关于MACSS的一点信息和疑问~

DorothyZheng11UID: 3825892

在urch里看到的关于Uchicago macss的信息。。看了之后突然觉得对这个项目的好感降低，感觉macss不是很适合经济phd?有没有正在读的小伙伴现身说法下？
http://www.urch.com/forums/phd-e ... -advice-needed.html

Although the Chicago MACSS program is new, I have some familiarity with their required/recommended coursework as well as mild exposure to the sociologists and political scientists on that MA program's committee (note that only one economist is a committee member; the other economists are only affiliates). Overall, I wouldn't consider their training as valuable or complementary with what is done in graduate-level economics. To the extent that they're teaching you data analysis, you'll either learn methods orthogonal to current econometrics techniques (and thus, unlikely to be useful in econ), or statistical/econometric materials that are weakly inferior to undergrad-level coursework at most universities.

Note that undergraduate coursework from UChicago, in fact, forms nearly half of the core in their quant-oriented recommended tracks (CMSC 121-3, MATH 19620, STAT 244A-B), and that the comp-sci and math sequences here are the lowest (easiest) sequence for 1st/2nd year undergraduate students. The stats sequence is admittedly the most rigorous of the three intro sequences offered at the undergraduate level, but is nonetheless generally taken as a freshman or sophomore by the undergraduate population. To be really blunt, even with the most quantitative track suggested, this is not a real graduate-level program in quantitative methods. Keep in mind that although you'll also be recommended to take graduate "methods" courses in the second year from psychology, poli-sci and sociology departments, I have sufficient exposure (some first-hand, some second-hand) to strongly believe that both their breadth and depth would be strictly dominated by, for instance, some convex combination of electives in undergrad statistics and applied econometrics at the same ranking of universities.

(It's important to consider that you also won't be learning anything tangential to modern economic theory for two years, although there are some decent positive political theorists on that committee.)

That's the mostly professional comment. Here's a more personal one: you shouldn't neglect the amount of bullshit ideas or totally wrong applications of basic statistics that you may have to suffer through during necessary coursework, even if you choose all the electives wisely. I have no idea who will teach the three first-year "core" courses, but it appears that it's neither the economists nor the positive political theorists, which means you have much cause for concern. In my own (very) substantial and continuously disappointing experience within other social sciences, it is difficult for an outsider to overestimate the gap between what is comprehensively presented in advanced undergrad or 1st-year grad level micro-econometrics, and what many "quant" faculty, in nominally high-ranked departments of social sciences, appear even capable of grasping, never mind utilizing in research. For instance, part of the recommended "advanced" coursework, in the program description, are the seminars on causal inference from the human development department, which you'll find are taught by a PhD in education who literally made up her own statistical terminology and used them as course titles as she went along (and where she contrasts the "cutting-edge" potential outcomes framework, i.e. the one from Rubin 1974 or generally Fisher 1918, with "conventional strategies in quantitative methods", where she then proceeds to list three different 'methods' that are all equivalent to OLS). Also recommended as your second-year MA coursework are entire sociology courses on "hierarchical linear models", otherwise known by real statisticians as random-effects models, which in turn is just a special case of fixed-effects models, generally covered within one week in standard econometrics. No, they aren't any deeper than this appears.

You should wonder, for example, why there isn't a single statistics faculty, nor any MLers/theorists from comp-sci, on a program's committee for "computational social science". I can reject an obvious hypothesis: this isn't a result of path-dependent disciplinary divide, because the statistics department at UChicago was one of many in the country which were initially founded by economists. To put it diplomatically, the reality is that most of the faculty in UChicago's non-econ social science departments, even those with strong quant leanings and who appear ​to be pioneering new methods, are ultimately on the bottom of the downstream/demand side of quantitative techniques, if they read any modern statistical literature at all, and thus have no actual complementarities with statistics or econometrics faculty. More specifically, a substantial part of what "quant social science" in sociology, psychology and education do is to constantly belabor themselves to invent new names for OLS + various arbitrary mixes of unspecified (and generally untested) structural assumptions.

To sum up, it looks like a fancily-motivated introduction to basic quant methods for people with very weak undergraduate backgrounds (e.g. many aspiring psychology, sociology, education, poli-sci PhD students), but I don't see it as remotely qualified preparation for economics grad school, and it is largely a waste of time if you already have standard econometrics training and received passing grades. I don't dispute there is indeed the opportunity to work with the 3-4 econ faculty affiliated with that program, but this seems like getting through a lot of crap to end up paying money to write your own paper under light supervision, whereas you can do real economics at LSE and then get paid for doing real research for an economist, for the year after that.

