1000-Year Flood of Statistical Ignorance



I ran across, but can't at the moment relocate, a piece that claimed something like, "We just saw a 500-year flood in Houston. And 100-year floods in X, Y, and Z. All in the last 5 years!"

Obviously, what we were meant to conclude is that there was no way we could have had four 100+ year events in a five year period without global warming being the cause. Here is a similar piece, in which uber-idiot Naomi Klein says that "The records being broken year after year..." prove that man-made climate change is real, and a disaster. Without the least bit of curiosity as to just how often we should expect weather records to be broken. The world is a pretty big place, and Klein tosses out four categories of records -- "whether for drought, storm surges, wildfires, or just heat" -- so it is very likely that somewhere in the world, a record for one of those things is being broken pretty regularly. Perhaps these record breaking events really are happening more frequently, but Klein doesn't provide a shred of evidence beyond, "Well, Jeez, just look at all those records, will ya?"

In contrast, here is a nice, calm (and non-"denialist") explanation of the meaning of "X-year" weather events:

"As it turns out, the country experiences multiple 500-year flood or storm events (that is to say, an event that in had a 1 in 500 chance of occurring in that given place) every single year."

In short, if we divide the United States into 100 metropolitan regions plus their hinterlands, on average, we'd expect one 100-year event per year, since we'd now have 100 chances for 100-year events. (I'm ignoring the fact that these events are not strictly independent, because I don't think that falsifies the real picture too much, e.g., Houston just had a 500-year flood, but Corpus Christi, San Antonio, and Austin were barely dampened.)

By the way, in an effort to explain the "X-year" weather event concept, writers for "statistical" sites such as 538 are... showing they don't understand basic probability:

"In the wake of catastrophic flooding on the Texas coast, the media has been working hard to explain the term, turning out dozens of articles explaining that a "100-year flood" is not a flood that you should expect to happen only once every 100 years. Instead, it refers to a flood that has a 1 percent chance of happening in any given year."

Hmm, and if an event has a 1% chance of happening in a year, what is its expected frequency? Once per 100 years!

Comments

  1. "A 100-year flood" is not a flood that you should expect to happen only once every 100 years"

    In any given 100 year period there is >50% chance that the event woth 1% probability will NOT occur exactly once - so it is not totally wrong to make that statement.

    ReplyDelete
    Replies
    1. Yes, that's why we call it "expectation," not "certainty."

      Delete
    2. Well, 538 isn't wrong to point out that 1% chance per year doesn't mean that the odds of twice in a century are all that low. Which is more or less the kind of nonsense the AGW alarmists are peddling. There are two senses of "expect" here.

      Delete
    3. 'Yes, that's why we call it "expectation," not "certainty."'

      Well yes, but my point is that the two sentences from the article you quote don't show any lack of understanding of basic statistics. The two sentences are both correct and totally consistent with your 'if an event has a 1% chance of happening in a year, what is its expected frequency? Once per 100 years!' statement.

      Delete
    4. So, "a "100-year flood is not a flood that you should expect to happen only once every 100 years" is consistent with the expectation of such a flood being once every 100 years?!

      Delete
    5. The key word here is only. People seem to think that if you see such a flood twice in a short period it cannot be a "100 year flood". But it can. If you roll snakes-eyes twice in a row it's wrong to conclude that snakes is not a 1/36 event.

      Delete
    6. Yes, OK, but then it should have been written better: "A 100-year flood is not a flood that CAN only happen once per 100 years."

      Delete
    7. Ken: Thanks
      Gene: Yes, I agree

      Delete

Post a Comment

Popular posts from this blog

Libertarians, My Libertarians!

"Machine Learning"

"Pre-Galilean" Foolishness