Due by noon Monday, November 22, 2004. You may work with one partner on this project.
For this asignment, you will obtain a data set, articulate two or three interesting questions about your data, and use the data analysis techniques from this class to provide some answers to your questions.
Your first job is to collect or find a data set on something that interests you. Your data set needs to include enough variables, both categorical and numeric, to support the use of a variety of the data analysis techniques you have studied this term. Your data should also include at least 30 observations. Since you will be performing inferential statistics, the observations should, ideally, be a random sample from a larger population. For example, your observations could represent people, players from a league, cities, days, trees, dogs, cars, etc. If you find an interesting data set that does not seem to be a random sample from a larger population, please consult with me about its suitability.
Your report should begin with an introduction describing the general research question or questions you have in mind. The introduction should grab your reader's interest by explaining why your data set and associated questions are interesting.
Your introduction should also describe the general features of your data set. Where and how did you get the data? What fields does it have, and in what units (if any) are they measured? How were the observations collected (assuming you know)?
Your analysis should use the following as appropriate:
Descriptive summary statistics for each variable (mean, median, standard deviation for numeric variables, counts for categorical variables, etc.).
Exploratory graphics (barcharts, histograms or stemplots.)
Exploratory analysis of relationships between variables (scatter plots, side-by-side box plots, grouped bar charts). You don't have to analyze every possible pair of variables (just the ones it makes sense to compare).
Use at least two of the following procedures:
Confidence intervals.
Two-sample t-test (or matched-pairs if relevant).
Least-squares regression (with and without information from one of the categorical variables). Don't forget to include discussion of residual plots.
Chi-square (for association or goodness of fit).
Discussion on any unusual features (outliers, extreme skewness, etc.)
End your report with a page or two of discussion and conclusion. Summarize your findings, criticize your methods, make your assumptions (especially the questionable ones) explicit, and discuss briefly how you might proceed if given more time and resources.
Ultimately, I want you to write this report as an honest persuasive essay. You will use your statistical analysis of the data to argue in favor of one or more conclusions. (The honesty is especially relevant if your analyses are inconclusive--I want persuasion if it makes sense, and admission of inconclusive results if not.)
Your grade will be based on:
Correctness.
The analysis of your results. Make sure to include a discussion of each procedure you perform, making clear what questions you are trying to answer, and evaluating the statistical results in the context of those questions.
The quality of your exposition. This includes clarity of expression, organization, mechanics (grammar, spelling, etc.), and so on.
Presentation. Is your report neatly presented? Are the graphs arranged well and labeled clearly? Have you eliminated unnecessary SPSS output?
Creativity and critical thought. Is the project carefully thought out? Are the research questions reasonable? Are the tests performed appropriate to the questions asked? Are the questions interesting?
If you question the suitability of your chosen data, please talk to me about it. A poorly chosen data set will make it hard for you to write a good report.
Here are some places to look for data.
Start early, have fun, and keep in touch.