Collecting Data (All In context(只能回答本问题，非广泛性))
SRS: Label -> Random number generator (Unique / Different)
Stratified -> Strata / Random Select from each stratas
Cluster -> Cluster / Select n from all clusters
Systematic -> Every 10th
Compare -> 夸一贬一
Bias ->
- Volunteer
- Response / non-response
- Under coverage -> In context (Under / Over : Estimate)
Experiment
- Factor (x)
- Treatment: List not assigned
- Response variable (Recorded)
- Experiment unites (需要写数字): 380 Volunteers
Block
特征分类，相似
减少Confounding Variable
Blind
Researcher
Subject
1.Random
Balancing
The effect of other variable
2.Replicate
减少 By chance 结果
3.Control
Same Condition
4.Compare
Two or more treatment
Control Group
As a baseline to compare
Placebo
减少心理因素影响
Randomly assigned
Cause and effect
Randomly selected (X Volunteer) (Not Representative)
Inference for population
Exploring Data
Describe Distribution (SCSO) (In context)
抄问题
Probability and Sampling distribution
用 Binomial 先提及（解释用BINS）
Normal mu sigma
Binomial n p
E(x) = np
Sigma(x) = \sqrt(np(1-p))
出现probability先定义 （A是什么，B是什么）
算概率都要过程
Sampling Distribution
xbar mu sigma/sqrt(n) -> CLT(Central limit Theory) 无论总体分布是什么形状，只要样本容量足够大，样本均值的分布就会近似服从正态分布
phat p sqrt(p(1-p)/n)
过程要写出来
Inference
4 Steps
Interval / Test 写名字
Chi - Square
- goodness of fit test (1 Cat Variable): H0: P_1 = ?, P_2 = ? / Same as claim
(O - E) - Independence (2 Cat Variable) H0: Independent (No association)
- Homogeneity (2 Cat Variable) H0: Proportion is same
Pvalue 定义: 得到一个 P (Statistic >= Value) =？
Type I / II
Consequence