Two friends have commented on the tanning bed example in my previous post. I’ll highlight the example’s weaknesses and add some information to clarify:
1. It reads as an apology or justification of tanning bed use. If you know me, you know that I’ve used tanning beds before so it’s easy for it to appear this way.
2. The argument might have been more clear if I quantified it. Let’s say 10 minutes in a tanning bed delivers the UV equivalent of about 30 minutes of real sun. My point is that 40 minutes in a tanning bed per week is probably equal in cancer risk exposure to 2 hours per week in the real sun, however the tanning bed user / cancer link makes it seem like tanning beds cause cancer in and of themselves.
3. The point could have been even more starkly made if I pointed out that the demographic who uses tanning beds is the demographic most likely to get skin cancer: fair-skinned people.
4. There’s a general weakness in the example because I’m speculating about something that might or might not be true. This makes it confusing because I’m asking readers to imagine a hypothetical situation while delivering a persuasive argument as to why the hypothetical situation might be real.
5. I thought it was a good example because it has all the components I wanted to include: a clear G, an elusive X, an A caused by erratic behavior, and a familiar correlation that’s often assumed to be causality.
6. A better example might have been divorce rate. “50% of marriages end in divorce.” That well-known statistic is based on the marriage rate being twice the divorce rate. The site below (I’m having hyperlink issues) states that the divorce rate is 41% for marriage #1, 60% for marriage #2, and 73% for marriage #3. So while a young couple getting married may be thinking that there’s a 50% chance they’ll divorce their partner, the truth is those odds are lower. The extremist group—people who get divorced 2 to 7 times—are raising the divorce rate data for everyone.
In an earlier post, I outlined what I called the extremist group fallacy. This is a logical fallacy in which group G is observed as having an attribute (or exhibiting a behavior) A, but in actuality A is attributable only to a subgroup of G. That subgroup is X: the extremists.
G is an easily identifiable group, whereas X is elusive. X has or does A so much that it elevates the rate of A for G. Therefore statistics show that G disproportionately has or does A more than the rest of the population P. These statistics may mislead and let observers conclude that being in G causes A (causality fallacy is at play as well). It could be that if we remove X from G, the remainder of the group (G-X) actually has A just as much as P.
Here’s a made-up example: tanning bed users (G) are observed as having higher rates of skin cancer (A) than the rest of the population. This makes us want to conclude that tanning beds cause cancer. In actuality, tanning addicts (X) will naturally seek out tanning beds because the real sun just isn’t enough to satiate their addiction. So they are an extremist subgroup within G who, due to their excessive UV exposure no matter the source, are very likely to get skin cancer. Tanning beds are blamed as causing skin cancer, but the real culprit is extreme and obsessive tanning.
Extremist subgroups and their annoying or erratic behavior cause a lot of problems. They get a lot of attention from the media because of their behavior. They become an unlikable caricature of their group, which foments tribalism. Here are some examples:
|When people dislike this group (G):
||They may actually dislike this behavior or these attributes (A) exhibited by an extremist group (X):
||corporate greed, racism, evangelicalism
||militant progressivism, political-correctness, Marxism
||corrupt leadership, self-righteous judgmental people
||smug elitism, amorality
||entitled, naive youth with fragile egos
In these cases, the extremists are a minority within the group—a very loud minority. It’s so easy to overestimate the predominance of X within G.
How do you mitigate this?
I don’t think you can do it by statistically educating people, because people only believe statistics that support their preexisting worldview. There’s one way to help people realize that their caricature of a group is inaccurate: force them to interact with those people.