Personally, I don’t believe in probabilities

That is, I don’t believe in assigning numerical values to them. That practice seems to me a mendacious attempt to quantify the unquantifiable.

This came up because of this email from a friend today:

When I was in high school, I once wondered whether a 50% chance of rain meant: A. there is a 100% chance that it will rain on 50% of the city or B. there is a 50% chance that it will rain on 100% of the city.  I think the real answer is that there is a 50% chance it will rain at least somewhere in the city (which is actually less than a 50% chance that it will rain in any one place).  It’s painfully obvious that my chances at a STEM major in college were much lower than 50%!

For my part, I don’t believe there’s such a thing as a 50 percent chance of rain, period. I don’t mind someone saying, “I have as much reason to believe it will rain today as I have reason to believe it won’t.” But don’t insult me by attaching a number to it. You don’t know enough to attach such precision to the matter.

You want to assign a numerical value to the likelihood of rain on Thursday? Here’s my advice: Wait until Friday. Look back, and if it rained on Thursday, there was a 100 percent chance. If it didn’t, there was a 0 percent chance. The rest is nonsense.

36 thoughts on “Personally, I don’t believe in probabilities

  1. Lynn Teague

    Apparently the Weather Service doesn’t feel up to explaining second standard deviations to people who say that they don’t believe in probabilities. They certainly don’t want to get into whether the Pearson Type-III (P3) distribution fits the observed data better than the G2 distribution (a quick Google search tells me that faculty members at the Engineering Dept at Tufts are big fans of P3 for good fit with the U.S. 30 year rainfall data). However . . . factor in the specific conditions affecting the probability of rainfall on a particular day, along with the price of catnip, and there is a curve predicting the probability of rain. If you are in the Atacama Desert you can multiply the resulting numbers by Lindsey Graham’s current poll numbers for a pretty good idea of how likely it is to rain.

    As to what the single simplified number that we get from the Weather Service means – according to NOAA, if they are 50% sure that precipitation will occur, and if it does occur it is probable that it will produce measurable rain over about 80% of the area, that translates to a 40% probability of rain.

    1. Brad Warthen Post author

      Ha, and also ha.

      I should clarify that I believe in probabilities (and there’s a distinct chance that I’m not using the term as statisticians do, which doesn’t particularly concern me) when they describe things that are more measurable.

      For instance, it doesn’t bother me when we say a poll has a 4.6 percent margin of error. In part, that’s because I know there’s a fudge factor. What that margin statement means is that 95 percent of the time, the result of canvassing the entire population would be within 4.6 percentage points either way. That’s demonstrable, and it’s based (I think) on experience. Of course, it assumes that you have a statistically valid sample.

      Anyway, it comforts me that there’s an admission that 1 out of 20 times, the result would NOT be within that spectrum…

      But the weather? Fuggedaboudit. Way too uncertain. Use words, not numbers. Say that, based on our best projections, there’s sufficient reason to believe the hurricane will head this way that we should begin preparations to evacuate. Say “high probability” if you like. Don’t give me percentages…

  2. Kathryn Fenner

    What is your stance on evolution?
    Weather forecasting is pretty sophisticated, based on mountains of data. It maybe used to be pretty loose-y goose-y, but I think it’s pretty valid nowadays. Just because you can’t wrap your math-challenged mind around it….

    1. Brad Warthen Post author

      First, I’m not math-challenged. My math SAT was impressive by almost anyone’s estimation.

      I just don’t find numbers as meaningful as words. And there are situations in which I believe the use of numbers creates unreasonable expectations of precision.

      And when I give in and trust such numbers, I am generally disappointed.

      Such as the time year before last when I helped my daughter move out of an apartment in Charleston. I loaded as much of her stuff as I could onto my pickup. Then I stopped at Walmart to get some bungee cords. I checked my weather app to see whether I needed to buy a tarp to protect her stuff. The app told me that there was 0 percent chance of rain in Charleston, and 0 percent in West Columbia.