Marshall3UID: 3696039

这个是我去年纠结选校时在urch发的帖子下面的回复，虽然我后来去了LSE，但对MACSS多少还是了解一些，所以稍微来说两句。

我之前也研究过MACSS的faculty member的背景，虽然不知道最后是不是他们教课，但你贴出来的回复应该算是比较中肯的，non-econ social scientists教的quant course对于经济背景的学生而言是没有价值的。

总的来说这个项目给我的感觉会比较适合两类未来想申phd的学生，一类是本科数学和统计背景较差 (eg.缺数学分析和高年级统概之类的课，且没法用summer school来补全的)，那么这个项目确实可以选芝大本科的分析和统计课来补充背景。第二类则是自己对未来的发展方向已经有了一定的认识，并且确知需要补充哪类的课程，而这些课程又很难通过自学来完成。比方我有个朋友今年去了MACSS，他感兴趣的领域主要是macro，而且自身的编程水平也不错，于是他waive了入门级的编程课选了C++，同时选了一个sequence的phd macro课。

除这两种情况之外的话这个项目意义可能并不是很大。诚然现在做实证需要很多的统计知识，而且编程的能力也是必不可少的。但对大多数学生而言，一年的graduate level计量其实在methodology方面已经是差不多够用了，剩下的即用即学就可以。而编程方面就我个人的经验而言，除非是macro这类对代码优化需求很大的领域，其他的分枝像是labor, development, trade之类的完全自学就可以了。Stata, R/Python/Matlab找几本教材跟着把代码都敲一遍，然后做项目的时候干中学已经足够了，完全没必要浪费时间和学费去学些data analysis的课。就算以后要用到ML, deep learning的内容，听听网课或者等到phd阶段再选课也不妨。

综上，如果想申phd的话同是芝大的项目MAPSS可能会更合适大多数学生。

DorothyZheng11UID: 3825892

Marshall3 发表于 2017-12-07 22:07
这个是我去年纠结选校时在urch发的帖子下面的回复，虽然我后来去了LSE，但对MACSS多少还是了解一些，所以稍微来说两句。

我之前也研究过MACSS的faculty member的背景，虽然不知道最后是不是他们教课，但你贴出来的回复应该算是比较中肯的，non-econ social scientists教的quant course对于经济背景的学生而言是没有价值的。

总的来说这个项目给我的感觉会比较适合两类未来想申phd的学生，一类是本科数学和统计背景较差 (eg.缺数学分析和高年级统概之类的课，且没法用summer school来补全的)，那么这个项目确实可以选芝大本科的分析和统计课来补充背景。第二类则是自己对未来的发展方向已经有了一定的认识，并且确知需要补充哪类的课程，而这些课程又很难通过自学来完成。比方我有个朋友今年去了MACSS，他感兴趣的领域主要是macro，而且自身的编程水平也不错，于是他waive了入门级的编程课选了C++，同时选了一个sequence的phd macro课。

除这两种情况之外的话这个项目意义可能并不是很大。诚然现在做实证需要很多的统计知识，而且编程的能力也是必不可少的。但对大多数学生而言，一年的graduate level计量其实在methodology方面已经是差不多够用了，剩下的即用即学就可以。而编程方面就我个人的经验而言，除非是macro这类对代码优化需求很大的领域，其他的分枝像是labor, development, trade之类的完全自学就可以了。Stata, R/Python/Matlab找几本教材跟着把代码都敲一遍，然后做项目的时候干中学已经足够了，完全没必要浪费时间和学费去学些data analysis的课。就算以后要用到ML, deep learning的内容，听听网课或者等到phd阶段再选课也不妨。

综上，如果想申phd的话同是芝大的项目MAPSS可能会更合适。

谢谢大神的回复～受教了

共由UID: 3689117

楼主可以看我去年发的关于macss信息整理的帖子，里面有引用到你提到的那段话。

那一段话黑其他社会科学黑的非常犀利到位，以至于我专门收藏了每次要吐槽时都会拿出来引用。
而且的确，macss 是一个非常社会学主导的项目，经济学方向的学生反倒有些另类。列在macss 官网上的那些课程也是非常不适合经济学学生。

