Our exclusive data analysis was published during Canada’s last federal election campaign and was the first news story to put concrete numbers on inequalities between political candidates based on their race and gender. We found that, while all parties recruited more diverse candidates than ever, white men who ran for office still received more money from parties and ran in ridings that were easier to win than racialized and Indigenous candidates. Beyond challenges related to funding and ridings, former and current candidates also shared experiences of racist attacks and discrimination on the campaign trail that discouraged them from running.
Our story was addressed by Canada’s Prime Minister Justin Trudeau while on the campaign trail the day it aired. At the time, Trudeau acknowledged that his party could do more to promote and protect candidates who weren’t white. The diversity of the House of Commons in Canada is an important topic that comes up at each election cycle, especially the aspect of race. Before our data analysis and stories were published, that topic was mostly discussed through profiles of candidates or calls to action from various citizen groups, but it wasn’t backed up with hard data. There was also a gap in academic literature and many were looking for a way to maintain accountability on the topic in a credible way every election. Our coverage became a point of reference for groups promoting diversity in all levels of politics. Our analysis and methodology was also noted by many Canadian researchers as an example of what credible data analysis could tell us on diversity in our country. We were even contacted by political science researchers who had read our stories and now had new ideas on how to push this analysis further, including by adding new variables like diving into the professions of political candidates before they ran for office. Lastly, our coverage also shined a light on our own possible shortcomings as media members: many federal candidates gave us concrete examples of times where they felt typecast by news coverage because of their race or were asked biased questions a white candidate would never have to answer by reporters. We shared this feedback with newsroom leaders, which gave us a chance to openly discuss ways to fight our own potential biases and offer better coverage for the 2021 elections.
For our data analysis, we used the computer language R. For the digital story, we coded an interactive project, with the main data visualization made in 3D with Three.js and the others charts with d3.js. To analyze the gender and race of candidates, we used the dataset on the demographics of Canadian Federal Election Candidates (2008-2019) compiled by four Canadian academics with funding from the Social Sciences and Humanities Research Council of Canada, the first one in the country to match candidates with their race and gender. To identify party strongholds, we used election results from 2004 to 2019 published by the Library of Parliament. We consulted with several political scientists before deciding that, for the purposes of our analysis, a stronghold would be a district won two times in a row by a party during federal elections with at least a 10 percent margin. To calculate how much each candidate had received from political parties, we scraped the reviewed financial filings submitted by candidates made available by Elections Canada and matched each candidate with their electoral results, electoral funding, gender and race. To compare the weight of political candidates with their representation in Canada’s population, we used the breakdown of individuals who self-identify as white, racialized or Indigenous in the Canadian population from Statistics Canada’s 2016 Census, the most recent census available at the time of publication.
What was the hardest part of this project?
The biggest technical hurdle for our team was to merge data for more than 4,000 candidates running federally. The information was scattered in various datasets from different sources. By combining them together, we were able to produce this exhaustive and exclusive analysis, which includes race, gender, electoral results, political strongholds, funding and the likelihood of being elected for the first time in Canadian politics. After the analysis, we decided on a step by step data visualization that would follow the audience as they scrolled down the piece the way electors follow their candidates along the campaign trail all the way to Parliament. On the reporting front, finding former or current political candidates who were willing to share sensitive stories of racist attacks in the middle of a political campaign, especially to two white reporters, was no small feat. By sharing their stories publicly, they could alienate political parties and ruin their chances of ever running again and expose themselves to becoming targets once more. It was key for our team to navigate those concerns with our journalistic responsibility for balance, while working as hard as we could to give each of Canada’s main five political parties a space in our story. Our team did research on each one of the 299 diverse candidates who ran in our last federal election in 2019 and reached out to about 50 while staying mindful of balancing male and female candidates with equal numbers for each political party. We ended up interviewing a half-dozen individuals who trusted us enough to relive recent and sometimes painful memories of microaggressions and racist attacks on the campaign trail. Those characters carried the story and helped us reveal yet another challenge for diverse candidates, which we couldn’t see in the data alone.
What can others learn from this project?
We think this project demonstrates how valuable merging data from various sources can be to tell new and important stories. It’s tedious but rewarding work and having the ability to spot how datasets can talk to each other will often lead to exclusive findings. We hope other reporters will develop that skill, especially if they are covering topics like social justice and racial or gender inequities. On the reporting side, our advice would be to step away from the often dazzling data and code and to really take the time to connect with potential interviewees on a human level, without referring to the data at first. Our coverage would never have had the impact or the depth it did without the honesty and emotions shared in our interviews and that did not come from numbers. Another valuable lesson for us was that you can’t be married to your data or your findings if another key element of the story is revealing itself. What interviewees were sharing often didn’t have anything to do with data or funding or their likelihood of winning a riding. Many times there were harrowing anecdotes about racist attacks they’d suffered while on the campaign trail and sometimes the lack of support they received from their own parties. They also often felt that media coverage on their campaign focused solely on their race and that reporters rarely asked them about their views on policies the way they did white candidates. That never appeared in our data and it became an essential element of our coverage with good reason. A last point to keep in mind: you can make beautiful and engaging broadcast coverage (TV/Radio) with data analysis too! We added a link to our TV coverage so you can see the difference.