I intensely dislike dishonest presentations of data


One could make the point that there’s an interesting and sort of natural experiment going on here: two adjacent countries with strong cultural similarities but different approaches to health care are addressing the pandemic, and spontaneously generating lots of data we could use to evaluate their methodologies. But this illustration isn’t it:

Oh, look. The US is a smear of red. Canada just has a few dots…but wait. Those dots look like they correspond to the approximate centers of Canadian provinces. Are we seriously supposed to compare a coarse province-level aggregation of data to what looks like a finer-grained county level visualization of US cases? That’s highly misleading. It may be that Canada has managed the pandemic better than the US, but you can’t assess that from such a bad map. For shame, whoever made that.

Comments

  1. Pierre Le Fou says

    Yeah, exactly. A few months ago on a Slack channel we use at work, I posted a screenshot of the Canadian province of Quebec from that very same site. And my caption was: “hey look guys, all the cases we have in the province are over there way in the northern regions! We’re safe here in Montreal!”

  2. says

    And there’s no need to be so misleading, either. The USA has approximately 9 times the population of Canada, so we’d expect, perhaps, to see nine times the number of cases and deaths.

    Canada had its first death on March 9th, and its 1,000th death on April 15th, and now has had nearly 9,000 deaths attributed to Covid-19, with new deaths a tiny fraction of what they were at the first peak.

    USA had its first death on February 29th, nine days earlier than Canada, and its 9,000th death (i.e. same per population) on April 4th, 11 days earlier, and now has 167,000 deaths attributed to Covid-19, twice the Canada rate, with new deaths at about half what they were at the first peak, but possibly still rising.

    In short, the USA has failed to contain the virus, probably due to inadequate political leadership.

  3. Howard Brazee says

    I am much more upset when people on my side have bad arguments than people on the other side do. My side should be better.

  4. HappyHead says

    My first thought on seeing that image, without reading any text, was:

    Oh, it’s a comparison of population by province/state/whatever. Huh, looks like they did provinces in Canada versus individual counties inside of states in the USA… I suppose that’s fair in terms of head-count. Lots of US counties have more people than most Canadian provinces. (More US counties than Canada has provinces actually… our whole country has less people than California.)

    If that’s what they were doing, I’d have called it a good chart, but no, it’s a bad effort at labelling corona virus spread on a chart that makes it impossible to figure out the actual relative amounts. I think the disappointment is worse because at first I thought it was something semi-useful.

  5. raven says

    …but different approaches to health care are addressing the pandemic,…

    News to me.
    AFAACT, the USA never has addressed the pandemic.

    The Federal government ignored it for two months, then tried to get together a response that went nowhere.
    Then around June, they decided the pandemic was winding down, it was boring, and they just lost interest.
    Shortly afterwards it flamed up again.

    Here it is August and the Federal response is still nonexistent.
    It consists of ordering school children to go back to school, get sick, spread the Covid-19 virus further, and have more people dead and permanently disabled.

    What the Feds did do, was push the Covid-19 response down to the states to deal with.
    The states have both fewer financial resources and less science-medical expertise but they did what they could.
    In some cases, the states also gave up and pushed the pandemic response down to the counties and cities.
    Which of course, have even fewer resources than the states.

  6. rojmiller says

    You don’t need any fancy illustrations (or not-so-fancy ones) to compare Canada and the US (from worldometers.info):
    Cases per 1M population:
    US 16,019 Canada 3,187
    Therefore US infection rate is 5X Canada
    Deaths per 1M population:
    US 506 Canada 238
    Therefore US death rate is 2.1X Canada
    Canada’s death rate is relatively high because in the early stages COVID-19 got into a lot of long term care homes in Ontario and Quebec, and killed a lot of elderly people.

  7. timmyson says

    Yeah, I complained about this when it came up on Twitter. And I think the county-equivalent data exists for Canada: We know that Windsor/Essex (right across the border from Detroit) is hit pretty hard, also the greater Toronto area. Maybe it’s not being published as widely or accessibly?

    @pierre #5 I would love to know your data sources.

  8. wzrd1 says

    The graphic is a low resolution graphics from the Johns Hopkins pandemic tracker, which was snapshotted at low resolution (one can zoom and gain more granular results) and not even given attribution or context beyond what the abuse of that map provides.
    https://coronavirus.jhu.edu/map.html
    Technically, it’s a snapshot, likely excised by cliptool to simply cut the continental zoomed out image, as the page starts out with a global map and one zooms to areas of interest, which does go down to town/county level.
    It’s an abuse equal to taking a snapshot of one of PZ’s spider habitats, blowing up its thumbnail until details cannot be made clear and captioning “PZ Myers idea on a proper human household looks like”. And both would be not only not right, but not even wrong.

  9. says

    you can basically show correlation between any 2 things in US showing number of cases on US map, because all of them will correlate to population distribution

  10. says

    I’m a Canadian, and I think it’s a very misleading representation, but also, it’s not. How Canadian!
    I don’t understand the data that drives this picture. There are 9 dots in total (at least that I can make out) that appear to be in Canada, and I have to assume that this is because they’re going with one per province (and excluding the territories). But if that’s the case, then the US side should have 50 dots of varying sizes. Hell, just looking at the map, it would appear that Calgary has no cases, but there’s cases in Edmonton (Calgary has WAY more cases than Edmonton) If it’s not the case, and this refers to total cases or successes or deaths or whatever, then there should be many more dots. More, as you point out, most of the dots are kinda central, and most of the people live in the south (see https://qph.fs.quoracdn.net/main-qimg-073611f762cf3ef329ee65a5d65baa58 for an example). So it seems to me that this picture is total garbage.

  11. jrkrideau says

    @ 8 timmyson
    I think you want the stats from the various Public Health Units. The units are are not necessarily I unit / county but the stats seem analogous in Ontario to US county stats.

    Countries do not really mean much in Ontario compared to US counties. For that matter much of the province has districts not countries and our counties have townships which are much more important to the rural population than a county.

    Presumably someone in the province understands how it all works.

    Here is the link to the Windsor-Essex County Health Unit.
    https://www.wechu.org/

    This seems to give a list of all the units in Ontario.
    http://www.health.gov.on.ca/en/common/system/services/phu/locations.aspx

  12. wzrd1 says

    Personally, I’ll stick with Johns Hopkins tracker, which track all health department sources, local, county/parish/district/commune/etc, as well as state/province/department/etc, as well as nationally for 188 nations and has a team dedicated exclusively towards that task. That and their experience, beginning with the 1918 influenza pandemic and have shown a highly effective and accuracy that’s exceptional, just makes life a bit easier, with only a few other trackers that I’d need to follow to get an overall picture of what is going on.
    If I need more, I have a friend who is an epidemiologist to help fill in any gaps, of which there are many. For, epidemiologists are master of the darkest of the darkest arts of statistics. ;)
    And someday, when this is all past us, I do owe him a few beers and a fine dinner wine for him and his wife.

Leave a Reply