Pens and Polls in Nicaragua: An Analysis of the 1990 Preelection Surveys

Abstract

In order to understand the problems of interpreting political surveys in highly polarized societies, we analyze the results of 17 preelection polls carried out during the three months preceding the 1990 Nicaraguan election. We also draw on findings from a survey-based experiment on response bias administered in the last days of the election campaign. Both sets of data are used to explore five hypotheses that might account for the large inconsistencies between many of the preelection surveys and the actual election results: the volatility of the Nicaraguan electorate; serious problems in sampling a developing country in turmoil; effects of the administration or context of questions; intentional partisan bias on the part of polling organizations; and response effects due to the partisanship of the polls as perceived by the Nicaraguan population. Our analysis indicates that the main explanation for the failure of some polls and the success of others lies in an unusual interaction between respondents' vote intentions and the perceived partisanship of a poll. Paradoxically, bias on the part of a survey organization was necessary to reduce bias on the part of respondents.