General Comments from Paper Grading

Document Sample
General Comments from Paper Grading Powered By Docstoc
					General Comments from Paper Grading

Scoring was from zero to five points per part, with the exception of Model and Results
which was from zero to ten points. If the presentation within a section was equal to my
expectations for a typical paper in this course, the score was 3 (6). A score below this
meant that there was some element lacking. A score above this meant that either the
ideas and information presented or the quality of the presentation itself exceeded my
expectations. A score of 5 (10) indicates an exceptionally clear and concise presentation
of excellent work and ideas.

The sections of the paper should be as follows:
0. An Executive Summary providing a brief description of the issue analyzed, a sentence
or so telling what data was used and a couple of sentences describing the results,
especially which explanatory factors had negative and positive estimated effects.
1. A Background discussion introducing the topic, explaining why it is interesting or
important, summarizing some previous work done and explaining what is to be
investigated in the paper.
2. A Data Description section giving the source of the data and describing it briefly,
listing the variables used and presenting mean values and standard deviations for them in
a table. Any variables that were calculated using original data should be clearly
explained.
3. A Model and Results section describing the model(s) estimated, commenting on
explanatory power, giving coefficient estimates and levels of significant in a table and
briefly describing what was positive and negative and what was significant or not.
4. A Discussion of Results that tells what the results mean and what is important,
interesting or unusual. This section should also wrap things up and conclude the paper.

An executive summary should be a brief description of the topic, the data used and the
results, including which factors had positive effects and which had negative effects. The
idea is that a very busy person should be able to look through the summary and know
what the question was and what the answer was in a couple of minutes.

When using sets of dummy variables, be sure that one group is excluded and in your
discussion be clear about which is excluded and about discussing the fact that the
estimated coefficients on all the remaining groups represent effects relative to this
excluded group.

When describing data, it is good to offer some means and standard deviations in a table,
along with a few paragraphs discussing what each variable is.

                                     Mean             St. Dev.
          Male                       0.52
          Education
               Some College           0.10
                    Bachelors         0.26
            Graduate Degree           0.15
          Age                        38.52             20.37
          Height                     65.25             10.25


Try to present regression results in concise tables and then discuss them. For example,
you might imagine a model of income that a person is trying to use to establish whether
or not age and height are significant factors. This person estimated a model with income
as the dependent variable without age and height, a model with income as the dependent
variable with age and height and a model with ln(income) as the dependent variable with
age and height. In the second model, age and height were not significant but in the third
they were, although height had a negative impact.

Sample Presentation of Regression Results
Dep. Variable        Income                   Income                 LN(Income)
R2                   0.23                     0.28                   0.65
               Male 2578.32*                  1536.25*               2.65*
       Some College 863.25                    638.25                 0.39
           Bachelors 5863.28*                 5268.45*               1.26*
    Graduate Degree 6985.47*                  4756.25*               0.68*
                Age                           34.16                  0.10*
             Height                           -0.52                  -0.03*
* p < 0.05

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:0
posted:10/1/2012
language:English
pages:2