Project for Biostats Course VHM 801 at AVC
The last home assignment (worth 15% of the course mark) may be replaced
by a "project" consisting of the statistical analysis
of a suitable dataset of your own, as well as a report
describing the analysis and drawing conclusions. It is
optional to replace the last home assignment by a project, but if you want to
do so the project needs to be prepared in advance of the last home assignment.
Choice of project data
Maybe you have already thought about a project related to your own
research for the course project. One requirement is that the data are
"new", in the sense that they have not been analysed statistically before by
yourself or any of your colleagues. Furthermore, the project should be of
suitable size and difficulty, and the methods of the course (Sessions 1-11)
should be applicable to the data. If your
dataset is too large or complex, the project may be confined to a
subdataset and/or a simplified objective. As a general and rough rule,
the data should usually contain at least 30 and not more than 200-300
observations (cases, subjects, or worksheet rows).
I'll discuss individually with each of you
whether your dataset is suitable.
Outline of project work
- At least one week in advance of the day the 4th home assignment is
handed out, you'll need to submit a project description, see below.
- After having read the description and discussed it with you, I'll
formulate a project objective, which should guide you in the data
analysis and the writing of the report.
- The project report (see below for suggestions on how it could be
formatted) is due on the same day as the 4th home assignment, and will
be marked by a scheme as similar to that of the home assignment as
possible.
Project description
You write 1-2 pages containing answers to (some of)
the following points, and give it to
me no later than one week before the 4th home assignment is handed out:
- Title (preliminary),
- Short introduction: problem and experiment/data collection,
- Data:
- how many observations/experimental units?
- which response variables (outcomes) measured for each observation?
- which explanatory variables (predictors) measured or recorded for
each observation?
- Purpose:
- which relations or comparisons would you like to study?
- which hypotheses would you like to test?
- your expected findings (if any), based on the literature and your impression of the data,
- Statistical model (draft):
- which are your response and explanatory variables,
- proposed type of statistical analysis,
- perhaps mention of some points that might complicate matters.
Report
Generally speaking, the report should follow the format of an article, e.g. containing the
following sections: Summary, Introduction, Materials and methods (including
statistical methods), Results, Discussion/conclusion, References.
It is important that you
describe and document the statistical analysis in detail (more than would
be needed in an article for an applied journal within your field). You
are required to submit an electronic version of the dataset (e.g. a Minitab worksheet)
together with the report, and the report should in an appendix contain some
documentation of the analyses (e.g. Minitab prints, or a Stata do-file).
Henrik Stryhn
(hstryhn@upei.ca) 2018-10-08