Friday, April 3, 2020

A curious graph and the future of New York



That's from National Review and the credit tells us that the author of the piece it is attached to, Daniel Tenreiro, is responsible for it. Before discussing the problem I have with it, let's first read the paragraph that introduces it:
Yesterday, more than 1,000 Americans died of coronavirus, the highest daily death toll yet recorded. The number of confirmed cases is above 215,000 in the U.S., with serious outbreaks across a number of states. While New York and New Jersey remain the domestic epicenters of the outbreak, Michigan, Louisiana, and Massachusetts are all seeing their per capita case numbers skyrocket. Florida governor Ron Desantis issued a statewide stay-at-home order yesterday.
Now go back to the graph and have a look at the markings on the vertical axis. Weird! I don't know what Tenreiro intended but if you set out to design a graph to misrepresent the data to make it look like all these states are on the same path, that's what you'd do.

In fact, the numbers for New York State are far worse than the others and the difference is almost entirely because of New York City. An honest would show huge differences.

As has been discussed here in the past, cities like New York, London, Paris and Berlin no longer make sense. This disease is highlighting one kind of problem but there is far worse. There is no longer any sane economic or cultural basis for metropolitan areas.

There is, however, a massive political base in these cities and it has a huge influence on elections.

3 comments:

  1. I'm going to offer a comment on this post... please take it in the spirit of a kind corrective. This may be hard to hear, and I am sorry for any pain that arises from this, but I feel it must be said. I hope to get you to understand your error here, in the hope that you can learn from this, and do better.

    This is not intended to attack you in any way... it's to point out a too-common human frailty: we tend to see malice behind things we do not understand, which are confusing and lead us to suspect that others have bad intent, when in fact they are doing good instead. This is a species of "fundamental attribution error", and one we should all try to guard against in ourselves. I believe that when we indulge this erroneous impulse, we make the world a worse place.

    I get the distinct impression that you're not familiar with semi-log plots, why they are used, and why they are MUCH more valuable to showing real relationships of interest in situations involving exponential growth. Please Google “Semi-log plot”.

    There is nothing remotely "weird" about what Daniel Tenreiro did, here. He actually chose an IDEAL form of plot, to show how the GROWTH of cases has happened over time, in different locations.

    Most people can't think properly about how exponential growth works, what it means, and how best to represent it visually. That's not a slight on them... human brains don't naturally deal well with exponential growth, despite it being a commonplace feature of the natural world.

    It takes experience and training to gain the skill to have it become intuitive. The lack of that training quite often leads people to not understanding what's really important in something like a pandemic: finding out what's happening to the RATE OF GROWTH OVER TIME.

    The semi-log plot is ideal for representing and comparing that information. A consistent rate of growth ("doubling every X days") will show as a straight line on a semi-log plot like this, with the slope of the line corresponding to the rate of growth. That's what makes the semi-log plot so useful. Knowing those things lets you cleanly and easily see those complex relationships across multiple locations. That's why it's so-often used.

    It is the "normal plots" that, if anything, are "designed to mislead", because of that. You'll often see them used by people who ARE trying to create a panic. They are LESS informative, which is why they are common propaganda tools.

    I think that unfamiliarity has led you to needlessly slander Daniel Tenreiro with your implications that he was somehow TRYING to mislead people by using this bog-standard form of representing epidemiological data.

    This is exactly how it is SUPPOSED to be done, and how it's been done in this field for many decades. You've taken something that is completely good and proper, done by someone who is "doing it right"... and you've allowed your own ignorance about what's being done to trick you into thinking they're trying to do something malicious instead.

    You owe Daniel and all of your readers the deepest and most sincere of apologies for what you've done here, even though you likely did it out of ignorance rather than malice yourself. This is the opposite of what we should be doing. This behavior is spreading falsehood, and causing you to bear false witness against your neighbor.

    Please... think about this and repent over what you have done. This feels like egregious sin to me. I'm not your confessor, but I can imagine what they might say if they understood what you have done, here:

    Correct your error, publicly step forward and own the wrongdoing you have committed. Ask for forgiveness, perform the necessary acts of good contrition to set this right, and then go forth and sin no more. Please do what is RIGHT here, if only for yourself. I offer this in loving kindness, please accept it as such. Thank you.

    ReplyDelete
    Replies
    1. I don't disagree with you that this is an appropriate way to present data if, and this is a big if, you are presenting to a specialist audience. If you are writing for a larger audience, you need to explain why the data is not presented in an intuitive manner, particularly if the sentence leading into the graph is as follows, "While New York and New Jersey remain the domestic epicenters of the outbreak, Michigan, Louisiana, and Massachusetts are all seeing their per capita case numbers skyrocket." That suggest that these states are catching up to New York, a false message was reinforced by the graph. A note pointing out the calibration of the vertical axis should have been included.

      Delete
  2. It may come off a little indignant to say (I don't take it that seriously), but, as regards your condemnation of the metropolis in relation to COVID...this did not age well. It is interesting to ponder, however. Intuitively, one would believe massive populations in a small area means more bad things...but I think the curious thing about the 21st century and the advancement of technology is that we are finding out the exact opposite is true, particularly when it comes to wealthy nations.

    ReplyDelete