A key challenge in studying democracy is understanding who runs for office and what factors influence their success, particularly when it comes to descriptive representation – the extent to which a legislature reflects the demographic makeup of the population. Until recently, however, researchers had limited data on Canadian political candidates. Professor Semra Sevi of the University of Toronto has created the largest dataset of its kind, documenting every candidate who ran in Canadian federal and Ontario provincial elections since 1867. This comprehensive dataset has already enabled several groundbreaking studies on Canadian democracy. Read More
The dataset is remarkable for its size, covering more than 46,500 federal candidates and over 15,500 provincial candidates. It includes key details about each candidate – such as their name, date of birth, occupation, indigenous origins, identification with the 2SLGBTQIA+ community, gender, party affiliation, election date, electoral district, incumbency status, and vote share. Crucially, each candidate is assigned a unique identifier, allowing for career tracking across multiple elections – whether they won, lost, switched parties, or ran again.
Professor Sevi has used this dataset to publish numerous studies. One of her early analyses, with colleagues at the University of Montreal, addressed whether women get fewer votes. By examining data on over 21,000 candidates since 1921 (when women first ran for Parliament), they found that women candidates receive virtually the same vote share as men. The difference is only about half a percentage point. While a larger gap of about 2.5 percentage points existed in the 1920s, the gap has since become statistically insignificant in recent elections. This suggests that political parties should recruit and promote more women candidates without fear of an electoral penalty.
Building on this, Professor Sevi used the dataset to explore the incumbency advantage (those already holding office) in Canadian politics and whether this advantage differs between men and women. Her analysis of competitive districts in federal elections from 1990 to 2021 shows that winning candidates are more likely to run again, with men and women incumbents showing similar patterns. However, women who lose an election seem more likely to quit politics than their male counterparts.
Professor Sevi’s research has also uncovered interesting trends in Canadian politics over time. For example, while the proportion of women candidates has surged – from less than 1% in 1921 to 38% in 2021 – they still make up only 30% of elected representatives. Additionally, the data shows that electoral turnover in Canada is higher than in many other countries, which could create more opportunities for new candidates to enter politics. These findings highlight both the progress and the ongoing challenges in achieving a truly representative democracy.
Professor Sevi’s work demonstrates the usefulness of her dataset and its incredible potential for future research. Some of her other research using these data include examining: legislative log-rolling: using a proposal lottery to identify causal effects, the incumbency advantage in Canadian elections, working-class descriptive representation in Canada’s federal parties, the effects of proposal power on incumbents’ vote share (with updated results from a naturally occurring experiment), gender and political campaign contributions in Canada, whether women get fewer votes in Ontario provincial elections, whether lawyers receive more votes, and legislative party switching and the changing nature of the Canadian party system from 1867 to 2015.
Each of these studies sheds light on different aspects of Canadian politics, from gender and party representation to electoral advantages and voting behavior. This pioneering dataset is advancing our understanding of the complex forces that shape who represents Canadians in Parliament – and what this means for the future of Canadian democracy.
