1. For discussion during lecture #6:
Problems from the text: 2.8, 5.2 (for part c just calculate the
ICC, not its SE), 5.4, 6.3 (not c and d) , 6.11.
Suppose you are investigating
the efficacy of pasteurization of milk by regional dairies over a period of
several months.
Considering a completely randomized design, describe a scenario
in which a fixed-effects model would be appropriate,
and also describe a
scenario in which a random-effects model would be appropriate.
2. For discussion during lecture #14:
i) For the data in problem 2.1, select a set of contrasts and use SAS to show that the SS for the set of contrasts equals the treatment SS. Also find a set of orthogonal contrasts, use SAS to calculate the SS for each contrast, show that the sum of the SS's equals the treatment SS. ii) 6.11a, b, d iii)For the data in Table 6.13, assume all are random effects. Use the computer to get SS's then conduct an approximate F test for temperature manually - check your results against the computer. iv) 7.3 v) 8.1 a,b,c,e,f vi) 8.5 a,b,d vii) 9.1 a,c,d,e
3. For discussion during lecture #22:
Problems from the text: 9.6, 9.7, 14.6, 15.1, 15.5 (a-g, also Proc Mixed modeling)
4. For discussion during lecture #30:
Problems from the text: 11.1, 11.5, 11.7, 12.2, 12.6
5. For discussion during lecture #36:
Problems from the text: 17.2 (on d, you don't have to calculate the average
standard error),
17.1 a-g (Part c here same as d on 17.2; Also compare strength means and test for differences
at the three quartile values for weld diameter)