© 1998 World Association for Public Opinion Research
research-article |
SURVEY DATA COLLECTION METHODOLOGY FOR ELECTION MISSIONS: HOW PROFESSIONAL FORM DESIGN HELPS TO CONTROL A SURVEY
Correspondence to Statistics Netherlands (StatNeth), P.O. Box 4481, 6401 CZ Heerlen, The Netherlands; on questionnaire design methodology to Dr. J. C. Akkerboom, Questionnaire Design Resource Center, Department of Data Collection Methodology (GWM), and on election monitoring methodology to J. J. G. Schmeets, Department of Social-Cultural Surveys (KPE), HAKM{at}CBS.NL and HSHS{at}CBS.NL.
The present case study discusses some difficulties in applying survey methodology to the observation of elections by international observers. In election missions data collected through observation forms should justify general statements on the freedom and the fairness of the election observed. The paper focuses on observations of the voting during visits to polling stations and the observation forms used to collect the information gained in these visits. Important differences between election observation surveys and regular self-administered surveys are, to varying degrees, lack of control over (a) sampling, (b) what is to be observed and reported, and (c) data processing and analysis due to time pressure and logistic circumstances. It is shown what progress has been made on points (b) and (c) between the 1995 Russian State Duma Elections and the 1996 Russian Presidential Elections. The whole statistical process was reviewed, various experts were consulted, and redesigned forms were tested. The resulting new forms in general conformed to the Jenkins and Dillman (1997) principles referring to (o) form content, (1) instructions and explanations, (2) form compositon, and (3) guidance through the form. Theoretical and empirical evidence asserts that professional form design helps to optimize response rates, increase reporting accuracy, and decrease processing times. In our study, form return rates increased from ca. 50 percent to ca. 90 percent, data quality improved, and preliminary results were available within 48 hours.