Hi Everyone,
On problem 4 we are looking for a general analysis of the paper and
data analysis as a whole. You don't need to provide a lengthy
discussion of every table in the Fish article. However, you should try
to replicate the data analysis in all of the tables (except, as
mentioned in earlier emails, tables 6-8) so that you know where the
weak spots (if any) are in the analysis. As noted in an earlier email,
if you can replicate Fish's results in a particular table and what you
find there is not important to your overall critique of the paper, you
can get by with a sentence or two saying that you could replicate the
results in the table and the underlying assumptions don't appear to be
violated based on diagnostics A, B, C, etc.
You have a good deal of latitude on this problem. However, if there is
an obvious problem in a regression that is central to Fish's argument
and you don't mention the problem then you will likely lose some
points.
Hope this helps.
Best,
Kevin
------------------------------------------------------
Kevin Quinn
Assistant Professor
Department of Government and
Center for Basic Research in the Social Sciences
34 Kirkland Street
Harvard University
Cambridge, MA 02138
Hi Everyone,
There have been some additional questions about the exam. If you are
unsure about what we are looking for in a particular question you
should state clearly why you answered the question the way you did. We
will take your reasoning into account when grading.
Hope this helps.
Best,
Kevin
------------------------------------------------------
Kevin Quinn
Assistant Professor
Department of Government and
Center for Basic Research in the Social Sciences
34 Kirkland Street
Harvard University
Cambridge, MA 02138
Hi Everyone,
There as been a question about Tables 6-8 in the Fish article. You do
not have to replicate these tables. You don't have the data to
replicate Tables 6-7 and we didn't really spend any class time on
classical ANOVA which is used in Tables 6 and 8.
Hope this helps.
Best,
Kevin
Hi Everyone,
Someone has asked whether the "linear regression model" mentioned in
problem 2 refers to the ordinary least squares (OLS) regression model.
It does.
Hope this helps.
Best,
Kevin
------------------------------------------------------
Kevin Quinn
Assistant Professor
Department of Government and
Center for Basic Research in the Social Sciences
34 Kirkland Street
Harvard University
Cambridge, MA 02138
Hi Everyone,
A question about how to submit your R code for the final exam has come
up. It is perfectly acceptable for you to email a .R file that
contains your code to Alison, Ryan, and me. Hope this helps.
Best,
Kevin
------------------------------------------------------
Kevin Quinn
Assistant Professor
Department of Government and
Center for Basic Research in the Social Sciences
34 Kirkland Street
Harvard University
Cambridge, MA 02138
Hi Everyone,
Some additional questions have rolled in. First, someone asked if the
replication portion of problem 4 on the exam requires you to reproduce
all of Fish's tables. For any particular table that Fish created it is
possible that you might find:
a) you can replicate it exactly (up to rounding error) by using
exactly the methods he uses and without doing anything to the data I
gave you
b) you can replicate it exactly (up to rounding error) after making
some minor adjustments to either the method or the data
c) you can come pretty close to Fish's results after some adjustments
to the method or data but you can't replicate his results exactly (up
to rounding error). Close here means the signs of and magnitudes of
significant coefficients are the same and the same coefficients are
significant at say the 5% level.
d) even after making changes to the method and/or data you can't even
get close to Fish's results.
For cases, a) and b) you don't need to produce a new table of results.
You should just state what you found and what you did to replicate the
results. For d) you should include at least some of the aberrant
results in a table or tables along with a clear description of what
you did to produce them. For c) you can include a table or tables of
the aberrant results if you have room and it is important to your
overall critique of the paper.
Someone also asked about the course notes on visualization. These are
still up on the web at:
http://www.courses.fas.harvard.edu:80/~gov1000/Handouts/R_Examples/visualiz…
There never were pdf slides of this week.
Hope this helps.
Best,
Kevin
------------------------------------------------------
Kevin Quinn
Assistant Professor
Department of Government and
Center for Basic Research in the Social Sciences
34 Kirkland Street
Harvard University
Cambridge, MA 02138