Friday, March 30, 2012

Another pet peeve I thought I'd share

...because I feel like it's been a while since I've shared a pet peeve.

So I'm sitting here getting some work done with the TV on, and some car insurance commercial comes on.

I hate car insurance commercials. Why? Because they're either consistently bad at interpreting data or they're good at it and are being deliberately dishonest - and I don't like either of those things.

So on this one they talk about how the average person who switched from Allstate to this company "saved an average of $547". Argh - nails on a chalkboard. I'll let readers take a shot at why this bothers me before I answer.

UPDATE: So the problem here is a substantial selection bias. They always say that $X are saved by "switching" to their company. Often they'll go as far as this ad and even say $X are saved by switching to their company from a specific other company. That's all well and good, but the problem is you only switch companies if you get a better deal! There is no indication of how your expected savings will compare to this. Let's say insurance companies assessed everyone exactly the same premium on average, but there was some random variation across insurance companies for any given person. So no insurance company gives a "better deal" than any other insurance company. Now let's say you initially buy insurance after some kind of limited search process. In the future, you may switch. Because there's some randomness to it, there may be another company out there that would actually give you a better deal than you initially got. But the point is you'll only switch if they give you a better deal. So any insurance companies that assess their quality on the basis of the savings of switchers are always going to be able to tell you they can save you a bunch of money!

Different coverage is offered by different companies, etc. - and that complicates reasons for switching. But for me at least, insurance company advertisements are basically giant flashing billboards for selection bias. Enjoy the geckos and laugh at Flo, but forget whatever numbers they spew at you. Don't believe anything about any "savings" until you actually shop around for it yourself.


  1. Because you're a socialist and you can't stand competition in insurance markets?


  2. I have no idea, but I'll make two guesses,

    1) some type of criticism of the use of the word average.


    2) dishonesty in that they imply you'll save the average.

  3. Oh my gosh are you seriously going to make us keep guessing? Landsburg had a thing in one of his books, complaining about grocery stores that would have ads saying, "If you take the shopping carts of 100 of our customers, and bought the same items at Competitor X, it would cost y% more." Landsburg said that was dumb because they'd have different sales and of course people would pick the cheap stuff from each store.

    Is that what you're getting at? I.e. the people who switch are probably able to find a better deal? But I'm not sure it works that well in this case, versus the grocery stores.

  4. Savings aren't the only reason people might switch insurance companies. An insurance company could earn customers by consistently offering better customer service, or it could by offering more extensive coverage (not just savings, but coverage that other companies don't provide or any price at all -- for example, AAA was at one point one of the few, or the only, company which offered roadside assistance). There are insurance companies which do consistently offer lower rates. Some may offer lower rates, but poorer services -- they do this because they target lower income clients.

    1. Right - like I said different coverage (and you can throw customer service in there too) complicate matters, but it doesn't change the selection bias, which has got to be the major component of this. What percent of switchers do you really think pay more money for the same product? Because that's the other thing - once we get into discussions of customer service and coverage, we're talking about a somewhat differentiated product.

      The selection bias in that number is still substantial.

  5. It should be "the average person saved $x," not "the average person saved an average of $x."


All anonymous comments will be deleted. Consistent pseudonyms are fine.