      So I set out without a tarp, and halfway there, the heavens opened. Admittedly, I had neglected to check the forecast for points BETWEEN Charleston and W. Columbia, but in the end that was immaterial, because the downpour continued all the way to my driveway. A lot of her stuff was ruined. I felt terrible about it, especially since I NORMALLY don’t trust forecasts and use a tarp anyway. I was just being a cheapskate that day.

      So don’t tell me I’m being some kind of hammerhead who doesn’t believe in science. If those forecasts are an example of the kind of science you’re talking about, then science is bunk…

      1. susanincola

        Ah, the popular confusion that precision == accuracy. (Not for you, but for those who think having a more precise number automatically makes the information more accurate).

        1. Brad Warthen Post author

          Perhaps I expect too much of numbers, and they disappoint me, which makes me bitter. Because to me, if numbers can’t be precise, they’re useless. Precision is what numbers are FOR.

          For instance, I hate balancing my checking account, because it doesn’t come out right. It’s been 20-something dollars off for several years now, and way back when it happened, I was unable to pin down the error. And now it would take asking the bank for records that I don’t have any more, and a long weekend, to work it out, and every time I threaten to do it, my wife says I’m crazy.

          When inconsistencies like that come up in her accounts, she just shrugs and goes with the bank’s total. I cannot do that. I’m too much of a number Nazi.

          Here’s the thing about numbers: If I use them, I expect them to be dead-on balls accurate. That’s what numbers are for — for precision, not just being in the ballpark. If I actually have $235.64 in a particular account, it’s not cool for my spreadsheet to be coming up with $257.38. It’s the difference between RIGHT and WRONG. I either spent that penny or I did not.

          I resent that my account doesn’t come out right. It ticks me off. (Oh, and don’t think I’m blaming the bank; I assume the error is mine. I’d just like to reconcile it.)

          1. Brad Warthen Post author

            It’s kind of like the way I got disgusted with Democrats in the Legislature because they wouldn’t stand up and advocate for a gas tax increase to fix roads.

            If DEMOCRATS can’t stand up for a tax increase when there is every reason to do so, and there simply is no credible argument to make against it, then what use are Democrats?

            And if numbers can’t be precise, then what use are numbers?

      2. Norm Ivey

        A probability of rain forecast is only good for a set period of time, often only for a few hours. While the chance of precipitation in West Columbia may have been zero when you checked it in Charleston, in the two hours it took you to drive from one to the other, the probability changed.

    2. Brad Warthen Post author

      As for my stance on evolution…

      If you say to me, “The available evidence overwhelmingly indicates that humans are descended from such species as homo erectus,” I’ll agree with you. (And then I’ll make like Beavis and Butthead and say, “Huh-huh — she said ‘homo erectus!’ Huh-huh-huh…”)

      If you try to tell me there’s an 85 percent chance that that’s true, or 90 percent, or 99.44 percent, I’ll ignore you and your numbers…

  3. Brad Warthen Post author

    You know what I wish my weather apps would do? I wish they would tell me what the weather WAS, instead of what they think it’s GOING to be.

    I often wonder, “How hot (or cold) did it actually get to be today?” And that info is a little less readily available, which I find disappointing.

    Maybe it’s because I’m a journalist, or because I had a second major in history. I’m just more interested in what has happened than in projections of what WILL happen, however scientific they may be…

      1. Brad Warthen Post author

        Good for The State! That didn’t used to be there.

        But you know what I mean. I get a flood of info about forecast, and I generally have to deliberately seek out the info about what has occurred…

        1. Bryan Caskey

          “I get a flood of info about forecast, and I generally have to deliberately seek out the info about what has occurred…”

          Well…yeah, that’s because only a handful of crazy people like you care about what the weather was yesterday.

