Wednesday, April 11, 2012

Innumeracy

Illiteracy is a pretty minor problem in Canada these days.  It is quite the problem for those who are illiterate of course but there are not very many of those people now.  Innumeracy, on the other hand, is still a huge problem and honestly the majority of the population suffer from it.  Consider the following front page news article in a local newspaper:

Data shows not all local hospitals are stacking up when compared to peers across the country. - 11 of 18 GTA hospitals fall below the national average at preventing inpatients from dying within 30 days of being admitted for a major heart attack.


Taken from the Toronto Metro weekend edition, April 8.

This data is presented as if somehow it is useful or meaningful when it fact it is neither.  Since we expect 9 of 18 hospitals to fall below the average *no matter how good hospitals in general are* the chance of having 11 of 18 below average for any given area is extremely high.  Average is not bad, it is normal and there is no reason to panic if we aren't better than average.  We also have no idea how far below average the GTA hospitals were.  If the 11 'bad' hospitals were running at 99% of average and the 7 'good' hospitals were running at 200% of average we could safely say that the GTA hospitals are amazing... on average.  Since we have no idea how they actually compare to the national average though this statistic is completely worthless.  The article goes on further to talk about Lakeridge, the worst performing hospital, which falls below the national average on 13 of 21 indicators.  If the worst hospital is actually ranking above the national average 8 of 21 times I think the only thing we can conclude is that hospitals are generally of pretty consistent quality when it comes to keeping heart attack patients alive!

What drives me bonkers is that there is some real news in this article that is completely overshadowed by the statistical spewing.  It is good to know that hospitals have developed a series of indicators to determine how well their patients do and they are looking at each other's performances to attempt to improve.  Great!  That sounds like an idea that will help keep people alive and healthy simply by sharing information effectively.  Once that has been communicated however there is little to no value to spewing meaningless numbers around aside from stirring up the populace and making people upset.

There are two things that we can take away from this:  One, that people like to read useless statistics and can't properly figure out which ones are meaningful and which are noise.  Two, the writers and editors of newspapers know this and deliberately cram their stories full of numerical junk that can't be falsified but doesn't contribute anything to the discussion either.  I wish that media stories would be better about this sort of statistics porn but the fact is that until we get a population that is educated enough to reject this tripe in favour of real data (or an admission of not having any real data!) it is going to fill the evening news, the internet and local rags forever.

1 comment:

  1. Gonna use this w/ my Data Management class:) Great stuff.

    ReplyDelete