Quality & Quantity (2005) 39:559–579 © Springer 2005
Who Perceives the Collective Past and How?
Are Refusals on Open-ended Questions
Department of Political Syst. and Political Sociology, University of Stuttgart; Breitscheidstr.
2, D-70174 Stuttgart, Germany, E-mail: firstname.lastname@example.org
Many textbooks dealing with empirical social science or survey research
give only a brief overview concerning the use of open-ended questions.
Usually, the authors remark some advantages of the format and contrast
it with the closed-ended one. It is said that open-ended questions serve
the exploration of current changes in attitudes as well as the discovery of
frames of reference unknown to the researcher, the test of knowledge levels,
or the examination of thought processes as well as linguistic usage (Bab-
bie, 1973: 127; Bailey, 1978: 106f.; May, 1993: 95; Diekmann, 1995: 409).
Despite its acknowledged usefulness, this question type is seldom applied
(Kromrey, 1980: 352; Caplovitz, 1983: 119). Unwillingness of scholars to
work with this measurement instrument may be put down to the fact that
the closed format has several advantages especially with regard to costs and
time (Converse, 1984: 270; Fowler, 1995: 178; Schnell et al., 1999: 395).
Furthermore, the validity of open-ended questions is in dispute because
the missing rate of this measurement instrument is almost always inﬂu-
enced by degree of education (Schuman and Presser, 1981: 289; Roth, 1995:
164). Therefore, some scholars assume that this question type is an indica-
tor for the level of education (Craig, 1985: 65f.; Stanga and Shefﬁeld, 1987:
833). In this view, people who only went to school for a few years would
have limited ability to give well phrased answers. As a consequence, it is
in doubt whether attitudes can be measured by this question type in gen-
eral. In opposition to this hypothesis, John Geer (1988) insists that there is
a substantial relation between readiness for answers and the contents which
social scientists are interested in. In his view, people with little education
have simply little interest in subjects like politics, history, etc.
An empirical test of these assumptions has not been done, yet. No
doubt, if one wants to test for validity of the format one has to control for