THE VALIDITY OF FIELD DATA.

Abstract

In recent years all manner of research instrurnents and techníques and the quality of the data they yie1d have been placed under critical scrutiny. This artic1e brings a fresh perspective - the sirnultaneous comparison and evaluation of data secured with a wide variety of instruments in a single community setting-to some of the fundamental problems and questions involved in securing va/id responses. It is asserted that securing valid responses consistent with the behavior and phenomenology of the respondent in ordinary non-research situations is a sina qua non of the other forms of data oontrol to which socia'! scientists have tended to limit their attention. Because severa'! fie1d workers utilizing a variety of techniques were continuously able to check, re-check and cross-check information gathered from a particular respondent, a number of types of errors and sources of rnisinformation were revealed. These errors and misinformation result from purposeful intent, the temporary character of the tole of the respondent, the psychological characteristics of the individual respondent and from the involuntary inability of the respondent to meet the demands of the interview situation. Given these sources of error and misinformation, the problem confronting the analyst and theinterviewer is almost overwhelming. TraditionaHy the anthropologist has coped with this problem by assigning different weights to the responses of different respondents and to the responses from a single respondent gathered at diferent times and under differenr circumstanoes. In doing this he can never be sure that other procedures or further probing might not have yielded different information, and he has difficulties in objectively supporting the validity of his interpretations. But the same social psychological apparatus which produces different levels of response in free and depth interviewing also operates in other types of field instruments. The central problems lies in the fact that al'! answers to the same question in standardized interviews are not ofequal weigÍft, and cannot be treated as suchv In assigning equal weights the analyst simply adds up the conscious and unconscious misinformation, bias and accuracy and treats them aH as equal.Poll-type surveys secure and provide information in those specialized areas of mass society where otherwise such information is not readily available. Such surveys are valuable when they probe re'latively simple areas of choices among current alternatives avaílable at the public level; for example, presidentíal pol1s. However, when the object of research is to study at bottom the dynamics of a community or an institution, and where the problem of social and psychological levels of response is crucial to the research problem itself, other techniques are indispensable: forexampLe, in totalitarian societies even political and communications polling does not provide valid results since opinions which lead to such data are not adrnissible at the public level.