      2. Brad Warthen Post author

        And you know what this illustrates? It illustrates that I am far more dependent now on my devices than on the paper version.

        In fact, I seldom glance at it…

        1. Brad Warthen Post author

          It’s actually THERE, on my preferred app for The State — the one that shows the actual newspaper pages, which I tap to enlarge.

          But I never had reason to tap on the nameplate to see what was up there in the “ears.” I suppose I thought I knew…

          1. Brad Warthen Post author

            Stuff like that — say, the folio on each page, where you find the page number and date — is THE easiest thing for an editor to miss in proofing a page.

            It’s called “the furniture.” It’s easy to overlook when you’re obsessing over the body copy…

  4. Brad Warthen Post author

    Another example of what bugs me about the mania to quantify…

    My son-in-law is an economist. He once showed me a paper by some colleagues who had taken it upon themselves to come up with a way to quantify “bias” in news media. When I saw what they had done, it made me want to tear my hair out.

    It was all propped up on a list of words. The bias quotient would be based on the incidence of those words in news copy.

    Well, you know, as an editor, I could easily think of reasons why you might find those words in copy that had nothing to do with bias.

    It’s not the words… it’s how they go together. It’s context. It’s what’s left out as much as what is included.

    Bias in news is like obscenity. If you have good reading comprehension, you know it when you see it. And even then, you could be wrong…

  5. Brad Warthen Post author

    The problem is the preponderance of people in the world who believe that something is not a fact, is not valuable information, unless it has a numerical value.

    Just give me the words…

    1. Lynn Teague

      Now that much I can agree with. Always drove me nuts that there are archaeologists who think if they run their dubious data through a computer it is somehow purified and made whole.

  6. Norm Ivey

    Forecasters evaluate atmospheric and surface conditions (temperature, relative humidity, wind, air masses and fronts, barometric pressure). Based on data, they assign a number to their confidence. If it rains 50% of the time when these conditions exist, their confidence is 50%. As Lynn explains, the confidence factor is multiplied by how much of the forecast area is likely to be affected. If data/history tells them that only 50% of the forecast area will see rain, and the confidence factor is 50%, the chance of rain or Probability of Precipitation is 25%. They are reporting a correlation based on enormous amounts of data. It’s specific enough that the probability of rain for my area (NE Columbia) is often different from the probability of rain downtown.

    Your individual probability of encountering rain changes if you move from one area to another. The chance of rain is 50% for Columbia, and you are on Elmwood and there is no rain. You drive to Five Points. You have increased your individual odds of getting rained on above 50% even though the overall odds for the city have not changed.

    1. Brad Warthen Post author

      Speaking of the rain…

      I started this thread about probabilities (particularly with regard to weather), without being fully cognizant of all these predictions of rain of biblical proportions, which I only focused on last night.

      That’s because generally speaking, I filter out news about weather. If it’s cold when I go out, I’ll put on a sweater. If a hurricane’s coming, I’ll learn about it in spite of myself. Although I have an app for it on my phone, I don’t check it all that often.

      When I heard about all the hullabaloo about the rain last night, I remembered that that morning, when I turned on the tube to switch to Netflix to watch an episode of “Gotham” while working out, a weather girl was saying something about a hurricane, and I switched her off in mid-sentence. Which made me feel foolish later, like maybe I SHOULD have paid attention.

      When it comes to weather, I’m kind of like Simon Pegg in “Shaun of the Dead.” Did you see that? Remember all the indications that he ignores in the early scenes about the arrival of the zombie apocalypse — such as flipping past all the news stations where they’re talking about it?

      That’s me with regard to the current weather situation…

  7. Brad Warthen Post author


    Speaking of weather apps giving screwy info. I just glanced at mine, and it displayed the above.

    “Tornado,” eh?

    I kept waiting and it never did give the temp. It was too busy running the animation, making all that debris swirl around.


Leave a Reply

Your email address will not be published. Required fields are marked *