作者allen1985 (我要低調 拯救形象)
站內Statistics
標題Re: [討論] 關於ancova與T-test的結果
時間Sat Sep 10 21:20:37 2011
※ 引述《gioia000 (我沒有要引戰啊)》之銘言：
: 想請教版上高手..問到自刪
: 實驗數據共兩次測驗分為L1與L2
: 實驗組跟對照組 兩次測驗之間的ANCOVA都未達顯著
: 實驗組的成對T-test有顯著差異 對照組則無
: 所以實驗組組內前後測達顯著差異 可是組間卻沒有顯著
: 這實驗結果是失敗了嗎?還是應該如何解釋比較合理?或者有更好的跑法?
: 感謝回復
: ----
: 回應一樓:
: L1 L2是兩次測驗的成績 因為不同班 所以拿段考成績當作共變數分別跑兩次ANCOVA
: 兩次測驗都沒有顯著差異 前測平均72.9 VS 78.4 後測79 VS 81
: 段考平均則是誇張的561 VS 634 我以為後測變成2分的差距會有顯著差異
: 然後我用相依樣本T-test(同一母群) 測出實驗組有顯著差異 (平均明顯進步較多)
: 不知道這樣的解釋清楚了嗎?謝謝...不好意思阿記腦袋不太靈光XD
Frankly speaking, I still cannot understand your question clearly.
First, what is your response (y) ? average grade?
Second, what is your independent variables?
And what is your conti. independent variable and your discrete independent
variable?
Can you write down your linear models?
For example,
ANCOVA
post = beta_0 + beta_1X_1 + beta_2X_2 + error,
where X_1 means different class and X_2 means pre grades.
Besides, the testing result of beta_2 is not significant under alpha%
significant level.
Paired T test
Post - Pre = mu + error
where mu means the average difference between post and pre.
So, the testing result for mu is significant that mu is not equal to 0.
I don't know whether I understand your questions already.
--
※ 發信站: 批踢踢實業坊(ptt.cc)
◆ From: 75.180.59.249
※ 編輯: allen1985 來自: 75.180.59.249 (09/10 21:21)
※ 編輯: allen1985 來自: 75.180.59.249 (09/10 21:23)