This project is an opportunity for you to conduct your own applied
statistical analysis of data of your choosing. You will work as part of
a group of 2 or 3, and you will prepare a 3-page written report
documenting your analysis. You will also deliver a short, 7-10 minute
presentation of your work to the class during the last week of
classes.
Objectives
During your project you should focus on demonstrating the
following:
- An ability to use data visualizations and descriptive statistics
that are appropriate for the variables involved in your analysis.
- An ability to consider and/or control for confounding variables that
are relevant to your analysis.
- An ability to appropriately integrate methods of statistical
inference (confidence intervals and hypothesis testing) into your
analysis.
- An ability to recognize and communicate the practical significance
and limitations of your findings.
Timeline
- Monday 10/12 - Form a group of your choosing containing 2-3
members
- Friday 10/16 - Submit a 1-2 paragraph proposal that briefly
describes the following:
- The research question that your analysis will focus on.
- The data you plan on using in your analysis (including information
on the primary source, or a plan for collection), including a list of
variables involved in your research question.
- A short justification on why this project has an appropriate level
of difficulty for this course and will allow you to fulfill the
objectives of this assignment (see previous section). You may outline
some of your planned analyses in this part of the proposal.
- Friday 11/6 - Final version of your data and a collection of data
visualizations and descriptive statistics is due. This should include:
- Your cleaned data and a brief description of its contents, as well
as any preparation/collection steps you undertook.
- 2-3 data visualizations involving variables relevant to your
research question.
- Descriptive statistics (mean/median, standard deviation,
proportions, correlations, etc.) relevant to your research
question.
- Monday 11/23 - Initial analysis is due by 11am. This should include:
- Refined, corrected, or polished versions of the data visualizations
you previously created
- Statistical inference, using an appropriate selection of hypothesis
tests, confidence intervals, and statistical models.
- Wednesday 12/9 and Friday 12/11 - In-class presentations
- End of finals week - Final reports are due
Details:
Presentation:
Presentations will be 10 minutes and delivered on data voted upon by
the class. Your presentation should clearly outline your research
question, how your data were collected, the relevant patterns and trends
you found in your data, how you used methods of statistical inference,
and the practical conclusions you drew from your analysis. the resources
section at the end of this assignment for more information on giving a
scientific presentation.
Paper:
Reports should be no more than three (3) pages in length
(single-spaced), including embedded figures and tables, but
not including an optional abstract, extra figures or tables
that appear in an appendix, or references. The paper should contain the
following components: a title, a background section, a methods section,
a results section, a discussion section, and references. If you are
unsure of how to structure these sections or what information they
should contain, please visit the links in the “Resources” section below
for guidelines and examples. Recognize that staying within the report’s
three-page restriction can take a fair bit of planning and you need to
be deliberate and focused in deciding what to include in your report. In
statistical analyses it is extremely common to not include everything
that was explored when writing up the final results.
Incentives:
If you collect your own data using something other than a
survey you will receive an extra 2% on your paper’s final score.
However, if you are collecting your own data you must collect at least 4
variables (such as two variables involved in a research question, and at
least two potential confounders, demographic variables, or other
additional features)
Mistakes to Avoid
- Avoid choosing a research question involving a single variable.
- Don’t pose a research question that cannot be answered by your data.
Unfortunately there are lots of great research question for which
appropriate data doesn’t exist and would be impossible to collect
yourself.
- Prioritize quality over quantity, fewer hypothesis tests are
generally better. A good analysis is deliberate and coherent, it
doesn’t haphazardly combine a large number of different tests and data
visualizations together.
- It is perfectly acceptable for a project to yield no statistically
significant findings. In fact, it is much preferred that you report no
statistically significant results than alternatives like retroactively
changing your research question or p-hacking/data
dedgring to find significance anywhere in your data.
DASIL
The Data Analysis and Social Inquiry Lab (DASIL) is an excellent
campus resource that can help you acquire and prepare data that might
otherwise require tools beyond the scope of this course. Click
here for details
USCLAP
The Undergraduate Class Project Competition (USCLAP) holds a paper
competition for introductory undergraduate statistics students. The
requirements of the competition are very close to those of this
assignment’s final report. A successful submission is an excellent
resume builder and numerous Grinnell students have won or placed in the
past. As an instructor I am willing to help guide revisions of your
paper (after you turn it in) in order to facilitate your submission to
USCLAP.
Click
here for competition details
Additional Advice:
- Many students choose to work with topics related to Grinnell
College. If you choose to do so and plan on submitting your work to
USCLAP, you’ll need to mask any explicit references to the school you’re
studying at in order ensure your submission is properly blinded (ie:
speak in general terms).
- Visit this link
if you need some help finding a data source or coming up with a project
topic.