Did You Know...



By thisistwitchy  •  March 1, 2005 02:15 PM

According to the Wall Street Journal, “Some eight million Americans now publish blogs and 32 million people read them, according to the Pew Internet & American Life Project.”

8 million bloggers! 32 million readers! John Hinderaker doesn’t believe the hype, and neither do I. Even mighty Instapundit averages less than 200,000 visits per day–does anyone seriously believe that he is capturing less than 1 percent of the blogosphere’s readers?

There seem to be a lot of implausible blog traffic numbers floating around lately. Part of the problem is confusion between “visits” and “visitors.” If I visit Power Line 5 times on a given day (which is not unusual), that counts as five visits but I’m still only one visitor.

Unfortunately, even the savviest bloggers blur this simple distinction. Hugh Hewitt, for example, is stretching the truth when he says “more than 10 million people” have visited his blog since its inception in 2002. His site may have been visited 10 million times, but that’s not the same as 10 million visitors.

So how many people read political blogs? Let’s make some back-of-the-envelope estimates. Start with the two biggest blogs:

-According to the Pew survey, 120 million American adults “use the internet.” According to Alexa.com, 245.5 people per million users read Daily Kos on any given day. That implies 29,460 U.S. adult visitors per day, which is surprisingly consistent with the figure (34,000) reported by Comscore Networks for November 1, 2004, here.

-Alexa says 208.5 people per million users visit Instapundit on any given day, which means Insty is visited by 25,020 American adults per day.

Let’s continue, again using Alexa’s reach figures multiplied by 120,000,000 users. Ignoring minors, we get the following results:

Power Line: 19,200 U.S. visitors per day
Little Green Footballs: 17,700 U.S. visitors per day
Talking Points Memo: 16,800 U.S. visitors per day
Wonkette: 15,240 U.S. visitors per day

These figures seem shockingly low when compared to the hit counts based on SiteMeter statistics. The main reason, again, is because SiteMeter is counting visits whereas the above figures refer to visitors. A smaller reason: the SiteMeter figures capture non-U.S. readers and minors whereas the above figures do not.

Here’s a validity check. We know an Instalanche typically produces somewhere between 800 and around 4,000 hits. Is the size of an Instalanche consistent with readership of 25,020 U.S. adults per day? I think so. In other words, if Instapundit were visited by, say, 75,000 – 100,000 people per day, I would expect the Instalanche to result in more than 800 to 4,000 hits.

There is one other issue that must be considered here–namely, the fact that most people read more than one blog. There is probably substantial overlap, for example, between readers of Daily Kos, Eschaton, Talking Points Memo, and Wonkette. Same thing with Power Line, Little Green Footballs, The Corner, Best of the Web, and Instapundit. By the time you start getting down to smaller blogs like mine (estimated at 10,200 visitors per day based on Alexa), I’d guess the percentage of readers who are “unique” visitors (i.e., visitors who do not also read one of the larger blogs) is very, very low.

If Alexa’s figures are in the right ballpark, the top 100 political blogs, in the aggregate, probably average about 100,000 U.S. adult visitors per day. That’s not small potatoes but it is a far cry from the inflated Pew figures cited by the Wall Street Journal this morning.

Update: A couple of readers have pointed out that the 32 million figure cited by the Wall Street Journal included non-political blogs. According to John Hinderaker, the Pew study suggests there are 4.8 million self-described “regular” readers of political blogs. That’s a lot more accurate than 32 million, but it’s still wildly inflated.

Update II: John Hawkins of Right Wing News thinks I’m off base. In an e-mail, he says, first, that the Alexa rankings are not very reliable. (An odd argument coming from someone who publishes rankings every year based on Alexa’s statistics.) For example, he says Alexa shows Daily Kos getting more visitors than Rush Limbaugh’s site, which Hawkins considers implausible. I’m not so sure.

He also points out that Daily Kos raised $537,079.79 for Democratic candidates last year. “I don’t think you can raise that kind of money with a few people chipping in,” he notes. That’s a strong argument, though it’s possible that Kos’s readers are more devoted than Hawkins gives them credit for.

Finally, Hawkins says he has a statistics tracker that shows him getting between 7,500 to 8,000 unique readers looking at his site on an average day. “That’s not impressions, that’s unique readers,” he says. “Now that can be inflated a bit, because I don’t know if they use IPs or cookies to track people, but it shouldn’t be off THAT much.” By comparison, Alexa currently shows his site getting 3,600 U.S. adult visitors per day (30 visitors per million users). So: if Hawkins’ tracking packaage is the gold standard, Alexa is undercounting his visitors by a factor of about 2. If Alexa is undercounting other sites by the same amount, that would mean Kos is getting 60,000 vistors per day rather than 30,000, whereas Instapundit is getting 50,000 vistors per day rather than 25,000. It would also imply that the top 100 political blogs attract somewhere around 200,000 U.S. adult readers per day, not 100,000 as I guesstimated above. Either way, we’re still talking numbers that are significantly smaller than those being tossed around by Pew.

