¹ Niemi, Richard G., Whitten, Guy and Franklin, Mark, ‘Constituency Characteristics, Individual Characteristics and Tactical Voting in the 1987 British General Election,’ British Journal of Political Science, 22 (1992), 229–54.CrossRefGoogle Scholar

² A feature of the questionnaire that generated our data was that it did not ask all of those whom we identified as tactical voters which party they ‘really preferred’. However, for some of those who were not directly asked, we were able to impute such a preference from the same answers as enabled us to identify them as tactical. See Niemi, Richard G., Whitten, Guy and Franklin, Mark, ‘People Who Build Glass Houses … A Response to Evans and Heath's Critique of our Note on Tactical Voting’, British Journal of Political Science, 23 (1993), 549–53.CrossRefGoogle Scholar

³ Evans, Geoffrey and Heath, Anthony, ‘A Tactical Error in the Analysis of Tactical Voting: A Response to Niemi, Whitten and Franklin’, British Journal of Political Science, 23 (1993), 131–7.CrossRefGoogle Scholar

⁴ Niemi, et al. , ‘Constituency Characteristics, Individual Characteristics and Tactical Voting’.Google Scholar

⁵ Voting for the likely winner in order to give that candidate a larger majority might be regarded as a variant of normal tactical voting in that it would be undertaken in order to achieve a different outcome (in some sense) than might otherwise have occurred. However, almost all (85 per cent) of the tactical votes for the largest party went to parties that won an overwhelming proportion of the vote in the constituency (the first two parties were more than 25 per cent apart).

⁶ For example, a moderate Conservative party supporter, living in a safe Conservative constituency, might vote for the Alliance candidate in the hope that a diminished Conservative margin of victory would help pull the party back to the centre. Or, in the 1992 American presidential election, one can imagine some individuals who preferred the Democratic or Republican candidate, expected one of them to win handily in their state, and who voted for Ross Perot in order to express indignation of the sort that Perot articulated. One can well imagine similar motives among many voters for Green parties in Europe.

⁷ In a recent article, Rose and McAllister also employ the terms instrumental and expressive to point out that ‘every vote has a double significance; it expresses the state of mind of an individual elector and it is an act in an instrumental process for determining control of government’ (p. 114). Our perspectives differ, however, in that we suggest the existence of expressive tactical voting, while Rose and McAllister see expressive voting as explicitly nontactical. See Rose, Richard and McAllister, Ian, ‘Expressive versus Instrumental Voting’, in Kavanagh, Dennis, ed., Electoral Politics (Oxford: Clarendon Press, 1992), pp. 114–40.Google Scholar

⁸ By construction, distance from contention would have a negative coefficient for expressive voters. (When included, it has this expected result along with a strong positive coefficient for instrumental voters.) The effects of excluding it are less obvious for other variables, so we tested an alternative specification with distance included; differences between that equation and the one shown in Table 4 are noted below.

⁹ For brief descriptions, see Demaris, Alfred, Logit Modeling: Practical Applications (Newbury Park, Calif.: Sage, 1992), pp. 61–71CrossRefGoogle Scholar, and Greene, William H., Econometric Analysis, 2nd edn (New York: Macmillan, 1993), pp. 666–72.Google Scholar

¹⁰ If distance from contention is included in the model despite its built-in relationship to expressive voting, competitiveness takes on a positive coefficient for both types, suggesting that competitiveness is still something of a puzzle. Note that the competitiveness measure was based on actual election results. As we pointed out in our original piece, one would ideally like to have individual pre-election expectations. See Niemi, et al. , ‘Constituency Characteristics, Individual Characteristics, and Tactical Voting’, fn. 15 for further discussion.Google Scholar

¹¹ This relationship weakens considerably if distance from contention is included in the model, suggesting that instrumental tactical voting may stem from both opportunity to affect the outcome and a motive for doing so.

¹² When distance from contention is included in the equation, the relationship of ‘caring’ with instrumental tactical voting is positive and that with expressive tactical voting is negative, both as expected. However, in none of the instances – with or without the distance variable – is the ‘caring’ coefficient statistically significant.

¹³ In a further effort to differentiate between instrumental and expressive tactical voters, we include a variable inquiring how the respondent felt about one-party versus coalition government. Those who preferred a one-party government less often voted tactically, with both coefficients clearly having statistical significance. Possibly this result was influenced by the fact that Alliance supporters had more reason to vote tactically and saw that the only way they would be involved in the government was as a coalition partner.

¹⁴ See, for example, Fiorina, Morris P. and Shepsle, Kenneth A., ‘Is Negative Voting an Artifact?’ American Journal of Political Science, 33 (1989), 423–39.CrossRefGoogle Scholar