adult households surveyed in Kingston. But if this is the explanation of the discrepancy
between the two estimates, it follows that a group of households whose income elasticity
of demand must be much lower than the general average was in effect excluded from the
survey sample. It must be remembered that the weighting appropriate to the averaging
of elasticities over different sections of the population is proportional not to their num-
bers, but to their expenditure, and this will give the under-represented wealthy consu-
mers a much higher weight than they would receive on the basis of their numbers. These
considerations seem too important to be ignored, and although the allowance for the under-
represented income-groups can only be made on a very rough basis it has been assumed
that its effect is to reduce the elasticities applicable to the whole population in 1958 by
about 10 to 15 percent as compared with the weighted averages resulting from the calcula-
tions previously described.