How much longer do compact fluorescent bulbs last?

Though a CFL bulb costs more than a standard light bulb, it has two features that make it cheaper in the long run: It uses less electricity per hour of use, and it burns for more hours before dying. I had been under the impression that both of these features would, in and of themselves, save enough to make up for the initial higher price -- this makes them doubly a good deal.

The electricity argument is true so far as I can tell, but a new Wall Street Journal piece claims that "The light bulb that costs 10 times as much does, it is true, last four times as long." This itself deviates from numbers from earlier in the article, which said regular bulbs cost 50 cents, as opposed to $3 for CFLs (six times the cost, not 10).

Wikipedia seems to agree with my initial impression:

Modern CFLs typically have a life span of between 6,000 and 15,000 hours, whereas incandescent lamps are usually manufactured to have a life span of 750 hours or 1000 hours.

The lower-end difference is eight times; the higher-end is 15 times. Both are higher than six times.

Every source I can find says that CFLs last longer to at least (approximately) the degree they cost more. Consumer Reports says the initial cost is closer to $2 per CFL. The American Lighting Association says CFLs both cost and last 10 to 15 times what normal bulbs do. The government says CFLs last up to 10 times longer.

Where did the WSJ get its numbers?

UPDATE: In The Weekly Standard, Andrew Ferguson writes, "With proper care and moderate use, they can last as much as six times longer than a typical incandescent. Even if you consider their higher purchase price--six or seven times the price of a traditional bulb--CFLs can lower your monthly lighting bill by as much as 20 percent." No citation there either, but that sounds more reasonable. (There's some mercury hysteria in the piece, though. Some states in fact allow CFL disposal in regular garbage, and the options for recycling are growing as CFLs become more popular. 3 mg of mercury will not kill you, so calm down.)

I totally beat The New York Times

With this article from The American about the Wii's exercise potential. A new study shows that the Wii burns more calories than other video game systems do, but that it (shocker!) doesn't burn the same calories as actually playing the sports the games imitate would.

In fact, I think I'm still one step ahead of the NYT: Like the study I cited in my piece (contrary to what the NYT says, data was not "lacking" until now), the new one used Wii Sports, which demands an unusual amount of movement for a Wii game. With most other games, you point the controller at the screen and push some buttons, as opposed to moving around and acting out sports maneuvers. Thus Wii-playing in general isn't all that great for your health, and the story totally misses the boat.

Finally, I have a hard time believing that tennis is the "most active" Wii Sports game -- for me, boxing caused a lot more aches and pains.

Mercury = death!

I hate environmental crusades as much as the next guy, but the right-wing attempt to vilify compact fluorescent light bulbs is starting to grate on me. Numerous times in the past few weeks I've read about how, because the bulbs contain a little bit of mercury, you'll have to call in the authorities! if you ever happen to break one.

It's too bad, because usually it's conservatives who put chemical risks in perspective: The dose makes the poison. There's about 3 mg worth of mercury in a bulb, about enough to cover the ball point of a pen. If you break a bulb, all you have to do is open a few windows, sweep it up and wipe the area with a damp cloth. Not a big deal, and not a deal at all if you don't break your light bulb.

Let's put that in perspective. A common guideline for fish is that you can have .5 mg of mercury for each kg (1,000,000 mg) of fish. A 6-oz can of tuna converts to 170,000 mg. A back-of-the-envelope calculation indicates that, if you consume a can of maximum-allowed-mercury tuna once a week for two years, you will have eaten the amount of mercury that's in that light bulb. Certainly it won't kill you to simply handle it for a bit.

Rather than pretending CFLs are a bad idea on net, conservatives should lobby against making them mandatory. Competing with standard light bulbs forces CFL makers to (A) convince the American people of the substantial cost benefits, (B) find ways to make the product cheaper and (C) work to cut down on the annoying brightness some (in particular, morons who can't spell "fluorescent") find with the bulbs.

UPDATE: I guess I missed this American Spectator piece when it came out, but it's about perfect on the issue.

Guns and the election article up at The American Spectator

Here it is.

Main point:

When Americans nominate their presidential candidates next year, the Second Amendment won't be the first thing on their minds. The issue didn't even appear in a recent CNN poll that found that the economy, Iraq, health care, immigration, and terrorism are the nation's biggest concerns.

But in a country where 36 percent of Democrats and 48 percent of Republicans have firearms in their homes, the issue is still a locked and loaded one for candidates of both parties.

Parenting and school performance

The Freakonomics blog has a post from Ian Ayres, who observes from a New York Times article:

...researchers used four variables that are beyond the control of schools: the percentage of children living with one parent; the percentage of eighth graders absent from school at least three times a month; the percentage of children age 5 or younger whose parents read to them daily; and the percentage of eighth graders who watch five or more hours of TV a day. Using just those four variables, the researchers were able to predict each state's results on the federal eighth-grade reading test with impressive accuracy.

It turns out that, in this regression, single-parent families have no significant correlation with test scores, and that's the point of Ayres's post. But of course, no one bothered to factor in something else schools can't control: students' races. I did so, expecting to find that, as usual, race is the elephant in the room. I was largely wrong.

Ayres is nice enough to provide the spreadsheet with which he proved single-parent families to be statistically insignificant. Here's my update of it, using the three variables he found useful and adding Census figures for the percentage of kids who were black and Hispanic in 2006. Some of the states were not available in both data sets, so I removed them, leaving me with 44 observations. For some reason these tended to be whiter states with high test scores, so this could weaken the correlations with the racial variables.

(I used American Factfinder to make a state-level custom table of total males 10-14, black males 10-14 and Hispanic males 10-14, then divided blacks and Hispanics by the total. I'd have included all kids instead of just males, but I couldn't find that number already made, and it seemed silly to clutter up the spreadsheet with six more columns that are almost the same as three already there. The black male share of the 10-14 male population will be virtually equal to the black share of the total 10-14 population.)

The result: The two race variables alone explain .43 (adjusted r-squared) of the variation, statistically significant but worse than Ayres's .63. In fact, when you put all five variables in one regression, "percent black" and "percent read to" become statistically insignificant. I toyed around with the numbers quite a bit, and the best adjusted r-squared I can get where all the variables have statistically significant effects is .66, a two-variable one with TV watching and percent Hispanic -- these are the only two variables that stay significant no matter what they're paired with.

The key with the Hispanic data is to remember that the test is of reading, so students who grew up with Spanish are at a significant disadvantage. Thus their population explains much of the variation in state test scores.

Also, the information implies that heavily black states' low scores are explained better by parental behavior (especially monitoring the TV) than by race itself. State-level data can be tricky, so this is far from conclusive, but score another one for nurture over nature.