Data for the wine evaluation experiments in Brien, May and Mayo (1987).

Four experiments (2--5) are discussed in Brien, May and Mayo (1987). The discussion in this paper is based on Brien (1980)

The data for the four experiments is available in the following files that differ in their formats:

R scripts that produce the anatomies for the design used is available in the following files:

A series of four previously published experiments concerned with the effects of viticultural treatments on wine quality were to be tested. There were gaps of one year between each of the first three experiments and two years between the third and fourth. In each experiment, viticultural treatments were applied to plots; the grapes of each plot were made into a separate lot of wine by small-scale winemaking techniques. The wines were then evaluated by at least six experienced wine-makers who used the judging system used for Australian Wine Shows, with which they were familiar. In this system, a single composite score is assigned to each wine within the range 0 to 20 (the maximum being composed of three points for color, seven for aroma and ten for taste characteristics). Every sensory experiment was a two-occasion experiment, i.e. all the wines were presented to the judges on each of two occasions, the second occasion differing from the first only in the order of presentation of the wines for the first 3 experiments. In the last experiment the wines also differed in the glasss type (clear [1] or black [2]) in which the wines were presented. Each occasion consisted of several sessions, at each of which 8 or 12 wines were presented for simultaneous evaluation. Wines were assigned to sessions so that all treatments were equally replicated at a session. See Brien, May and Mayo (1987) for further details.

The scores assigned in evaluating the wines, and the factors indexing them, are given in the files. The factors in an experiment are in each case a subset of the following factors:

  1. Judges
  2. Tables
  3. Occasions
  4. GlassType
  5. Sessions
  6. Glasses
  7. Blocks
  8. Plots
  9. Halfplots
  10. WineReps
  11. Bottles
  12. Treatments
  13. Rootstock
  14. followed by a subset of the following response variables
  15. Colour
  16. Bouquet
  17. Palate
  18. Score
  19. Riesling
  20. Sultana
  21. Shiraz
  22. Mataro

The data is ordered so that the factors Judges, Occasions or GlassType, Sessions, and Glasses are in standard order; the remaining factors are in randomized order.