Substate variations in political values in Canada

ABSTRACT

The article aims to make three methodological and substantive contributions to the literature on substate cleavages in political values. Considerable controversy characterizes this literature. The paper argues that this controversy is due to how indicators representing political values are chosen and constructed. The paper proposes to use factor analysis to select and construct indicators of political values. The analysis identifies five dimensions, which collectively account for 57% of Canadians’ political values. They include support for moral traditionalism, egalitarianism, pluralism, openness to immigration and personal responsibility. Second, the paper shows that there is only limited variation across provinces in political values. Third, the paper shows that this result holds when considering regional variations rather than provincial variations.

KEY WORDS:

• Regional differences
• federalism
• political values
• Canadian politics
• methodology
• political culture

Acknowledgments

I would like to thank professors André Blais, Christopher Cochrane and especially Peter Loewen for their advice, suggestions and support during the redaction of this article. I also acknowledge the helpful feedback of discussants and reviewers who read previous versions of the manuscript at conferences and during the submission process. Finally, I would like to acknowledge funding from the Social Sciences and Humanities Research Council, the Fonds de recherche du Québec - Société et culture and the Ontario Ministry of Research and Innovation’s Early Research Award program.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. Data from 2006 are not included because the mailback survey was missing from the 2006 election.

2. As a robustness check, a second factor analysis was performed only on variables that pertain to political values. The results are extremely similar and are available from the author upon request.

3. Age has been rescaled by dividing this variable by 10. For this reason, estimates for age should be interpreted as the effect of being 10 years older.

4. Detailed results are available from the author upon request.