Hostname: page-component-cd9895bd7-jkksz Total loading time: 0 Render date: 2024-12-28T07:03:57.303Z Has data issue: false hasContentIssue false
  • English
  • Français

Additional Evidence on the Effect of Incumbency in Canadian Elections*

Published online by Cambridge University Press:  10 November 2009

Michael Krashinsky  and
William J. Milne
Michael Krashinsky
Affiliation:
University of Toronto
William J. Milne
Affiliation:
University of Toronto
Get access Rights & Permissions [Opens in a new window]

Abstract

This work extends the authors' earlier work on incumbency by examining, riding by riding, results in 10federal elections from 1926 to 1980. Regression results indicate that incumbency generally had a significant impact, and between 1957 and 1980 incumbency was worth about four points for the Liberals, Conservatives and New Democrats, and significantly more for the Social Credit. The results vary and seem to be smaller than those observed in Ontario provincial elections, but this may be due to the greater instability in federal politics, since general shifts in party allegiance seem to reduce incumbency effects.

Résumé

Cet article élargit la portée de l'etude précédente des auteurs sur l'avantage éléctoral dont bénéficie l'élu, en examinant circonscription par circonscription les résultats de 10 élections fédérales tenues entre 1926 et 1980. L'analyse de régression indique que l'élu jouit d'un avantage considérable qui équivaudrait, dans la période 1957–1980, à quelque quatre points pour les candidats libéraux, conservateurs et nouveaux-démocrates, et à un chiffre encore plus significatif pour les créditistes. Ces résultats varient, et s'avèrent moins prononcés que ceux observés dans les élections provinciates de l'Ontario, fait qui serait attribuable à une plus grande instabilité en politique fédérale, puisque les fortes oscillations dans la loyauté que ressent l'électeur envers un parti fédéral tendent à réduire l'avantage dont bénéficie l'élu.

Type
Note
Information
Canadian Journal of Political Science/Revue canadienne de science politique , Volume 18 , Issue 1 , March 1985 , pp. 155 - 165
Copyright
Copyright © Canadian Political Science Association (l'Association canadienne de science politique) and/et la Société québécoise de science politique 1985

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 See Krashinsky, M. and Milne, W. J. ‘Some Evidence on the Effect of Incumbency in Ontario Provincial Elections,’ this JOURNAL 16 (1983), 489–500.Google Scholar

2 The reader is reminded that the random disturbance term has a complicated variance because the coefficients in equation (2) are the means of the random coefficients in equation (1). This is what requires the use of generalized least squares to derive efficient estimates. For a discussion of the estimation techniques used, see our earlier work Ibid..

3 The remaining elections in the period since 1920 were omitted because redistribution had made it impossible to compute the change in votes for a given riding between the election in question and the previous election (the wartime election and the one following were also dropped because of the distortion of party voting patterns in the wartime election). The first two elections were omitted for the NDP-CCF and the Social Credit because the parties did not exist at that time. The 1962 election was dropped for the Social Credit party because the party had no incumbents to run in that election.

4 Where the party in question had received less than 5 per cent of the vote in the previous election, there is the possibility that equation (2) might predict a change in vote that would result in a negative vote total in the next election. Of course, this problem could be handled by logit techniques, but we chose the easier approach of deleting ridings where this might be a problem. We also believe that ridings where independents receive more votes than the two major parties are ‘special’ and should be deleted (it would be unclear whether or not the independent was siphoning votes from the party being studied).

5 There are 35 equations, each with 7 estimated coefficients. Complete results are available from the authors.

6 In this note, the term statistically significant means that the null hypothesis (generally that the coefficient is zero) is rejected at the .05 level.

7 The coefficient is insignificant at the .05 level but would be significant if a slightly less demanding test is used. The coefficient is significant at the .062 level.

8 See Zellner, A. ‘An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests for Aggregation Bias,’ Journal of the American Statistical Association 57 (1962), 348–68.CrossRefGoogle Scholar

9 Equation (2) was run on the pooled data first constraining the incumbency coefficient to be equal across all elections and then allowing it to vary across elections. A Chow test was performed on the null hypothesis and the calculated F values were 1.295 for the Liberals, 3.851 for the Conservatives and 4.130 for the NDP (7 restrictions, 1943 degrees of freedom) and 9.427 for the Social Credit (6 restrictions, 1699 degrees of freedom). See Maddala, G. S.Econometrics (New York: McGraw-Hill, 1977), 198.Google Scholar

10 For 1972, the F values were 2.079 for the Liberals (4 restrictions, 237 degrees of freedom) and 1.771 for the Conservatives (2 restrictions, 239 degrees of freedom). For 1963, the F values were 0.887 for the Liberals (3 restrictions, 234 degrees of freedom) and 2.477 for the Conservatives (4 restrictions, 233 degrees of freedom). See Maddala, Econometrics. The number of restrictions vary because the parties often did not have incumbents running in a particular region, requiring the suppression of that regional incumbency dummy in the unrestricted regression.

11 The F values were 7.804 for the Liberals, 0.448 for the Conservatives, and 1.768 for the NDP (1 restriction, 193 degrees of freedom). Maddala, Econometrics. The unrestricted coefficients on incumbency were .0375 (federal) and .0971 (provincial) for the Liberals, .0262 (federal) and .0403 (provincial) for the Conservatives, and .0714 (federal) and .0397 (provincial) for the NDP. The federal coefficients differ from those reported in Table 1 because the data are limited to Ontario ridings. The provincial coefficients differ from those reported in our earlier work because the equation used omits any regional variables.