translated by damien from Marco Avarucci

Basic Econometrics - Unit 1 Introduction

Experimental and Observational Data, Causality and the Ceteris Paribus condition.
 实验和观测数据，因果关系和其他同等条件。

1.Experimental Data

 1.实验数据

• Suppose we are interested in evaluating the effects of a new fertilizer on crop yields.
   假设我们有兴趣评估一种新肥料对作物产量的影响。

• To determine the causal effect of the fertilizer amount on crop yields we can conduct an experiment.
   为了确定施肥量对作物产量的因果关系，我们可以进行一个试验。

• Experiment design: choose several one-acre plots of land; randomly assign different amounts of fertilizer to the different plots; compare yields.
   试验设计：选择几块一亩地，随机给不同的地块配不同的肥料，比较产量。

• Ideally, all the other factors that influence crop yields such as quality of land, rainfall, presence of parasites etc. should be constant across the plots (ceteris paribus: other things equal).
   理想情况下，影响作物产量的所有其他因素，如土地质量、降雨、寄生虫的存在等，应在整个地块上保持不变（ceteris paribus：其他事项相等）。

• Other things are unlikely to be equal across units!
   其他的东西不可能在不同的单位之间是相等的！

• However, if we randomly assign the fertilizer and we have many plots, the individual features of each plot will be washed out. We can still infer causality!
   然而，如果我们随机分配肥料，并且我们有许多地块，每个地块的个别特征都会被冲掉。我们仍然可以推断因果关系！

Key point: The fertilizer assigned to the plots is independent of other factors affecting yields.
 关键点：分配给地块的肥料与影响产量的其他因素无关。

• The Perry project is an experiment designed to assess the effect of high-quality education on children.
   佩里项目是一项旨在评估高质量教育对儿童影响的实验。

• 123 preschool children assessed to be at high risk of failure in school were randomly divided into two groups.
   123名学龄前儿童被随机分为两组。

• One group entered a high-quality preschool programme, the other group received no preschool education.
   一组进入了高质量的学前教育项目，另一组没有接受学前教育。

• The project was conducted in 1962-1967 in Ypsilanti, Michigan. Students were followed trough their lives.
   该项目于1962-1967年在密歇根州的Ypsilanti进行。学生们被终生跟踪。

• At the age of 40, the participants in the pre-school programme were found to be more likely to have graduated from high school, committed fewer crimes,…
   研究发现，40岁时，参加学前教育项目的人更有可能高中毕业，犯罪率较低，。。。

2.Observational Data

 观测数据

• The previous experiment can be costly and could be criticized as unethical.
   先前的实验成本高昂，可能会被批评为不道德。

• To investigate similar topics, e.g. return to education, most economic data is observational.
   为了调查类似的话题，例如回归教育，大多数经济数据都是观察数据。

• If a person is chosen from the population and given another year of education, by how much will his/her wage increase?
   如果从人口中选择一个人，再接受一年的教育，他/她的工资会增加多少？

• We can use a survey to collect data on education and wages of a sample of individuals. The data will be treated as realizations of a (bivariate) random variables.
   我们可以利用调查收集样本个人的教育和工资数据。数据将被视为（双变量）随机变量的实现。

Key point: people choose their education.
 关键点：人们选择他们的教育。

• More educated people tend to earn more than less educated people.
   受教育程度高的人往往比受教育程度低的人挣得多。

• Smart people usually earn higher wages. Smart people often choose higher levels of education.
   聪明人的工资通常更高。聪明人往往选择更高层次的教育。

• Being smart could be the fundamental reason for the higher wage of more educated people (positive correlation between education and wages could be misleading).
   聪明可能是受教育程度更高的人工资更高的根本原因（教育和工资之间的正相关可能会产生误导）。

• The level of education is not determined randomly (as the fertilizer), but influenced by other factors (e.g. intelligence) that affect wages, and are different across units.
   教育水平不是随机决定的（如肥料），而是受影响工资的其他因素（如智力）的影响，各单位的教育水平各不相同。

Takeaway

 课外

• Even if we deal with observational data, it is useful to think about an experiment one would have run if possible.
   即使我们处理的是观测数据，如果可能的话，考虑一个实验是有用的。

• The previous example (return to education) highlighted the difficulties that arise when inferring causality in applied economics.
   前面的例子（回归教育）强调了在应用经济学中推断因果关系时出现的困难。

• Using multiple regression we can still infer causal effects from observational data.
   使用多元回归，我们仍然可以从观测数据中推断因果关系。