Project Proposal.  Your term project will require you to:

1.     Identify a research question that can be answered with a designed experiment.

2.     Choose hypotheses to test, then appropriately select responses, conditions, and material that will best allow you to answer your research question.

3.     Conduct a thorough analysis of the data.

4.     Write an executive summary of your findings.

A few additional notes:

·        You need to approve your project with Prof. Christensen by the date noted on the class schedule.  Projects must be approved by meeting with Prof. Christensen in his office, either during office hours or during another pre-scheduled time.  (Not everyone in the group needs to be present, but having most of your group is recommended.)  Projects that are not approved by the proposal due date will receive at least a 20% deduction in points.

·        You need to work in groups of 3-5 people. 

·        Your design should have at least 2 factors (with interactions).  (That is, experiments associated with the term project will generally use one of the following designs: BF[2], CB[2], or SP/RM[1;1].)  Designs with only one factor will receive a maximum of 85 points out of 100.  Using a design with blocking (e.g., CB[2], SP/RM[1;1]) is a way of demonstrating additional sophistication, but blocking is not necessary for full credit.

·        To assist you in preparing for your project proposal meeting, review the Term Project Proposal Questions document and be ready to discuss these issues with me. 

 

Project Report.  Your executive summary should be no more than 3 pages (single-spaced, 1 inch margins, at least 11 point font) including tables, graphs, and photos.  Your report will follow the basic outline below.

1.     Introduction (approx 0.33 to 0.5 page)

a.     Outline the principle questions of interest, making reference to main effects and interactions (if interactions are included in your model).

2.     Design and Data Collection (approx 1 to 1.25 pages)

a.     Describe the experimental design you are using, including a statistical model for your observations.  A graphical representation of your design (such as the table on page 6.2 of the course notes) is a big plus. 

b.     [Optional] Power analysis to decide how many subjects you would need to detect differences of interest with specified power.

c.      Describe in detail how you gathered your data.  You should give enough detail to make it possible for me to reproduce your experimental procedures using your report.  Emphasize how you used randomization in assigning treatments and carrying out the experiment.  Instructive photos or diagrams are a big plus in illustrating your project.

3.     Data Analysis (approx 1 to 1.25 pages)

a.     Give a well-formatted ANOVA table for your analysis.  (Don’t just paste in raw SAS or R output, which has an ugly font and ugly formatting.)

b.     Interpret the effects in your model.

c.      Give a nicely-formatted interaction plot for at least one interaction, even if it is non-significant.  (If you have multiple interactions, choose the most interesting one(s)).

d.     Choose which means to compare and interpret such comparisons using appropriate methods.

e.      [Optional] Consider contrasts of interest to you.

4.     Conclusions (approx 0.33 page)

a.     BRIEFLY summarize important findings.

b.     Discuss ways in which your findings can be interpreted and generalized.  Is there a cause-and-effect relationship?  What populations can you reasonably draw inferences about?

c.      How might a future study be refined and/or expanded with extra time or resources?

5.     APPENDIX – includes only the following items (and does NOT count toward your 3 page limit)

a.     Well-commented code used for all analyses

b.     Data sheet (print out from Word, Excel, SAS, etc.)

 

** To turn in your project, deliver a complete copy of the report (including the Appendix described above) to the usual HW drop-off location.  Also, EMAIL a copy of your report and your data file to me william@stat.byu.edu

 

Project Grading.  The project grade sheet that I will be using is here.  Read this carefully!  But, in a nutshell, the following factors will yield a high project grade:

·        The report uses clear writing that is free of grammatical and formatting errors.  The document looks professional in every way.  NO report covers—just a staple in the top left corner.

·        Precise language is used when describing the design and the experimental procedures.  (If it isn’t clear what you did, I have to assume that your approach was also sloppy.)

·        The project report demonstrates that the students approached the data collection with sophistication and seriousness. 

·        The correct analyses are used, with a proper ANOVA table given.

·        Proper methods are used when comparing means.

·        Conclusions and interpretations of experiment are thoughtfully summarized.

 

Ideas for Term Project

·       Some Past Term Project Experiments in Stat 230.pdf

·       Stat230ProjectIdeasfromKahneman.pdf -- For those interested in psychological phenomena, I'm posting a summary of some studies described in Daniel Kahneman's fascinating book Thinking, Fast and Slow.  If you'd like to read more about any of these experiments to get ideas for your term project, you are welcome to borrow my copy of the book for a day.