STA429/1007 Assignment 7
Quiz on Thursday Nov. 11th
This assignment is based upon lecture material on random effects and nested
designs. Treatment in the online text is sketchy.
- Make up a study, preferably in your field, with two
independent variables in a factorial (non-tested) design. In this
study, both main effects (and as a consequence, the interaction)
are random, not fixed.
- Make up a study, preferably in your field, of a study in
which all effects are random, there are at least two factors, and
each factor is nested within the preceding factor(s).
- The quality control division of a wholesale grocery company
is investigating its canned corn. From a fleet of 840 trucks,
they sample 50 trucks at random. Each truck contains 580 cases of
corn, and they sample 12 cases at random from each of the 50
trucks. Each case contains 36 cans, and 3 of those cans are
selected at random and closely examined. The number of worm
fragments in each can is counted, and that count is the dependent
variable. These data are available in the file
worms.dat.
- What is the unit of analysis (case, sampling unit)?
- How many independent varables are there?
- Is this design balances? Answer Yes or No.
- Estimate the sources of variance in the number of worms
per can. What percentage of the variance is due to trucks?
Cases? Error? If you want to do the estimation two differenty
ways, that's fine.
- Test the effect of trucks and cases. For each, get an F
and p-value. Why are you not testing the effect of can?