Interviewing is the most frequently used technique for collecting data, but it
has its share of methodological problems. Thailand has few studies on methodology
regarding the effect of interviewer characteristics on the data quality, and so there is
little empirical research on methodology development or recommended changes for
improving the existing method of social survey.
The three objectives of this study are 1) to assess the quality of survey data
regarding item non-response, accuracy and response reliability, 2) to investigate the
effect of interviewer and respondent characteristics and social environmental attributes
on the quality of survey data, and 3) to investigate what are the determinants affecting
data quality as regrads the efficiency and effectiveness for contact strategy. This study
focuses on data collection of the Kanchanaburi Project of the Institute for Population
and Social Research Mahidol University with support from the Welcome Trust of the
United Kingdom. I used multilevel analysis and structural equation (path analysis).
I found that the interviewer being of a different gender than the respondent or a
third person presence at the interview adversely affected all three categories – item
non-response, accuracy, and response reliability. Other factors adversely affecting one
or more data quality indicators included: interviewer speaking a different native
language than respondent or having less interview experience, interview being held in
the morning, and respondent being younger or of lower education.
Ultimately, this study serves to show that characteristics of interviewers and
respondents and some social environmental factors affect the quality of survey data.
Results of this study would be used in setting up guidelines for interviewing to account
for the interviewer’s potential effect on the process, for example the effect of
difference of gender, language, etc. These serious problems may affect the objectivity
of research. A survey should start with a well thought out study design that includes a
method to monitor the quality of the survey.
|