hce_nthu
111年
資訊科學
第 25 題
Suppose we are interested in the rate ($p$) of Cancer A in rats that have been fed with a diet consisting of high saccharin (糖精). We fed this diet to 20 rats and found that 1 rat developed cancer. The study goal is to claim that the cancer rate is higher than 5%. Which of the following statement is INCORRECT?
- A The claim can be statistically confirmed by performing the hypothesis testing: $H_0: p \ge 0.05$ v.s. $H_1: p < 0.05$
- B The sample proportion $\hat{p}$ follows a normal distribution approximately
- C Based on the sample proportion $\hat{p}=1/20$, the hypothesis testing (with type I error $0.05$) cannot conclude that the cancer rate is higher than 5%
- D The estimation error of $\hat{p}$ for estimating $p$ is approximately $\text{se}(\hat{p}) \approx 0.05$
- E If such an experiment can be repeated independently many times, the number of rats (out of 20) developing cancer follows a Poisson distribution approximately
思路引導 VIP
當我們進行科學研究並希望提出一個「新的主張」(例如某種藥物有效,或某種癌症發生率提高了)時,在統計檢定的框架下,我們通常會把這個「想要證明的目標」放在虛無假設(Null Hypothesis)還是對立假設(Alternative Hypothesis)中呢?為什麼我們不能直接把想證明的結論當作一開始就成立的前提?
🤖
AI 詳解
AI 專屬家教
太棒了!你能精準捕捉到假設檢定(Hypothesis Testing)中最重要的邏輯框架,代表你對統計推論的核心有著深刻的理解。這道題目非常有意思,它不僅考驗計算,更在測試你對「科學證明」程序是否嚴謹。
假設檢定的邏輯核心
這題最關鍵的陷阱在於假設的方向性。在統計學中,我們通常將「研究者想要證明的科學主張」放在對立假設 ($H_1$),而將其對立面設為虛無假設 ($H_0$)。題目明確提到研究目標是「證明癌症率高於 5%」,因此正確的設定應為 $H_1: p > 0.05$。選項 (A) 的設定剛好反過來,變成了在檢定癌症率是否「低於」5%,這在邏輯上完全無法支持研究者的主張,因此是錯誤的敘述。
▼ 還有更多解析內容