The Family Expenditure Survey is a sample Survey, and it is therefore necessary to "inflate" the results of the sample in order to obtain estimates for the entire population and its subgroups.The numerous cases of "drop out" - households which should have been investigated but which, for different reasons, were not - can cause a bias in the Survey Estimates. This is specifically true with differential drop out rates for the different characteristics according to which the differences in household budgets were investigated. In order to minimize these biases as much as possible, the "inflating" process described below was set up.

1. Determining the Inflating Groups - The goal was to establish groups of households which would be as homogeneous as possible in their consumption patterns and which could be determined according to as few household characteristics as possible. These groups would serve as "inflating groups". For this purpose, a detailed analysis of the data of the 1979/80 Family Expenditure Survey was conducted. The characteristics according to which the division into groups was made were : size of household, size of dwelling and locality of residence. The division into groups was made separately for Jews and non-Jews. This division into inflating groups was made according to a hierarchical basis of the division characteristics, where size of household was the primary division characteristic. The others were determined according to an integration of the importance of the characteristic and the size of the group.

2. The Estimates of the Size of the Population in the Inflating Groups - There were no "external" (to the Survey) data on the distribution of households in the population according to the inflating groups which were established and which were revised for the Survey period. Therefore, a special operation was conducted in order to assess the required data, based on the current Labor Force Surveys (which are based on relatively large samples) and on the 1983 Census of Population and Housing.

3. Determining the Inflating Coefficients - All households in the sample which belong to the same inflating group received an identical "inflating coefficient" which expresses the number of households in the population which the household in the sample represents. This coefficient was calculated as the ratio N/n, where N is the estimate of the number of households in the population which belong to inflating group k, and n is the number of households belonging to group k which participated in the final processing.

4. Calculating the Estimates - The estimates of the average expenditure per household was obtained as a total of all the "inflated" expenditures of the households in the sample, divided by the estimate of the number of households in the population. The estimate of the percentage of consumption of a consumption item out of the total consumption was calculated accordingly as the total "inflated" expenditures of the consumption item divided by the total "inflated" expenditures for total consumption. The parallel estimates for income items are calculated in a similar manner. Other estimates, e.g. distributions (percentages) of households by classifying variables such as size and origin, ownership (percentage) of durable goods, housing density, etc. were calculated accordingly as ratio estimates and were obtained by dividing the "inflated" numerator by the appropriate "inflated" denominator.