De meting van politieke voorkeur: een methode tot vermindering van fouten
In: Acta politica: AP ; international journal of political science ; official journal of the Dutch Political Science Association (Nederlandse Kring voor Wetenschap der Politiek), Band 12, Heft 1, S. 90-110
ISSN: 0001-6810
The value of voter behavior surveys lies in the possibility of following the changes of the voters' preferences. In this type of survey, errors are found that are due to specific problems of sample composition. The problem is that one cannot weigh the sample with respect to the variable 'past voting behavior'. Rs answers to the question: "Which party did you vote for in the last election?" cannot be accepted at face value. NIPO includes that question in its weekly survey. These surveys are studied to discover how to protect against mistakes arising from a nonrepresentative sample in terms of past voting behavior. The method involves a comparison between actual electoral returns & voters' statements regarding past voting behavior as well as a weighing of the sample based on the results of the weekly survey from the last sixteen weeks. In Comment, J. W. Foppen (Erasmus U, Rotterdam Netherlands) objects that there are serious weaknesses regarding the NIPO surveys which de Hond uses as his bases. Moreover, de Hond's own work based on these surveys suffers from statistical blunders & computational errors. In Remarks, de Hond answers that the above comments either are not applicable to his own work or else can be directed at all survey research work. They are also based on misunderstandings & an unsympathetic reading. 7 Tables. A. Orianne.