Next: Data Disk
Up: Other Important Stuff
Previous: What Do I Put
- 1.
- Almost always when doing an observational study, such as a survey, students
have trouble getting the sampling right. This can also occur, but is less
common, in selecting experimental units for an experiment. Unless you
have a frame (list of
sampling units in the target population), and deliberately sample the
units from the frame by a probability sampling scheme, you are almost
certainly obtaining a suspect sample. If this is the case, in your final
project report you had better justify your sampling method (this can be
often be done in terms of lack of time or other resources), tell what is
wrong with it, and tell what you should have done to get it right. If you
are not clear about this point, see me.
- 2.
- Validate your sample whenever possible. This means collecting
data on the sampled units that can be compared to known population
quantities to help establish the representativeness (or lack thereof) of
the sample. For example, if your sample of WPI students is 60% female,
you would rightly suspect that the sample is not representative of all
WPI students. In your final project report, detail validation of your
sample, or tell why it was not possible to do it.
Next: Data Disk
Up: Other Important Stuff
Previous: What Do I Put
Joseph D Petruccelli
5/26/1998