笔记对应普林斯顿公开课程 Statistics One 的Lecture 1~2 .
中英字幕地址：https://www.bilibili.com/video/BV1sy4y1e7Eg?p=1
1. Experimental Research(实验研究)
Example : Polio Vaccine（骨髓灰质炎疫苗有效性实验）
研究种类：Randomized controlled experiments.
sample(样本)
initial：弗吉尼亚州的4000名儿童 final：来自44个州的180万儿童 population（总体）：美国的所有儿童
Independent variable（自变量）
Treatment：
Vaccine Placebo Dependent variable（自变量）
Polio diagnosis (measure of ana individual child) Rate of polio (measure of a group of children) Results
Rate(per 100,000) :
Treatment: 28 ; Control: 71 Summary :
True independent variables Random and representative samples No confounds [混淆] (impossible, but we try our best)
The major benefit of randomized experiments is they allow strong claims about causality(因果关系) Strong causal claims require:
The importance of Random
Eperimental research requires (for sample) :
Random selection from the population Random assignment to the condition
so that we have similar types of samples across the conditions in the experiment.
2. correlation research（相关性研究
Example 1 : Intelligence theories
Theories of intelligence have been proposed based on detailed analysis of patterns of correlations across different types of tests—— I refer to these as "studies" of intelligence rather than "experiments" because no variable is manipulated (no independent variable).
Example 2 : Effects of concussion(脑震荡)
May cause neural damage and cognitive deficits.
Quasi-independent variable(准自变量)
Treatment:
Suffered a sports-related concussion Control group Dependent variable
Neural measures Cognitive measures
Quasi-independent variable并不是一个真正的自变量，因为并没有对样本运动员进行随机分配至不同条件（脑震荡或没有脑震荡）。因此，在进行causal claim的时候，必须注意区分造成组间差异的是脑震荡还是其他因素。这些其他因素就是潜在的混淆因素。
Confounds Prior concussions Prior hits to the head (not necessarily concussions) Personality types more likely to be aggressive
由于Quasi-independent variable并没有进行随机采样（也就是样本对总体并没有很好的代表性），所以基于这个研究的casual claim并不如实验研究那么有力。事实上，虽然形式有类似之处，但是这种类型的医学研究通常来说依然属于“相关性研究”而非“实验性研究”。
