Where are we so far?
Our first two days have focussed mainly on the first of three components
of statistical studies:
- Collect data (statistical design)
- Organize data (data analysis)
- Draw conclusions from data (statistical inference)
Here are some highlights:
-
Many problems with statistical studies are not mathematical.
(7 critical components)
- Most studies "do the math" correctly.
-
You can't do good statistics unless you start with good data.
- Measuring can be non-trivial (validity, reliability, bias, variability)
- People are complicated (to get data about)
-
Good samples must be
- representative (use randomization to avoid/reduce bias),
- large enough (to reduce variability)
-
There are fundamental differences between observational studies and
experiments.
- Cause and effect can never be established on the basis of an observational
study alone.
- By randomly assigning treatments to individuals, experiments can give
strong evidence of cause and effect relationships.
Where to next?
Phase Two (data analysis): What do we do with all this data once we have it?
We will continue to revisit data collection issues as they arrise.
Last Modified:
Thursday, 11-Jan-2001 16:02:18 EST
Maintained by:
Randall Pruim
Back to Math W50 Home Page