At the time this article was written Bill Cross taught in the Political Science Department at Mount Allison University in Sackville, New Brunswick.

Most election study in Canada is focused on the leaders’ tours and debates, and the perceptions of the campaign as relayed by the national media. This view of elections, prevalent in many parliamentary democracies, leaves little room for significant involvement by local party activists. Recent case studies of local election campaigns during the 1988 federal election conducted for the Royal Commission on Electoral Reform and Party Financing suggest that this view may be incomplete. Several authors, in a collection of ten case studies, conclude that local campaign efforts had a significant impact on riding election results.¹ These findings, while somewhat impressionistic, support the hypothesis that local association “vitality” can make the difference between victory and defeat at the riding level. This study builds on these case studies by using survey data collected from local parties after the 1993 election to examine the relationship between local party “vitality” and electoral success.

This paper is based on results from a national mail survey of all constituency association presidents in the three parties receiving the most votes in the 1993 federal election – the Liberal, Reform and Progressive Conservative parties. Four hundred completed surveys were returned for an overall response rate of 52 percent. The response rate varied among parties with 62 percent of Reform presidents responding compared to 54 percent of Liberals and 41 percent of PCs. Despite the mailing of a bilingual survey and cover letter to Quebec associations, the response rate from that province was considerably lower than that from English Canada. The respondent associations are representative in regards to their performance in the 1993 election.

Respondents were asked a series of questions relating to activity of their local party association prior to the 1993 vote. An independent variable was created using an index comprised of five indicators of association “vitality”.

• 1992 party membership level,
• 1993 membership level,
• the number of 1993 campaign volunteers,
• the number of needed minus actual 1993 campaign volunteers,
• the number attending the candidate nomination meeting.

Statistics for each of these variables was collected but as many associations responded with a range or an estimate, the data was transformed for purposes of the index. Thus, the range of scores for each variable is one through seven, with seven representing the highest level of participation and one the lowest. Scores on the index range from 5 through 35. Associations with index scores between 23 and 35 are categorized as high “vitality” associations, those scoring between 17 and 22 are medium “vitality” associations, and those scoring between 5 and 16 low “vitality” associations. These classifications are designed so that one-third of associations fall into each category. The simplest method of analysis is to examine scores on the “vitality” index for each category of electoral result in the 1993 election. As illustrated by Table 1.

Associations whose candidate won the 1993 election had both the highest mean and median scores followed by associations whose candidate finished second, followed by third place finishers and finally those whose nominee finished fourth or lower.

Table 1: Scores on “Vitality” Index
by 1993 Candidate Finish

This pattern holds within each party. Winning Liberal associations received a mean score of 23.3 and a median of 23.0, while losing Liberal associations had a mean of 17.5 and a median of 17.0. Winning Reform associations scored a mean of 26.5 and a median of 27.0, while losing Reform associations had a mean of 17.0 and a median score of 16.5. PC associations finishing first or second had a mean score of 21.4 and a median of 21.5; PC associations finishing lower than second had a mean score of 16.1 and a median score of 15.0.

The findings in Table 1 suggest that there may be a relationship between candidate finish and association “vitality”. The existence and strength of this relationship can be tested more rigorously through a cross-tabulation analysis. This procedure reveals that candidate finish has a significant and strong positive relationship to association “vitality”.

As illustrated in Table 2, the higher an association ranks on the “vitality” scale the more likely it was to win the 1993 election. For example, 67 per cent of associations that receive high “vitality” scores won the election compared with 25 per cent of associations with medium “vitality” and six per cent of associations with low “vitality”.

A strong relationship between these two variables also exists within each party. As illustrated by Table 3, in all three parties, associations with high “vitality” were most likely to be electorally successful followed by those with medium “vitality” and finally those with low “vitality”.

Table 2: Constituency “vitality” index by 1993 candidate finish (actual cases)

Table 3. Association “vitality” by 1993 candidate finish for Liberal, Reform, and PC  associations (actual cases and column percentages)
(Won for PCs = 1st or 2nd place finish)

While the above table suggests a strong positive relationship between association “vitality” and candidate finish it is possible that these findings are strongly affected by regional variations. Given the regional disparity in candidate finish within parties in Canadian elections, it is possible that what is being reflected is association “vitality” within regions of electoral strength for each party and not any differences among constituency associations within a region. To test this, Table 4 examines the relationship between association “vitality” and candidate finish for constituency associations of the Liberal and Reform parties within the four western provinces. Both the Liberals and Reform won and lost a number of ridings within this region and Table 4 shows the existence of a strong relationship between association “vitality” and candidate finish within the West. These findings are meaningful as it is unlikely that a riding association’s ability to attract participation within the region varied significantly because of different degrees of popular support or constituent’s differing views of the parties’ chances for electoral success at the outset of the campaign.

Table 4. Association “vitality” by 1993 candidate finish for Liberal and Reform associations in Western Canada (actual cases and column percentages)

It is possible that there is no causative effect between association vitality and electoral success. Some may argue that both are caused by existing popular support for a party within each riding. (This is virtually impossible to test empirically as it requires data showing popular support for parties within each riding prior to the election). This explanation, however, would not likely account for the significant relationship found between association “vitality” and electoral success within the Western provinces. For example, there is no evidence that support for the Reform Party was substantially higher prior to the calling of the 1993 campaign in all of the Alberta constituencies it won than in those it lost.

A second factor minimizing the likelihood of the relationship being spurious is that party popularity was very volatile before and during the 1993 campaign. The use of variables representing participation in local parties over a two-year period should minimize this effect. The share of the popular support for all three parties shifted dramatically in the period leading up to, and during the course of, the campaign. The Reform Party began the campaign with support in the single digits while the PCs enjoyed a slight lead over the Liberals. If existing popular support causes “vitality” in constituency associations then its effect should be minimized by using variables that cover a long enough time span to minimize the effect of late surges or falls in popularity. For example, if current popularity levels determine association vitality, turnout at contested PC candidate nominations should be quite high, as the party was leading in the polls for much of the nomination period. This, however, is not the case. While it is likely that existing popular support has some impact on association “vitality”, it does not seem to be fully explanatory.

The suggestion of a relationship between local association “vitality” and electoral success should be of interest to political parties. While the analysis presented does not establish a causal link, it does suggest the existence of a strong relationship. A local party association cannot influence the national campaign and the leaders’ tours and debates, but it can have an influence over the size of its membership and how engaged it keeps its members both before and during election campaigns.

More research and analysis needs to be conducted in this area. Multivariate analysis can help determine more precisely the nature of the relationships being studied. As well, collection of similar data from subsequent elections will help to ensure that the relationship found is not limited to the 1993 experience.

Notes

1. Bell, David and Frederick Fletcher, eds., Reaching the Voter: Constituency Campaigning in Canadian Federal Elections, (Toronto: Dundurn Press, 1991).