Stats Canada and Their Foray Into Twitter Polls
Cam Loeschmann // Contributor
On Jan.24 2020, Statistics Canada released a Twitter poll that seemed innocuous at first glance. “How old were you when you bought your first house?”
The options began at “under 24” and ranged up to “over 46,” with the majority of those polled answering “25-34” (52.1 per cent at the time of writing). If one follows the fiction of the “American Dream” then this may seem reasonable, but personally, I only know one person who owns a home—my mother—and it’s a condo, not a house. She also bought it when she was in her fifties.
This poll brings up some interesting questions. Who is answering this poll? Who made this tweet? What conclusion are they reaching based on this data? Most importantly, how would the data differ if there was an option for “I do not own a house?”
The replies to a poll ran the predicted gamut of the 21st-century Canadian internet. Millennials overwhelmingly volunteered their renters’ status or commented “ok boomer,” while those able to participate in the poll talked about participation trophies and how much their homes from 1981 are now worth. One image posted by Twitter user Tristan A.F Long (@thelonglab) illustrated this issue poignantly, if you understood the reference. This image was an outline of a World War II military plane marked with red dots, representing gunfire damage, concentrated on the aircraft’s cockpit, tail, and the tips of the wings. Statistician Abraham Wauld famously rethought WWII bomber designs with his thinking on survivorship bias. Previously it was taught to reinforce those areas until Wauld considered the damage on the planes that could not even limp home. With that taken into consideration, those red points became a different set of information: how much damage an aircraft could take without falling to enemy fire.
With no option for Canadians who are not homeowners, and knowing that in this given market they might never reach this milestone, there is a huge sector of people who cannot place a vote. Survivorship bias, or in this case homeowner’s bias, has skewed the numbers in a way that cannot lead to accurate, thoughtful data.
How much do we trust our national statistics and data service? Since 1918, this agency has existed under the federal government to provide social and economic statistical information. It has done this via surveys, census and, apparently, Twitter polls. This brings us to an interesting question: how is one of these unlike the other?
Let us say, for example, that I needed answers about a particular aspect of Canadian life. For the sake of argument, I want to know how much Canadians spent in the last week, month and year on groceries. Let us also say that I, for whatever reason, do not trust any official channels to provide me with this information but instead prefer to ask Canadians directly. Now, approximately 37.8 million people live in this country. What would be the best, most accurate way to get the data I need?
The internet is hypothetically a great place to start as 91 per cent of people across ten provinces were reported to be active online last year. It is, however, a very large place and there is no one avenue or website that would be simple. Facebook is something that many Canadians have access to, as is Twitter and Instagram. For those too serious for everyday social media, I might do something on LinkedIn. What about people who are on several or all of these sites? Let us consider people who might be comfortable giving me a sampling of their financial data but would only be comfortable with this if it is anonymous. Already, my survey methods are in the mud. Ideally, I would find every single internet user in this country, exactly once, and collect this information without violating privacy or letting anyone answer multiple times.
Already we have another problem. If only 91 per cent of people across the ten provinces are active online this brings us to an approximate Canadian internet user count of 34.3 million. Very different from our 37.8 million total population count. To be sure, this discrepancy is not huge in the grand scheme of things. Let us not forget that the internet user data that Stats Canada provided me with did not include the territories. If the objective of my hypothetical grocery cost study was to look at the disparities in grocery spending between the provinces and territories, any internet-based study would be incredibly out of step as I would have no idea how many people would even see my study.
A foray into census by Twitter poll is Stats Canada’s way of trying to be more visible and involved in Canadians’ everyday lives. The lack of forethought about who would see this tweet and who would be able to respond comes across as insulting and abrasive. This is an organization funded with our tax money. In order to stay up-to-date in the internet age, they must remain reliable and nonpartisan—not succumb to the ease of polling Twitter users from their desks.