This one comes from Ellen Wulfhorst on the Reuters wire:

Unlike traditional, mainstream media, blogs often adopt a specific point of view. Critics complain they can contain unchecked facts, are poorly edited and use unreliable sources.

I sense a great disturbance in the Schwartz, as if a million monitors were just sprayed with soda. (Well, no, I don’t have that many readers, but Dave Neiwert, Athenae and Coturnix have already picked up on it.) And here’s another puzzling statement from the same piece, describing the poll which found that “a majority of Americans do not read political blogs.”

The poll was conducted online from January 15 to January 22 among 2,302 adults. Harris said it does not calculate or provide a margin of error because it finds such figures can be misleading.

Is anyone else concerned by the sampling bias which this procedure could entail? If you ask people online what websites they read, you’re going to get (at best) a measure of what people who spend time online read, not what Americans in general are reading. Sure, that’ll probably increase the percentage of blog readers, skewing the poll towards “new media,” but it’s still bias. (They claim to have used “propensity score weighting” to “adjust” for this.)

On their website, Harris Interactive lays out their rationale for not reporting margins of error. Basically, they assert that people are too poorly educated to know what “margin of error” means: people don’t know that the phrase refers only to sampling error, not to other possible sources of obfuscation (which are harder to get a quantitative handle on). Therefore, people will assume that polls are more accurate than they really are; to avoid this problem, and to save the wear-and-tear on a newscaster’s mouth which the tediously long phrase “margin of sampling error” would produce, Harris will not admit fallibility at all.

Gee, how nice of them to make that decision for us, so that even the people who know statistics can’t get the figures. I just loooove suffering for the sins of innumerate America.

UPDATE: I also get a kick out of this:

Just one in ten (19%) Echo Boomers (those aged 18-31) regularly read a political blog

One in ten. Nineteen percent. Oops.

## 6 thoughts on “Your Funny for Today”

1. gg says:

Unlike traditional, mainstream media, blogs often adopt a specific point of view. Critics complain they can contain unchecked facts, are poorly edited and use unreliable sources.

I’ll grant that the mainstream media seems much more impartial than most political blogs, but that’s because they never report on anything of substance!

I wouldn’t know, for instance, anything about the current FISA telecom immunity battle without the political blogs. If I read CNN today, I find that the stocks were up for the day and that Spitzer’s escort ‘loves who she is.’ If I were to exclusively read/watch FOX, I’d have learned all about how Bush invaded Iraq because Saddam was only days away from making nuclear and chemical weapons in mobile homes and flying them over to the U.S. on unmanned drones for Al Qaeda to use to bomb the library tower, and that because of the invasion Saddam moved all his WMDs to a turkey farm in Syria and got the Iranians to help expel U.S. troops.

As Atrios would sarcastically say, time for a blogger ethics panel…

2. Just assume it’s counting error, Blake. Google tells me that sqrt(2302)/2302 = 2%, so there you have it.

3. Everything in life comes back to $$1/\sqrt{N}$$, doesn’t it? Hooray for normal distributions.

4. Question:

Has anybody done a study on how the perceived reliability of polls is affected by including terminology like “margin of error”? That is, if people are not reminded that errors are possible, do they tend to discount the possibility? I couldn’t find any research on the topic, but I missed lunch today and am therefore likely to be both cranky and inept.

5. yapparister says:

Don’t know of any studies but I have been thinking about the 1 in 10 and 19 percent… reminded me of the fact that the most humans have more than the average number of legs. On that basis I suppose it’s more accurate to say 1 out of 10 for 19 per cent – as 2 out of 10 is plain wrong, and ‘1.9 out of 10 people’ makes no sense at all. Also hungry and lacking sleep…

6. But “2 out of 10″ is roughly right, and “1.9 out of 10″ does make sense — it just means “19 out of 100.”