The results of an unpublished Alzheimer's study are reported in Hand and Taylor 
(1987). There are two groups of participants, one receiving a placebo 
and one receiving a drug called lecithin. Each participant took two 
tests to measure memory and recall ability. During each test, the 
participant was asked to memorize a list of words. The outcome was 
number of words correctly recalled a moment later. The difference 
between the two tests was the list of words; the first test used the 
same word list every time, while the second test used a different but 
'equivalent' word list every time. The tests were both given at each of 
five time points: 0, 1, 2, 4, and 6 months. There are 48 patients, 22 
randomly assigned to drug and 26 to placebo.

It is expected that any treatment effects may be small and slow to
appear because of the nature of dementias. The administration of the
memorization "test" on multiple occasions also introduces a "practice
effect," that is, a subject may improve over time merely because of
carryover effects from one occasion to the next. This may be due to
repeated exposure to the same words (for the first test only) and/or 
repeated exposure to the testing situation (for both tests).

Questions of interest are:

  -- Is there a time trend in recall ability across the five testing 
       occasions?  Is it, for example, decreasing?
  -- Does any time trend differ by treatment?
  -- Does any time trend differ by test type?

The data set contains treatment in column 1 (0 for placebo and 1 for
lecithin), test 1 results in columns 2 through 6, and test 2 results 
in columns 7 through 11. There are no other covariates available.
This is a doubly repeated measures data set; see also alzheim.readme
and alzheim.dat for the data using only the first test.

Data source: Hand and Taylor (1987). Multivariate analysis of variance
and repeated measures. London: Chapman and Hall, Inc.