Appendix C

Random within house selection techniques

Kish (1949) selection tables

  data collection face to face surveys appendix c image 01

The tables provide unbiased estimates by giving each respondent a weight based on the number of adults in the household. This guarantees that the selection within a household is random for a combined total random sample across the housing units (addresses) that were randomly selected in the first place.

Instructions for Kish tables

  1. Assigning Kish tables to the sample file: one of the twelve tables is randomly selected and assigned to the first line in the sample file. The series of twelve is then run through twice, assigning tables to the sample lines. Then again a table is randomly selected and the series is run through twice. This procedure is repeated until all sample lines have an assigned Kish table.
  2. Household listing: a household listing of eligible adults (age 18 and over) who reside in that household is taken at each of the sample addresses. Usually the males are listed first in order of decreasing age, and then the females in the same order.
  3. Using Kish tables: the table assigns a number to each member of the household listing. Sample Kish tables are shown above. In the first column the interviewer would circle the total number of eligible persons. The corresponding number in the second column of the Selection Table denotes the person selected to be interviewed.