Guidelines for Soc 280B and Soc 381 presentations.
* Presentation outline
* In-Class Presentation
For this in-class presentation, you must do some original data analysis..
If you are doing work on a dataset already, use that data. If you donít have a dataset in mind, start searching the ICPSR archives, http://www.icpsr.umich.edu/icpsrweb/landing.jsp
You can also check ipums, at http://www.ipums.org/
And there are lots of project-specific data websites out there; I have a project on How Couples Meet with data here: http://data.stanford.edu/hcmst
1) In-Class Presentation:
* The in-class presentation should be a 10 minute presentation. It is important to stick to the allotted time, so practice and time yourself before hand.
* The presentation will consist of one empirical question, which should be stated clearly at the beginning of the presentation.
* The empirical question must be tested with a dataset. The dataset can be CPS data, or any other dataset. Other publicly available datasets are available via ipums. You may use your own data, even proprietary data, as long as you explain, very briefly in the presentation, where the data come from.
* A proper test of the empirical question can be made with two tables, or two slides [soc 381 can use up to 5 substantive slides, and as many appendix slides as you want. Soc 381 presentations have included more slides over the years, which is OK as long as you keep your focus on the main question. If you lose the thread of your main question, that is not good].
* Post your slides to the course software package (currently Canvas) before the presentation, and I can put them up on screen for you.
a) The first slide will include summary statistics. If the question were, for instance, whether women were healthier than men, you would show average self-reported health scores for men and women, perhaps for a few different age or education categories. In this first slide, if appropriate, present some bivariate statistical tests of your main hypothesis.
b) The second slide will include some type of statistical test of your hypothesis, with a couple of control variables if they are appropriate. Some kind of multivariable regression is usually appropriate for the second slide. Comment on the fit of the models, the meaning of the relevant coefficients, and so on.
††††††††††† * An appendix slide with all the references that you have cited. You donít absolutely need references and citations in your presentation slides, but if you refer to Jones (2010) you definitely need a reference slide that points the reader to the source for Jones (2010).
* A brief conclusion, and thatís it.
* A small bit of the presentation grade will be for the presentation style. By presentation style, I mean: donít read from printed text. You may refer to notes, but please look up and make eye contact with your audience. Speak loud enough to be heard. Try to limit the use of space-filling words and phrases like ďI mean,Ē and ďUmm.Ē A few of these are unavoidable, just try to keep them to a minimum. Also, please avoid saying ďsorry.Ē
* Your presentation slides should list clearly what dataset is being used, what the relevant (unweighted) sample sizes are, and precisely what variables are being considered or compared. If you use scales or indices, be very clear on every slide where the scale or index is used, as to the range and direction. So: Political liberalism (scaled 0-7, 7 being most liberal).
* Follow the advice of Edward Tufte and make your slides as clear as possible, and ďabove all else show the data.Ē Following Tufte, try not to have large fonts, sentence fragments, 3-D shading, or other non-data ink taking up space in your slides.
††††††††††† * Plots of predicted values are sometimes useful, but plots of actual data are generally better. Predicted value curves and actual data on the same plot can be especially effective.
* Remember to make slides and tables self-explanatory. Label variables and comparisons in a way that would be clear to a general reader. Donít use ipums variable names or Stata jargon in your table.. Be sure to include units for every variable, and indicate the excluded category for categorical variables in regression output.
* Are your data a random or representative sample of some larger population? If your data are from a population sample, identify and define the larger population. If your data are not a random and representative sample (for instance, a convenience sample or an entire population set), explain why not, and explain whether or how you think this lack of representativity limits your ability to do inferential statistics and hypothesis testing.
††††††††††† * Among the problems to avoid: If you frame your presentation around a theory, be sure that you donít ignore situations when the data donít support the theory. This is a class about the data. Donít get so attached to your theory that your confirmation bias leads you to think the data support the theory when the data are in fact ambivalent, or when the data are in opposition to your theory.
2) The project outline comes before the presentation, but I list it here second because the purpose of the project outline is to get you working on the presentation, and so you need to think about the presentation as the goal, and the outline as a step along the way.
*On the project outline due date, upload to Canvas approximately two pages of text and two substantive slides [for Soc 381, this can be up to 6 pages of text, and up to 3 substantive slides]. See the Presentation description above for a description of what the slides should be. The text should deal with the following questions:
a) What is the question? This brief introduction to what is the question can include a brief literature review if you want to, but be sure you have bibliography that lists every source cited.
b) What is the data? How was it gathered? Define the sampling frame. Define the key variables, and explain their relationship to the key concepts you want to test.
c) What do your analyses show?
d) What is your conclusion?
e) What limitations do the data impose on your hypotheses or your conclusions? Understanding the limitations of your data is crucial! Having problematic data is not necessarily a handicap for your Soc 381 project, as long as you understand and explain all the ways in which the data limitations curtail your ability to answer your key question. If you have problematic data but fail to explain why the data are problematic, that is a potentially serious flaw (for Soc 381), and will adversely affect your project (outline and presentation) grade.