The third for-credit "Statistics in Action" session will be held in class on Wednesday, week 8.
You should prepare for the session; it will be based on the situation outlined below. Before class, read the outline and get together in your group to discuss the problem. You should also use SPSS to analyze the data: you will need the RESTAURANT data file. Then, when you come to class you should be prepared to participate in a class-discussion, using what you've already discussed in your group. You can (and probably should) make notes on what you've discussed in your group before class. You should bring these notes to the class-discussion.
As a slight change to previous statistics in action sessions, there is something to turn in for this session by 4pm Tuesday, week 8. Send the instructor (by e-mail at ipardoe at lcbmail.uoregon.edu) the R2 and s values (accurate to 3 decimal places) for your group's "best model" (see below), together with sufficient information for the instructor to replicate your results, i.e. which predictors, interactions and/or transformations are in your best model. If you do not send this information to the instructor by the deadline your group will automatically receive zero credit for this SIA session. (The instructor will e-mail you to say your results have been received, so do not assume your e-mail went through successfully until you receive this confirmation e-mail.)
Grading for the sessions will be on a zero/full credit basis. Each member of a group will receive full credit for that session if the group obtained a model almost as good as (or better than) the instructor's "best model" (i.e. R2 at least as high and s at least as low), or, if not, if at least one of the group makes some relevant remark in the ensuing discussion. If the group was unable to come up with a model as good as the instructor's and no-one in the group makes a useful contribution to the discussion everyone in that group gets zero credit for that session. Groups that fail to e-mail the instructor their model results on time or who do not provide enough information to replicate the results automatically receive zero credit.
In class, we'll discuss building a regression model for this dataset and also anything else that comes up that you think is relevant or interesting in the context of the problem. To keep the class-discussion orderly and the grading fair, you must raise your hand before saying something. The instructor will ignore anything you say unless you've raised your hand first and been asked to speak. The instructor will do his best to allow the first person to raise their hand the opportunity to speak each time. If you keep your hand up, you will be given the opportunity to speak once the current speaker has finished making their point.
When you make a relevant observation, suggest a useful approach to answering a question, or raise an interesting question not previously considered, the instructor will make a note of which group you are in, and keep a tally of which groups have participated and which have not. Remember, you only need come up with a model as good as the instructor's, or, failing that, make one relevant remark to get full credit for your group. The instructor will decide what is relevant and what is not, and his decision is final - no arguments.
You've been asked to find out how restaurant profits are affected by certain characteristics of the restaurants. You have data on 39 restaurants, and would like to build a regression model for predicting Y = annual profits (in thousands of dollars) from 5 potential predictor variables:
Build a suitable regression model. You may want to consider the following topics in doing so:
You may use the following for terms in your model: