98
MORRIS I. BERKOWITZ
obtained would be a good sample of households with young children but not of all households. Accepting that limitation, we sampled from this list with the intention of "snowballing" our sample as we continued the work.
The purpose of the household head sample was primarily for the gathering of research data but secondarily it was concerned with increasing the spread of our total effort. We had early decided to study the reaction to the resettlement of several segments of the population, divided by social status, and we used the household heads as our data source for the other statuses. We were concerned with studying families across the entire age dimension and settled (because of the known patrilineality of the Chinese family) for asking complete information about all of the members of the household head's family. The groups we were primarily interested in were: Household heads; Housewives; Children; Aged; Young Unmarried Adults; and other Household Heads without school age children. We assumed that resettlement would have different impact on different status groups and therefore each of these needed separate examination.
Once having assembled our list of household heads and randomly sampled from it, we faced our second major methodological difficulty: information contagion within the resettlement area. We developed and pretested our basic instruments on other Hakka speaking people, and then devised a plan for "sweeping" the resettlement area for all of our primary interviews in the shortest possible time to limit the spread of the news that research was going on and the communication through the gossip of the villagers of the content of our instruments. Two days (using sixteen interviewers) was the allowed time for the collection of the scheduled 35 interviews. More than half were completed within the first evening. The schedule itself is a bulky one, with 129 questions, not all of which, however, would be administered to any one respondent. We had trained our interviewers to be prepared to administer it in an average of 45 minutes. As it turned out, the average time for Household Head and Housewife interviews was 58 minutes (N=35). The schedules had been prepared in English, translated and practiced in Cantonese, and frequently had to be administered through the use of a translator: a total of 14 translators were used, most frequently the children or other relatives of the household head.