Use of Rank Correlation Methods to Determine Food Preferences

Abstract

Certain statistical techniques, useful in collecting and interpreting ordinal data, are described and their applicability in plant breeding explained. It is suggested that these statistical techniques may be particularly valuable to plant breeders in determining taste preferences. A series of experiments are described, investigating the taste preferences of the Iteso of Uganda, who show a strong preference for a mixture of cassava, sorghum and Eleusine millet for food. They do not apparently prefer white sorghum grains for food, nor coloured grains for beer, although such preferences have been recorded elsewhere. They dislike maize and Pennisetum millet. It is suggested that an experimental approach to quality in the breeding of tropical food crops is both desirable and possible, using rank correlation techniques.