Quantities

time-series data vs cross-sectional data:

1. time-series data: observations taken over a period of time at a specific spaced time intervals 在一个时间段内以某个时间间隔划分得到的观测值
2. cross-sectional data: observations taken at a single point in time 在一个时间点下的多个观测值

longitudinal data vs panel data:

1. longitudinal data: several features for one object duirng a time series 一个时间序列下同一观测实体的多个特征
2. panel data: one feature for several objects during a time series 一个时间序列下多个观测实体的一个特征

central limit theorem
population with distribution ($\mu ,{\sigma }^2$), then the mean $\bar{x}$ of the samples of size $n$ from the population has the distribution of ($\mu ,\frac{{\sigma }^2}{n}$) as the smaple size becomes large (sufficiently larg $n\ge 30$).
对于一个总体，其分布为($\mu ,{\sigma }^2$)，那么从中采样，样本数为$n$，随着样本数变多（$n\ge 30$），样本的平均数$\bar{x}$满足分布($\mu ,\frac{{\sigma }^2}{n}$)。

standard error of the sample mean = $\frac{\sigma }{\sqrt{n}}$

degree of freedom in sample with size n: n-1
样本数量为n的样本，自由度为n-1，因为对于某一个分布，最多只有n-1个样本可以自由取值，剩余1个样本的值由那n-1个样本决定。常见例子为，已知3个数字a, b, c的平均值为4，最多只有2个数字可以自由取值，最后一个数字的值将由这2个数字决定 (12-另外2个数字之和)。

Usage of z-statistic or t-statistic

null hypothesis vs alternative hypothesis

1. null hypothesis: $h_0$, want to reject
2. alternative hypothesis: $H_a$, want to conclude

对于$h_0$, 可能为$\mu ={\mu }_0$或者$\mu \ge {\mu }_0$或者$\mu \le {\mu }_0$，总是包含等号

one-tailed test vs two-tailed test

1. one-tailed test: $x>0$等单向条件
2. two-tailed test: $x\ne 0$等双向条件

对于z-distributed test statistic

• 拒绝$H_0$假设当z-statistic值不在z-value范围内
• 无法拒绝$H_0$假设当z-statistic值在z-value范围内