Sociology 5811:  Final Project Assignment

 

Proposal Due:  Tuesday, November 15

Final Paper Due:  Start of class, Tuesday, December 13

 

Overview

 

Your final project will be to produce a brief paper in which you apply the statistical skills you've learned in this class to a topic of your own choosing.  You will locate a source of data, develop hypotheses, test them, and interpret the results.  In short, you will conduct a small research project.  Because this is not a course in theory or research design, I do not expect a literature review or an elaborate interpretation of the social meaning of your analyses.  Rather, emphasis will be placed on developing hypotheses that can be tested with data, utilizing appropriate analyses, and providing clear interpretations of the results.

 

The Topic

 

You may choose any topic you wish, as long as it is one that can be properly studied using multivariate regression analysis (either alone, or in combination with other methods we discuss this semester).  Since you will not have time to collect new data (unless you are extremely motivated), you must ensure that your project can be completed with an existing dataset.  Looking at the available sources of data may help you narrow your research topic.  I encourage you to talk with professors if you don't have your own research ideas.  Of course, the TA and I will be available to discuss both topics and sources of data.

 

Data

 

The university has a significant library of data files with variables relevant to social science analysis.  Many are available on a CD compiled by David Knoke and Karl Krohn.  Some info is at: http://www.soc.umn.edu/~krohn/Handouts/DataSources.htm

 

Part I:  The Proposal

 

Your first task is to develop a proposal that will get you started on your work and provide an opportunity for me to give you feedback.  In this proposal, which will be about 3 pages in length, you should inform me of the following:

 

1. Indicate your area of interest and general research thoughts.  Be as brief as possible.  What are you interested in studying and why?  What sociological process/theory is the source of your hypotheses?  Make sure you include some argument, however rudimentary, for the hypotheses that follow. 

 

2. List the specific hypotheses that you will test in this project.  You must test at least two research hypotheses.  Depending on your topic, you may have several research hypotheses.  Indicate both the null and alternative hypotheses in each case.

 

3. Discuss your data.  Indicate the data source, as well as a description of the specific variables.  Mention whether they are nominal, ordinal, or interval, as well as the total number of cases (N).  Note:  To use multivariate regression, you will need a dataset with at least one interval/ratio measure (or possibly an ordinal variable with many categories) that can serve as your dependent variable).  Keep that in mind when looking for data. 

 

4. Indicate the statistical methods that you will use to test each hypothesis.  You must use multivariate regression to test some hypotheses, but you may also need to use other methods.  Also, list the assumptions that your method requires.  (Assumptions of regression won’t have been covered in class yet.  Look ahead in Knoke for information.)

 

 

Part II:  The Final Paper

 

If you have completed the project proposal, you should have the following components of your project assembled:

 

1. A general topic.

 

2. A theory, piece of reasoning, or argument about the social processes which you hope to study, as well as specific hypotheses that you hope to test.

 

3. A set of data that you can use to test your hypotheses.

 

4. A list of statistical techniques that you can use to test your hypotheses.

 

You now have all the raw materials to begin your analysis. Below I will outline the major steps of your project. The steps may also serve as a general outline for your write-up of the project.

 

I. Introduction/Theory/Hypotheses

 

I want you to develop and clarify the argument or theory that you think is responsible for the outcomes you expect in your hypotheses.  (Note:  You may have already done a good job of this in your proposal.)  It isn't enough to say “I want to look and see if X affects Y”. Instead, explain why you think X affects Y, and/or why it will not.  Be specific in your arguments.  Also, make sure you provide arguments necessary to support all of the hypotheses that you hope to test.  In addition, you should state your hypotheses in this section.  Make sure that they reasonably follow from your theories/arguments.

 

II. Data/Indicators

 

Next, you should describe your data.  This will include a brief description of the source, variables to be used, and total N.  Make sure your variables make sense given your theory/hypotheses.  Provide explanations for any ambiguous indicators.  That is, for most cases, your variables will link with your theories in a straightforward, obvious manner-- such as using a variable of a person's reported salary to reflect “income”.  However, if your variables are more ambiguously linked to your theory (e.g., using income as a measure of “career success”), provide and explanation/justification.

 

III. Exploratory Analyses

 

Statistical research typically begins with exploratory analyses-- i.e., descriptive statistics, histograms, and scatter plots in order to understand your variables and their distributions.  This is important for two reasons.  First, it helps you refine your understanding of the data, as well as your hypotheses.  Many important sociological insights have come from exploratory analyses.  Second, this information helps you determine if the assumptions of your method are met -- e.g., Are the equal variance assumptions of ANOVA met in your data?  Is the linearity assumption of regression met in your data?  And so on…

 

At a minimum, you should present and interpret exploratory analyses that test the assumptions of regression analysis for at least one of your independent variables – to demonstrate that you know how to do that.  (You need not present assumption checks for all variables in the final paper, as that would fill pages and pages for a large multivariate analysis.  But, you should do all the checks and report anomalies.)  Also, if you are using methods in addition to regression, make sure to present exploratory analyses that check relevant assumptions for those methods as well.  Finally, you may wish to report additional exploratory analyses that are of significant interest – but that is optional.

 

IV. Hypothesis Tests/Conclusions

 

Next, you must employ hypothesis testing procedures.  Make sure that you are using procedures that are appropriate for the data you have.  If you are making any assumptions (e.g., that a 10 point scale is an interval variable), note them.  Choose an alpha-level and determine whether you are using one or two-tail tests before you begin.  Next, run appropriate analyses and interpret the results.  Don't just say whether you can reject the null hypothesis-- report relevant statistics, such as T-values and/or P-values, etc.  Then, indicate your conclusions.

 

Finally, take time to describe possible sources of bias or errors which might have affected your conclusions-- arising from measurement, sampling, sample size, violated assumptions of your methods, outliers, etc.  This is important!  Don’t forget to do this!

 

The Write-Up

 

I want you to do brief project write-ups, of approximately 7-8 pages (not including tables).  I suggest you use the sections, above, as an outline. Include printouts of any relevant exploratory analyses (plots, histograms, etc) and hypothesis testing procedures as appendices.  Include only that which is relevant to your paper.  Be clear and accurate in your descriptions and interpretation of methods and results.  Good Luck!