14.33  Research & Communications in Economics

Dear all, just fyi -- below is a potential outlet for your 14.33 papers, if you are interested.  The deadline is January 12th.
 - Heidi Williams

-------- Original Message --------

To Whom It May Concern:

Columbia University’s /Journal of Politics and Society/ is currently
accepting student papers for publication in our Spring 2015 /Journal/ ,
and we request your help in passing on our call for papers to your
undergraduate students. The /Journal/ has consistently published
multiple papers by your students in the past few years, and we would
like your assistance in continuing to discover and print high quality work.

We are an academic journal of the social sciences and history,
distributed nationally among academics on EBSCO, Google Scholar, and the
Library of Congress. In its 25^th  year of publication, the /Journal
/features exclusively undergraduate work on political and social issues.
We are seeking outstanding student research papers from seminars,
upper-level electives, independent study and thesis programs.

If you believe your students have produced work that would be suitable
for the /Journal/ , it would be greatly appreciated if you would let them
know about our call for papers (typically between 20 and 50 pages,
although all submissions will be considered). Please feel free to share
this information with your current students as well as those with whom
you have worked in the past.

_The deadline for student paper submissions is January 12th, 2015_. To
download a Call for Papers PDF, please visit: http://www.helvidius.org/cfp

If you care to learn more about our organization or to view past
editions of the /Journal/ , please visit _www.helvidius.org
<http://www.helvidius.org/>_.

Everyone -- again, fantastic job on your presentations.  You should be really proud of your papers and presentations as a capstone on your work this semester.  Great job, and good luck finishing up your final drafts.

I know that everyone is very busy at this time of the semester with final papers and exams, but I am writing to ask that you please take a few minutes to fill out the end-of-term subject evaluation for this class.  This is our opportunity to get your feedback on how we can improve the class in future years (e.g. one Stata session would have been sufficient vs. you would have preferred three Stata sessions rather than two).  We're going to be working hard to continue to try to improve the course for future cohorts of your peers, and your feedback and suggestions are the most helpful input we have to try to do that.

The evaluations need to be filled out by Monday 15-December at 9am:

http://web.mit.edu/subjectevaluation

Again, thanks from Sarah and I for a great semester.
 - Heidi Williams

dear all,

in my feedback on your first drafts, I mentioned to several of you that you should think more about whether your regression was more appropriately specified in levels or logs of the dependent variable.  Because a few of you had questions on this, I wanted to post a few additional notes to clarify.

Many variables are very right-skewed, rather than normally distributed -- medical expenditures would be one example.  The way to look at this in your data is to create a histogram of your dependent variable and to see whether the distribution looks approximately normally distributed, left (negative) skewed, or right (positive) skewed (http://en.wikipedia.org/wiki/Skewness).  I think the most relevant case for most of you is a right skew. 

When you have a right skewed dependent variable, estimating a linear model such as OLS with the level of the outcome variable is often inappropriate because it restricts the effects of your right hand side variables to be additive in levels, which often doesn't make sense from an economic perspective.  For example, say that medical expenditures vary tremendously across US states, and you want to relate medical expenditures to a treatment variable and also condition on state fixed effects.  It is much more natural to assume that a given treatment will shift the medical expenditures in a given state up by an equal percentage in all states (e.g. 50%) than to think that a given treatment will shift medical expenditures in a given state up by an equal amount in levels (e.g. $50).  The log model naturally transforms the data in this way which is often more appropriate from an economic perspective.

If you are looking for a reference on this, I would recommend Stock and Watson's "Introduction to Econometrics" textbook, pages 267-276.  Here are some other notes that may be helpful: https://www.kellogg.northwestern.edu/faculty/dranove/htm/dranove/coursepages/Mgmt%20469/nonlinear.pdf.

Concretely, the way to decide whether a log or level specification is more appropriate in your context is to look at the distribution of your dependent variables: if they look approximately normally distributed then a level specification is appropriate, whereas if the log of the outcome looks more closely like a normal distribution than a log specification is appropriate. 

Finally, note that a small issue arises if you have zeros in your dependent variable since the log of zero is undefined.  For the purposes of your paper, you can just use log(y+1) rather than log(y) if this is the case.

Hope that helps.
 - Heidi Williams

REMINDER: Class tomorrow (Monday 12/1) will start early, at 8.40am.  Please be sure to arrive on time for the in-class presentations.

I finished reading all of your first drafts, e-mailed you my comments, and posted your grades on the course website.

I was very pleased with these drafts, and just wanted to highlight a few general comments as you navigate revisions for your final paper (due Wednesday 12/10).

First, given that you turned in your papers last Monday, I was clearly a bit rushed on reading and writing up comments on all of your papers by today -- but wanted to prioritize getting these back to you today so that you would have time to work on your revisions.  I say that just to say: if any of my comments come across as harsh, please try not to take them that way, and instead chalk that up to me being in a bit of a rush to write these up.

Second, I gave many of you the advice to cut your "literature review" section and instead integrate an abbreviated literature review into your introduction or elsewhere in the paper.  It is becoming more common for research papers to be structured in that way (you can look at my paper that we read in class as an example), and that will also give you more space to clearly explain your analysis and still come in under the page limit.

Third, several of you turned in papers that were above the page limit.  I tried to give specific suggestions on things to cut, but please keep in mind that your final paper must come in under the page limit.

Fourth, I would encourage you to read over my comments soon, to get a sense of what I'm asking you to do.  For each of you, some of my comments are small (easy-to-implement) writing suggestions, and others are comments that require new empirical specifications.  You should start thinking about the latter sooner rather than later.

Finally, I tried to focus my comments on constructive, feasible suggestions for how to improve your papers on a time scale that is realistic between now and when your final draft is due.  As is detailed on the syllabus, grading of the revision will be based entirely on how well and how completely you respond to the comments you are given on your first draft.  That said, in addition to my specific comments each of you should try to give your paper several "read through edits" that try to streamline the text and make the paper more readable for the final draft -- that is a macro-comment that is relevant for all of your papers.  This type of editing is hard for me to offer specific advice on how to do, and that requires personal initiative on your part, but it is a very important part of the revision process.

Again, great job.  I look forward to seeing the rest of your presentations this coming week.  As a reminder, please try to arrive promptly so that we can stay on schedule with the presentations.

 - Heidi Williams

Dear all,

Just a reminder that tomorrow we will be starting class early, at 8.40am.  Because we will be starting the in-class presentations tomorrow, it is very important that everyone arrive at class on time and be on-deck ready to present.

I also wanted to clarify that your in-class participation grade (20% of your grade) will put particular weight on attending these last three classes with the in-class presentations, as it is very important to me that everyone attend class and give feedback on the other students' presentations. 

Very much look forward to your presentations starting tomorrow!
 - Heidi Williams