但是这不妨碍macss同样可以是一个非常好的项目。因为我们的选课同样是全校任选的，并不是作者误以为的仅限于列出来的那些课程。
铺开来讲太长了今天恐怕没时间。（寒假时本来更新了下那个帖子，结果闪退没保存，一怒之下放弃）
简单地说，MACSS vs MAPSS 的比较可以抽象为：

如果你编程底子好：2年任选课的项目 vs 1年任选课的项目
如果你编程底子不够扎实：1年的cs训练+mapss vs mapss

（对，macss第一年限制了3门perspective 是必修，最多只有6门选修余地，但是相信我，你就算在mapss也不会真的想把三门课都选econ的硬课而不修些水课的）

PS：我就是二楼提到的那个朋友。

ctlienUID: 3656714

這種曇花一現型態的項目幾個共通特色
1. 錄取門檻低  (相較於經典正統項目)
2. 選課彈性高  
3. 出路依賴個人特質
4. 未來program會消失
上面兩位從各方面很清楚Evaluate如何使用這些項目，以及在哪些狀況下是optimal choice。

DorothyZheng11UID: 3825892

ctlien 发表于 2018-1-8 17:47
這種曇花一現型態的項目幾個共通特色
1. 錄取門檻低  (相較於經典正統項目)
2. 選課彈性高  

hhh感觉录取门槛不算低吧~~楼上的两位都是大神啊
感觉mapss和macss还是比lse econ和duke econ难录的吧~

ctlienUID: 3656714

DorothyZheng11 发表于 2018-1-8 19:20
hhh感觉录取门槛不算低吧~~楼上的两位都是大神啊
感觉mapss和macss还是比lse econ和duke econ难录的吧~

我沒黑這program意思  不要錯誤解讀 lol

我意思是這種趨勢program會比本program簡單
比如Chicago MACSS 會比 Chicago MA Econ (如果存在) 簡單一些..
至於跟其他校比不是我原本意思了。

然後你講的幾個 LSE大概=Chicago ，這討論串的人基本上都是同時拿LSE+Chicago的
duke不太清楚

畫個重點: 这不妨碍macss同样可以是一个非常好的项目  (共由 2018)


ps 我跟上面兩樓都在Chicago(帶獎)和LSE(不帶獎)之間掙扎過的同道中人。

共由UID: 3689117

ctlien 发表于 2018-1-8 19:34
我沒黑這program意思  不要錯誤解讀 lol

我意思是這種趨勢program會比本program簡單

哈哈其实ctlien总结得很对。
MACSS 的项目门槛的确比 MAPSS 低。 MAPSS 应该就是UChi的MA Econ 了（列在econ dept grad program里的）。
我的个人直观感觉是，录取门槛上，MAPSS approx. = EME > Duke approx. = LSE Econ > MACSS。
但是，这两届的共同特点是，都有超级强的同学放弃 MAPSS、EME 来 MACSS 了，导致MACSS的方差特别大，“出路依赖个人特质”。

锦梓UID: 3527026

共由 发表于 2018-1-9 00:31
哈哈其实ctlien总结得很对。
MACSS 的项目门槛的确比 MAPSS 低。 MAPSS 应该就是UChi的MA Econ 了（列 ...

想知道MAPSS和MACSS的placement结果～等待楼上的朋友们来发offer榜哈哈

板栗球球球球UID: 3858694

共由 发表于 2018-1-9 00:31
哈哈其实ctlien总结得很对。
MACSS 的项目门槛的确比 MAPSS 低。 MAPSS 应该就是UChi的MA Econ 了（列 ...

你好！请问可不可以告知一下macss的大致timeline呀～看寄托上大家放的notification date都是二月底到三月初，不知道是不是1.4那个round的呢？感谢！

zshmathUID: 3840758

板栗球球球球 发表于 2018-2-10 01:03
你好！请问可不可以告知一下macss的大致timeline呀～看寄托上大家放的notification date都是二月底到三 ...

之前去见过mapss和macss的staff和教授，没记错的话他们是说三月初第一周出结果，应该默认的是priority deadline的那一批 至少是这么跟我说的

板栗球球球球UID: 3858694

zshmath 发表于 2018-2-26 17:45
之前去见过mapss和macss的staff和教授，没记错的话他们是说三月初第一周出结果，应该默认的是priority de ...

啊感谢感谢，超靠谱！