经济article被教授狂骂太垃圾无法看懂，求大神们授业解惑！

背景：不怕大家笑话,目前在美国读研究生。

教授给我的评价原文大概是：
"If you are a bad writer, people will believe you are a bad thinker--muddled words are the surest sign of muddled thoughts. You tend to add longer and more exotic words until your idea is lost and you ultimately write something that is false or unsubstantiated. It leads me to believe you want to distract rather than illuminate. "


我写的是对一篇关于IV(instrumental variable)的经济论文的review。我就摘两段自认为问题所在的段落，这两段是在说明为什么此论文的研究不能使用ols模型进行估计。如果哪位大神可以指点迷津，我不胜感激啊。


“Initially if we measure the veteran impact on income using OLS model, the
independent variable would be a binary variable: "whether the person i is a veteran
at time t in cohort c" and the dependent variable would be the person’s
income. Control variables in the model should include all the variables that
affect a person’s income, but are uncorrelated with the independent variable. In
addition, the cohort effect dummy variable and period effect dummy variable
should both exist in the model.


However, a self-selection issue occurs when certain unobservable variables which
also affect the binary veteran variable fail to be included in the model. Perhaps
people who would gain most benefits from military (they might obtain promises
to get high income as a veteran) choose to join the Vietnam War. Or their decision
to fight in the war is correlated with other characteristics that affect earnings.
For example, people (even from a poor family) with less IQ or ability might have
higher possibility to join the war if their parents believe they will earn more than
not being veterans. These joining-war decisions and the get-most-benefit situation
end up contained in the error term since they are hardly measurable to
be control variables. Due to the correlation with unobservable control variables,
the independent variable turns out to be endogenous. In addition, the veteran
impact on income may be overestimated since the research subjects are veterans
who have ensured higher income; or the model might underestimate the veteran
impact: Some veterans with less IQ or ability will earn even less than common
veterans because of the lacking of several-years’ education and working experience.”

赞成mchen117的说法 你这样写很中国
美国是要段落内容总分总  第一句一定要写行还是不行 后面是证据论点 最后一句也是要写总结句的
lz可以去take几堂本科的英语课 他们会教学术写作

非大神 ~~ 但是教授的意思应该就是不要堆砌复杂句式 先想清楚逻辑关系 层层递进

美式论文比较常见的套路是 第一句先总结 比如 OLS这个模型对于xxx是行不通的 后面紧接着跟上原因12345 每个原因后面接一个 for example 没有考虑到xxx 而xxx会造成什么影响 最后结尾再reiterate一遍 这就是为什么OLS不是最适用的模型

楼主还可以考虑把长句剪成短句 尽量保持全文structure的明确 这样方便让读者follow你的逻辑框架

其实楼主应该感激这样的教授，可以从教授的评语里看出她/他是个很有讲究的writer。 多去writing center以及系统的上一些writing 课。听说读写都没有捷径，只能苦练。加油！

你这写的像中文思考后翻译出来的

doodleknight 发表于 2019-4-8 06:28
你这写的像中文思考后翻译出来的

那能不能举个例子 比如某一句话怎么改能改成一种美式思考出来的句子呢？ 他说我写的导致他无法理解，是因为chinglish的原因还是语法或者别的呢？谢谢

先问下lz是否自己写的时候像教授说的那样故意把事情写得复杂化呢？是否为了写的高深而写？还是说这些都是表达清楚自己意思必要的东西。教授的态度确实过于严厉，但是也可以反过来想想是否如他所说。

binary var 后面的解释有点迷啊

我觉得这两段写得还行啊，都可以看懂，也没有什么exotic的词语。

zhefan 发表于 2019-4-8 07:37
先问下lz是否自己写的时候像教授说的那样故意把事情写得复杂化呢？是否为了写的高深而写？还是说这些都是表 ...

您好，谢谢您的指点

我解释一下这两段的写作思路。我第一段其实就在列出我们原始的ols模型的各个组成部分 “解释变量，控制变量，error term和被解释变量” 以及他们的属性。
第二段我就反驳第一段ols model的error term在这个研究中并不满足它的正确属性。因为它和解释变量存在关系。
perhaps....or 这两句话就是在说可能存在的关系。for example后面就是对上面两个可能的关系进行通俗化的举例。一个是参军的人可能已经承诺了以后退伍后会挣更多的钱（我自认为可能有些军官的孩子就通过参军退伍这种路子来继承他父亲的军官位），另一个是低智商低能力的人，可能本身就挣不了多少，参军以后耽误两年工作经历或者学习经历，退伍以后就挣得更少。
最后我就阐述了上面两种例子。第一种例子会导致估计的系数比真实的要高，毕竟不可能都是军官的孩子。第二种就会导致估计的系数比实际的小，毕竟不是所有参军的人都低智商低能力。
我觉得对于这两段话，每一句话的信息都很重要（虽然词语用的肯定很不对）。所以我并不觉得某一句话可以删除掉或者做简化。不知道您是怎么认为的呢？

currant 发表于 2019-4-8 08:03
我觉得这两段写得还行啊，都可以看懂，也没有什么exotic的词语。

谢谢您，那您能否再看一下下面一段话呢？这段话是紧接着上两段的，意在解释为什么使用iv可以改变存在的问题，是一种正确的估计方法。谢谢了！

“The author applies the instrumental variable approach to solve the identification
problem. He selects the draft lottery for the Vietnam War as the binary
veteran variable’s instrument variable. The number of draft lotteries people
draw in each year and each cohort determines whether they are chosen to join
the Vietnam War. Since the lottery numbers are drafted randomly to registered
people (so-called Random Sequence Number (RSN)), the draft lottery being as an
exogenous variable and is independent with error term seems plausible. As the requirement
states that people with draft lottery RSN below a ceiling (determined
by the Defense Department) are selected to join the war, this implies a correlation
between the draft lottery and veteran explanatory variable. In conclusion, the draft
lottery is considered to be an instrument variable. while being uncorrelated
with the error term in the model, it is correlated with the endogenous explanatory
variable and indirectly affects the dependent variable income. By replacing the
veteran explanatory variable in the OLS model with another estimation model of
veterans with the draft lottery as an independent variable, the new model then becomes
a good estimation model (meets all Gauss Assumption: BLUE, unbiased and
efficient).”

pengcheng1993 发表于 2019-4-8 07:58
binary var 后面的解释有点迷啊

您好，不好意思我解释一下第一段。

我是在陈述原始的ols模型的各个组成部分 “解释变量，控制变量，error term和被解释变量” 以及他们的属性。
“binary var：”后面其实就是对这个binary解释变量进行描述："某个人在第几年在第几cohort是否是退伍军人"。然后是被解释变量“人的income”。
接着我陈述说这个model里面的控制比那两也是必不可少的，并且它需要满足什么条件（这样才能是正确的）。in addition后面我又说因为有不同年份不同cohort，我们model里面还需要有年dummy和cohort dummy（这样才能是正确的一个model）。
但是我并没有在这一段写到error term的事情。那是因为第二段我开始反驳说error term在这个model里的使用并不正确。因为它与解释变量相关了。。。。后面便是对“为什么相关”做分析。
不知道这么解释能不能理解呢？