| H O M E | ||
Academic |
||
 |
A Brief Biography | |
|
Curriculum Vitae | |
|
Student Area (Emory) | |
|
PODCASTS | |
|
Political Music Videos | |
|
Data Sets and Computer Programs |
|
Publications |
||
|
Scholarly | |
|
Speculative Nonfiction | |
Consciousness Studies |
||
|
The Farsight Institute | |
|
Links of Note | |
|
Book Reviews | |
PUBLIC SPEAKING |
||
|
Topics | |
|
Schedule | |
|
Videos | |
|
Photos | |
|
Media Statement | |
| MUSIC | ||
|
Live Sacred Music | |
|
Studio Music | |
| ACTING | ||
|
African Television | |
PERSONAL |
||
|
Poetry | |
|
Peace Corps | |
C O N T A C T |
||
Statistical Modeling Final Paper/Project
Instructions for the third draft ...
Now that you have a topic, a data set, some hypotheses, and a research design, you need to get some results. Using the statistical tools you have learned in class, conduct some statistical tests to evaluate your hypotheses. Try a variety of statistical approaches. For example, think about how you might productively utilize a test for the difference between two means, or a bi-variate regression, or a multiple regression. Do you want to use dummy variables? If so, why not try using an F-test to test the difference between slopes and intercepts? When conducting multiple regression analysis, always remember to check for excessive multicollinearity using a correlation matrix and even ridge regression in R, or TOL and VIF in SAS. Finally, try to construct a really nice plot that is relevant to your analysis.
More about Tables and Figures:
Tables and figures need to be able to stand alone without support from
the text. Thus, you need to give all relevant information in the tables
and figures. Figures usually require a lengthy title that tells the reader
what the figure is all about and perhaps even how to interpret it. But
tables require even more information. Do not just put in the parameter
estimates. You also have to list information that relates to significance
levels, and perhaps even multicollinearity results if appropriate. For
example, it is typical for many good tables to list the parameter estimates,
the standard errors for these estimates, and some indicator of whether
or not the estimates have passed a given threshold significance level
(the p-value). All tables must also contain the N, or sample size, from
which the estimates are based. If you are conducting F-tests for the difference
between dummy intercepts and slopes, include the F-statistics as well
as their significance levels. If multicollinearity is an issue, include
the TOL values.
It is always a great idea to look at examples from published work that is available from major journals. I suggest looking at examples of tables from the American Political Science Review. You can see such examples online at JSTOR which is available from Emory Library Databases.
Discussion
You need to discuss your results. You do this in a "Results"
section. This means that you need to step the reader through the tables
and figures to explain what is in them and why the results are important.
Try to note the important points, and relate all this back to your hypotheses.
Always think in terms of the big picture. What is the main point of your
research? Have you addressed the big questions that you started out to
answer? Include either a summary or implications section at the end of
your report. Here is where you are able to hammer home the substantive
importance of your findings.
Hand your draft in on time. This is important.