Man bites pundit
Regular reader Garrett asks for the house's take on this from electoral-vote.com:
Despite Delaware and studios full of pundits saying the voters are furious and 2010 will be a terrible year for incumbents, the reality is that 98% of incumbents have won their primaries. Only seven incumbents have gone down, but the media are completely consumed by those seven and have ignored the hundred of incumbents who have survived challenges. "Man bites dog" makes for better news than "dog bites man" but the latter is still a lot more common, despite the odd headline about the former.
The short answer is "yep," but the long answer is more fun. "Man bites dog" isn't exactly the right analogy, though. That's just the J105 value of novelty: the odd Hapsburg working at the Burger King. I like "shark bites man" better; it's a rare event that sometimes turns into a trend story and sometimes doesn't, without regard to whether or how the underlying numbers have changed.
You might recall that up through about (ahem) the first week of September, 2001 was (as Time put it) the "summer of the shark," even though shark attacks were actually down compared with the previous year. The occasional news story would come along to throw a little perspective on the numbers,* to little avail. Shark attacks were a Trend, and when when the fishwrap or TV station down the road has a trender, the odds are you want one too.
Shark attacks are rare, but they're rare in fairly predictable amounts. From 1999 through 2009, the US averaged a bit over 41 a year; Florida, which usually accounts for most US attacks, averaged almost 27. 2001 was on the high side of normal, with 50 attacks nationally and 34 in Florida.** But "a little on the high side of normal" isn't much of a hed compared with SUMMER OF THE SHARK!!!!
There are other numbers you don't see either. Evidently, the best thing you can to to avoid being bitten by a shark is -- not go to Florida! And if you go, stay out of the ocean. But given that you probably survived a trip by car (37,313 deaths last year -- a five-decade low) with some pretty dangerous people (more than a fifth of murders in 2002 were classed as family murders) to get there, I can see how you'd want to go in the water just to snap your fingers at fate.
A reasonably good piece in the Times today gets to the main point: People overestimate rare (but prominent) events and discount common ones. Journalists are playing SUMMER OF THE SHARK!!!! again, rather than putting the numbers into context. And some of that context involves noting that many risks are not normally distributed -- there are parts of most states where you simply aren't going to get bitten, no matter how hard you try.***
The SHARK phenomenon is a nice example of a pure agenda-setting effect. Almost from the outset, critics of the agenda-setting hypothesis have complained that the distinction between "telling people what to think" and "telling people what to think about" was specious -- as Bob Entman put it, when you tell people what to think about, you're telling them what to think. Here, there's not necessarily a valence. When you put your "Summer of Peril for Incumbents" story on the front, you aren't telling people it's a good or a bad thing, just that it's something they need to take into account. That doesn't put you in the clear, though. You're still obliged to show that what you're telling people to think about is true.
I like the story about whether incumbents are in trouble -- enough to think it ought to be fully reported before it's run.
* And a special Elongated Yellow Fruit award to CNN for "the cartilaginous fish"!
** Just over and just under one standard deviation, respectively.
*** Yes, a seven-term incumbent lost a primary just down the road. She's Kwame Kilpatrick's mom, if that helps.
You might recall that up through about (ahem) the first week of September, 2001 was (as Time put it) the "summer of the shark," even though shark attacks were actually down compared with the previous year. The occasional news story would come along to throw a little perspective on the numbers,* to little avail. Shark attacks were a Trend, and when when the fishwrap or TV station down the road has a trender, the odds are you want one too.
Shark attacks are rare, but they're rare in fairly predictable amounts. From 1999 through 2009, the US averaged a bit over 41 a year; Florida, which usually accounts for most US attacks, averaged almost 27. 2001 was on the high side of normal, with 50 attacks nationally and 34 in Florida.** But "a little on the high side of normal" isn't much of a hed compared with SUMMER OF THE SHARK!!!!
There are other numbers you don't see either. Evidently, the best thing you can to to avoid being bitten by a shark is -- not go to Florida! And if you go, stay out of the ocean. But given that you probably survived a trip by car (37,313 deaths last year -- a five-decade low) with some pretty dangerous people (more than a fifth of murders in 2002 were classed as family murders) to get there, I can see how you'd want to go in the water just to snap your fingers at fate.
A reasonably good piece in the Times today gets to the main point: People overestimate rare (but prominent) events and discount common ones. Journalists are playing SUMMER OF THE SHARK!!!! again, rather than putting the numbers into context. And some of that context involves noting that many risks are not normally distributed -- there are parts of most states where you simply aren't going to get bitten, no matter how hard you try.***
The SHARK phenomenon is a nice example of a pure agenda-setting effect. Almost from the outset, critics of the agenda-setting hypothesis have complained that the distinction between "telling people what to think" and "telling people what to think about" was specious -- as Bob Entman put it, when you tell people what to think about, you're telling them what to think. Here, there's not necessarily a valence. When you put your "Summer of Peril for Incumbents" story on the front, you aren't telling people it's a good or a bad thing, just that it's something they need to take into account. That doesn't put you in the clear, though. You're still obliged to show that what you're telling people to think about is true.
I like the story about whether incumbents are in trouble -- enough to think it ought to be fully reported before it's run.
* And a special Elongated Yellow Fruit award to CNN for "the cartilaginous fish"!
** Just over and just under one standard deviation, respectively.
*** Yes, a seven-term incumbent lost a primary just down the road. She's Kwame Kilpatrick's mom, if that helps.
Labels: journalism, polls
1 Comments:
In CNN's defense (not that they deserve it), sharks are pretty much the canonical example of Chondrichthyes....
Post a Comment
<< Home