The non-declared vote in the surveys: The Spanish case in the 1980s

Abstract

Do individuals conceal their voting intentions? If so, why? In this article the author attempts to answer both these questions through a case study of declared voting intentions for the two main Spanish political parties in the 1980s. His analysis shows that the distribution of the hidden vote has not remained constant over time. Whereas in the early 1980s conservative voters were more likely not to disclose their vote, later in the decade a number of decisions taken by the Socialist government led some leftwing voters to hide their partisan preferences from pollsters.

Introduction

The use of pre-electoral surveys poses numerous problems. This articles focuses on two of these: first, the lack of correspondence between declared voting intentions and actual election results; and second, “the limited statistical analysis of non-responses or missing data” (Varela et al., 1998: 274–275). The first problem, namely opinion polls' failure to provide accurate predictions of the distribution of the vote has fuelled criticism and distrust of survey results. However, the disparity which often exists between predictions and reality cannot simply be attributed to statistical error. On the one hand, problems may also arise from the misrepresentation of poll results by the media.¹ On the other, with respect to the incidence of non-responses to some questions, researchers are only just beginning to develop convincing statistical solutions to the problem (Berinsky, 1999, Berinsky, 2002a, Berinsky, 2002b, Breen, 2000, Varela et al., 1998).

In this article I will address both these problems. I will try to discover pre-electoral surveys sometimes fail to reflect the real distribution of partisan preferences. More specifically, I will base my argument on an analysis of non-responses to questions about direct intended vote in various studies produced by the Spanish Centre for Sociological Research (Centro de Investigaciones Sociológicas, henceforth CIS). In these surveys, many respondents take refuge in the “Don't know/No answer” response to this question. Why is this so? In some cases, these respondents are deliberately concealing their partisan preferences. In others, their response will reflect genuine indecision. Significantly, however, both the hidden vote and indecision are not always uniformly distributed among voters for the various parties. That is, respondents sympathizing for a given political party may be more likely to remain silent than those supporting one or various of its competitors. This may be because the voters for the party in question are less sure of how to vote, or because they are more likely to hide their vote, than other voters. In this type of situation the sample will be biased if we treat these individuals as missing cases. This is the key issue I am going to examine in this article.

Spain constitutes an exceptionally interesting case through which to study this issue. As will become apparent below, the nature and scale of the non-declared vote in Spain has changed significantly since the return to democracy, as a relatively consistent reticence to reveal partisan preferences for rightwing parties has given way to a pattern in which the undeclared vote is more closely linked to particular situations and circumstances. Nonetheless, these developments have not been analyzed in the literature. Moreover, some Spanish authors have gone as far as to deny the existence of a hidden vote in Spain altogether (Arango and Díez, 1993). For both these reasons, the Spanish case would appear to be particularly relevant for the study of the hidden vote.

The article is divided into three parts. First, I will analyze participation in surveys and the declaration of voting intention, basing my argument on Spanish pre-electoral polls carried out by the CIS between 1982 and 2004. Having established the patterns of responses to the survey question about direct voting intentions, I will present the statistical model and data used here to analyze the “No answer” problem. Finally, I will test the main hypothesis put forward to account for the non-declared vote through reference to the main Spanish political parties in the 1980s.