Militant Nationalist Electoral Support: A Measurement Dilemma

Abstract

It could by no means be argued that efforts to predict election results from public opinion data have an unblemished track record. From the disastrous misprediction of the U.S. presidential election in 1948 to the unheralded victory of Hamas in the 2006 Palestinian election, pollsters have been criticized for ‘getting it wrong’, with shortcomings in the United Kingdom in 1992 and in France in 2002 attracting particular academic attention (Curtice, 1997, p. 317; Miguet, 2002, p. 207). In principle, serious failure in election forecasting should be a source of greater worry to newspaper editors than to academic political analysts, who are more concerned with understanding long-term trends than with offering ephemeral insights into the future. There are, however, circumstances where there is a serious problem of systematic bias in opinion data, an issue taken up further here. Northern Ireland is such an example: politically militant views have tended to be consistently understated (Whyte, 1990, p. 4; Evans & Duffy, 1997; Hayes & McAllister, 2001, p. 912). So much is this the case that it has been argued that ‘if the public had been anywhere near as moderate as they have generally represented themselves to pollsters during the past three decades there would not have been a Northern Ireland problem’ (Mitchell, O’Leary, & Evans, 2001, p. 732). The three following sections tackle this question by discussing in turn the nature of the problem, the challenge of explaining it, and its implications for the analysis of militant ethnonationalism.