This study is based on data collected under auspices of the Food Council of The Netherlands, between October 1966 and September 1970, from 2000 households in The Netherlands. It is a critical analysis of this particular investigation in food research and of food research in general, in which emphasis is placed on the statistical aspects.
Of the data collected only those relating to the use of milk and milk-products have been statistically analysed. The data on milkusage were obtained in 1966 and 1967: for each of 41 weeks approximately 50 households were sampled, 2000 households in all. Only usage at home' was of interest. For each week the daily usage patterns were obtained both for each household and each household member. A detailed breakdown for each type of milkproduct was obtained only per household.
This research is subject to a number of important limitations; these are discussed further in the chapters 1, 2, 3 and 5. In chapter 5 the choice of method of analysis of the household usage is motivated in detail.
As there were serious doubts as to whether the sample was representative of The Netherlands, arising for example from the 50% refusal to cooperate, statistical tests have not been used. The analysis is purely descriptive therefore and the most important results are illustrated by tables. This is, as Tukey put it, an exercise in data analysis. Differences between levels of household characteristics were not in disagreement with the findings of other food research but the absolute levels could not be meaningfully related (chapter 9).
In chapter 6 the weekly intensity is analysed. In chapter 7 the daily usage is examined. In chapter 8 the results of the enquiries made on the usage patterns of 9 milk-products on Sunday, Saturday and the other weekdays together, are given.
Besides making a large number of crosstabulations, a number of multivariate analyses were performed. Although the clearest results of the principal component analyses seemed to be valid, the results were of restricted value. Without other information, it is not clear how much value can be attached to them. In applying the cluster analysis although a few distinct usage groups could be formed on the basis of their patterns of milk-product usage, not all households could be assigned to a well defined group (chapter 8). The results of the AID-method were interesting (chapter 7). In general, apart from the association of frequent usage of most milk-products with households having children the associations established were not great.
In chapter 9 a review of other possible analyses is given, and the prospects for future research are considered. It is clear that much fundamental statistical research into the design of food research projects has still to be done. With the material of the present research it was not possible to do much in this direction.
Many results could not be given in tableform. Further information may be obtained from the Food Council.