Applied RegressionFinal Project

Due Date: Monday, May 9, 20165:00pm

The final project requires each group to perform a regression analysis of a data set
selected by the group and to produce a written report detailing the analysis techniques and
the findings of the project. Below I outline the details of the project:
Groups: You are allowed to work in a group of up to three students; individual projects
are also allowed. Groups (of more than one person) are also required to submit a memo
detailing the contributions of each member, so that I know who did what.
Data Set: The data set may be your own or may be obtained from an external
source. If you are having trouble finding a data set I will be happy to help you find
one. The data set must be appropriate for a multiple linear regression (not simple
linear regression) and there must be enough data to perform necessary analyses.
You must have your data set pre-approved by Prof. Johnson before handing in the
final project. Its better to get the data set approved earlier rather than later, just
in case the data set is denied. To get your project approved, you need to write up a
one-paragraph summary of the data set, including information about the size of the
data set and the research questions you will try to answer.
Analyses: The multivariate data analysis methods utilized need to be compatible
with the research goals and the data. Assumptions implicit to the analysis need to be
evaluated. If assumptions have been violated, it is OK to proceed with analysis, but
you must state the limitations of the results.
Report: The report should not be too lengthy, but needs to be long enough so that I
understand what the data is, the conclusions you make, and know how you arrived at
the conclusions. Ten pages is probably more than enough.
I have offered one way to organize your paper below, but feel free to do what works
best for you, just make sure that the organization is easy to follow.
AbstractStates the main results of the paper (not the objectives) and possibly
the methods used. It should be short; four or five sentences is probably enough.
IntroductionA brief (approximately one page) general description of the problem, including:
Why the problem is of interestyou might refer to previous studies using
this, or similar data sets.
Information about the data set, e.g, the source of the data, how many observations, number of variables, etc.
A brief summary of the analysis performedtell me what methods you have
utilized, e.g., We perform a multiple linear regression to examine the differences in prices of cereals across the shelves on which they are displayed.
Detailed information about the methods should be left for a later section.
1

 The main results of your analysis.

Exploratory analysesThe introduction introduces the data set and gives the
reader a brief taste of the results that will be presented later in the paper. This
section should provide the reader (me) with graphical and numerical summaries
of the data, paying special attention to summaries that provide evidence for the
results youve mentioned in the introduction.
MethodsThe introduction tells the reader (me) what the data is, what method
(on a general level) youve used, and the main results of your analysis. The
methods section should expand the description of the methods used. Topics that
should be covered in this section include:
If appropriate, explicitly define the regresion models used in your analysis.
Make sure to state why you think the method(s) is (are) appropriate for the
data.
Discuss and evaluate the assumptions of the regression model. If youre data
does not quite conform to the assumptions, make note of it, and discuss the
implications.
State the hypotheses your are testing, both in terms of the applied problem,
and in terms of the model parameters; also state which testing procedure(s)
you are using. For example, We perform a hypothesis test to determine
whether the mean price varies by display shelf,
H0 : 1 = 2 = 0
H1 : j 6= 0 for some j,
with a partial F-test.
Detailed ResultsThis section expands the explanation of the results, and
includes, where appropriate, tables and figures providing evidence for the conclusions youve stated. You can also report any secondary results youve found.
DiscussionSummarize the findings one last time, paying close attention to the
limitations of the analysis. You can share thoughts with the reader about how
you might expand the study, improve on the model youve used, and what are the
long-term implications of the findings.
Grading: Your score on the final project will be comprised of the following:
Clarity, 15%. If I scratch my head and ask myself, what the heck are they
trying to say? several times when reading the paper, then its probably not very
clear. Delete long sentences with complex structure in favor of ones that are
relatively short and easy to understand.
Appropriateness of the method(s), 30%. This is the most important piece
of the project. When evaluating this portion of the project I will be asking my
self, Is this what I would have done? Is there a better way to perform this
analysis?
2

 Ability to draw the correct conclusions, 25%. So youve used the right
method(s) for data analysis. Did you use that method correctly to draw conclusions?
Thoroughness, 25%. Did the group perform a complete analysis? Was an
adequate exploratory data analysis performed? Is there something in the data
that the group failed to discuss? Were all of the model assumptions discussed
and evaluated? Were limitations of the method(s) discussed?
The wow factor, 5%. Extremely well-written papers will be rewarded. Did
the group go beyond the call of duty in their analysis? Is the paper extremely
well-written? Did the group suggest ways to extend the work or how the analysis
could be improved?
Submission. The final project must be submitted by uploading an MS Word or PDF
file to Moodle by no later than 5:00pm, Monday, May 9, 2016.