會講 z-test, 跟住係 t-test, 之後係 F-test, 最後係 chi square test, 因為你既問題都幾多, 如果我講少左或者 你有咩唔明的, 你可以叫我再補充啦

z-test:
(可以點用）
1) 可以test 兩個sample 既 mean係咪相同:
--Null hypothesis: mean1=mean2 (或者係 mean1-mean2=0, 兩個null hypothsis都得)
--alternative hypothesis: mean1 is not equal to mean2
2) 可以test兩個population mean (p1,p2) 係咪相同:
--Null hypothesis: p1=p2 (或者係 p1-p2=0, 兩個null hypothsis都得)
--alternative hypothesis: p1 is not equal to p2


z-test
(assumption): 同上次我講既差唔多啦, 你可以睇番
I) data are normally distributed (一定要) and
II) sample must be independent to each other,(一定要) and
III) standard dev. of pop. is known or ( 當ｎ細過30 既時候, 只要個standard dev of population 係知道, 我地都可以用 z-test)
IV) sample size > 30 (當我地只知道sample. standard dev. , 只要個sample size 大過30, 我地都同樣可以用 z-test)
總之一定要係 normal, independent 同埋 III) 或者 IV)其中一個就用得z

我assume你知道 1), 2) 既FORMULA, 我唔係度講，你唔知再問喇


＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊


t-test
(可以點用)
1) 同z-test一樣功用, 可以test 兩個sample 既 mean係咪相同:
--Null hypothesis: mean1=mean2
--alternative hypothesis: mean1 is not equal to mean2
　>>>>assumption有d唔同z-test
　>>>>t-test 有四種, 不同情況用不同既option（我之後會講）
　　　　　　　　　(a) both data are independent to each other( variace is equal)
(b) they are independent( variance is not equal),
(c) they are dependent to each other( variance is not equal),
(d) they are dependent to each othe (variance is equal)
2) [比較少用] 可以test correlation coefficient (r)---(即係當有 x 同 y 兩組data既時候, 我地要睇下佢地兩個有無relationship, 當x增加,y又會唔會增加 or 當x一路增加, y又會唔會一路減少,明唔明?)
xxxxxxxxxxxxxxxxxxxxxxxxxx
t-test
(assumption):
1) both data are normally distributed (or approximately normally distributed) and
2) sample standard dev. is known (no need population standard dev.) and
3) one or both sample size is less than 30

xxxxxxxxxxxxxxxxxx
t-test(當決定用t-test 黎compare two means, 我地要睇某d情況，有四種option)
第一: 睇佢地係咪independent 定係 dependent
----independent:
呢個好清楚,就係當兩組data 係獨立的, 佢地無任何關係影響到另一組數
----dependent:
兩組data係有關係的, 例如.....有廿個男人去做一個cholesrerol level test, 實驗前度既 CL 就係 data1, 佢地要食一種藥兩個月, 兩個月後呢十個男人再度 CL, 就係data2, data係由同一班人搵出黎, 所以佢地係有關係

第二: 再睇佢地variance 係咪相同:(用F-test 黎check)

所以, data係咪 in/dependent一睇就知, 我地開始計 t-test 前, 都會用F-test check左variance 先, 之後再用番相關既formula 計 t-test , (total 4種test), 唔同既formula 有唔同既 degree of freedom 要小心.
xxxxxxxxxxxxxxxxxxxxxxx

t-test (for correlation coefficient ):
呢個太少用,唔講咁多, 主要check x 同 y 個relationship 強唔強
例如: x 代表學生一日用幾多時間睇電視, y代表學生呢次考試既得分, 睇下睇得多電視,成績會唔會差 etc....


＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊

2007-04-04 06:10:05 補充：
因為過左4000字, 其他既去呢度睇啦http://h1.ripway.com/koolinfinity/ftest-chisquaretest.doc