The aim of the project is to follow a statistical analysis from the initial idea through
to the final report. The detail at each stage will of course depend on your involvement
in the project. If necessary, you can use statistical techniques beyond those covered
in this course, but talk to me first. A well explained simple analysis will be marked
better than a poorly explained complex one. You should do your own analysis of the
data, but if you use data from others, you can refer to their papers for methods. You
can also critically analyse their methods.
I’m not interested in computer output but you can copy tables and figures etc from the
output to include in your report. A concise project will receive more marks. Try for four
or five pages. If appropriate, you can refer to the literature for detailed background.
I don’t really care what science problem the project addresses, but it should allow you
to demonstrate your skills. One way of thinking of it is “If someone audited my project,
could I justify the whole thing?” All projects are different, but something along the
lines of the following
1. Context of the problem, including hypotheses. What did it plan to do? (You
may wish to have some simple hypotheses to address in 6.)
2. The statistical model (or models).
3. Experimental design.
(a) Type of experiment.
(b) Sampling procedure.
(c) Data collected.
4. Data collection and management.
5. Exploratory analysis.
6. Confirmatory analysis. Results of hypothesis tests, confidence intervals etc.
7. Final report. What was found? Eg. an executive summary for management, or
a report for your collaborators to use to prepare a paper for publication.
8. Including a screen dump of your SAS process flow as an appendix is appropriate.
Data sets from the book “Data” by Andrews and Herzberg, if you need it in this website