By the way, Hawkins is quitting his job and will be blogging professionally. If you’ve got a little extra dough, please consider sending some his way.

Update III: Charles Johnson disagrees with my analysis here. As he points out, Alexa’s statistics are based on surfing patterns of the small subset of internet users who use Alexa’s strange browser. Johnson considers this subset to be unrepresentative of the broader population. He implies that Alexa users are less likely to visit political blogs than internet users who use other browsers and concludes that SiteMeter’s count of daily visits is a “much more accurate” proxy for the number of daily readers.

Of course, Alexa’s statistics aren’t perfect. As I suggested in my response to John Hawkins (see above), Alexa appears to understate readership by a factor of about 2. (This is consistent with information sent to me by another top blogger.) But at least Alexa measures the correct metric, i.e., the number of readers per day. SiteMeter, by contrast, is measuring the number of visits, a metric that by definition exceeds the number of visitors unless every visitor checks in exactly once each day. Obviously, I can’t speak to Johnson’s internal tracking data, but in general I would expect the discrepancy between visitors and visits at the top political blogs to be quite large. Especially so at a site like Instapundit that is frequently updated or at sites like Daily Kos, Eschaton, and, yes, LGF, which allow reader comments.

In pointing all of this out, I don’t mean to diminish the success of these blogs in any way. We all want readers to check in repeatedly.

But we should also want the MSM to report accurately on the true reach and scope of the blogosphere’s audience. And a reality check–ain’t that what we’re for?–was in dire order.

Finally: Although I think Pew’s statistics are wildly inflated, I am not, not, not pooh-poohing the importance of the blogsphere. Bloggers’ influence depends in part on the number of readers, yes, but the more important factor is who our readers are. If I didn’t believe in the power of blogs, I wouldn’t be here. 🙂

Update IV: Mick Stockinger makes an interesting point:

I think MORE people are reading blogs, than are credited by services like Sitemeter. Sitemeter only counts actual visits to the webpages, ignoring RSS feeds. I can read Malkin’s entire blog without having to visit her site and so according to Sitemeter, I never did.

It should be noted that Alexa also fails to capture this traffic.

Update V: Michael Totten pointed out in January that different statistical tracking services provide rather different visit counts. In his case, SiteMeter said the average number of visits per day on his blog was 3,200, whereas Webalizer said his daily average was almost 6,000 for the same time period. If these statistical packages cannot even agree on the number of visits a site gets, then using one of them as a proxy for the number of visitors, a completely different metric, seems absurd. (Hat tip: Pundit Guy.)

Update VI: Reader Gerald Brown writes:

I am an avid reader of political blogs. I read through yours each morning, and I keep The Corner and TKS up all day. And, I occassionally visit
others, as they are specifically recommended by you and others. However, I
set my browser to not accept cookies from third parties (with a few
exceptions) when I visit a site. That means that sitemeter (and ad sites)
cookies get rejected by my machine. And, I know I am not the only one who
does this. So, every time I show up to a blog in a day, I most likely count
as a “unique” visitor – no matter how many times I have visited that day.
With privacy concerns being what they are, I am sure a lot of statistics
about site visitors is skewed by people like me.

Update VII: My web guru Mark Jaquith writes in:

“Gerald Brown” says that he disables cookies, and so might count as a unique visitor each time he views a page. This might be true if you were using a tracking system that used cookies, but SiteMeter does not. SiteMeter uses IP addresses to track visits. This has its own error factors, of course. People who browse using AOL may have their IP address change frequently, and may appear to be more than one person. People who are ultra security conscious will browse using anonymous proxies which could skew readership up (if they switch proxies) or down (if multiple people share the same proxy). Finally, SiteMeter uses JavaScript to log viewers, so people who do not have JavaScript enabled don’t get counted at all.

Bottom line: SiteMeter may overcount or undercount visits. Regardless, I still say that using the number of visits as a proxy for the number of visitors is very sketchy.

Update VIII: More information here about how hard it is to track unique visitors to a site based on IP addresses:

Some ISPs (Internet Service Providers) will change someone’s IP address in the middle of a visit or even on each server page requested. On problem that all “log” based tracking systems have is they can’t tell when a visitor’s IP address changes in the middle of a visit. A change in the IP address will cause the log based tracker it think the page views from the new IP address are from another visitor and that’ll cause it to overstate the number of visitors. AOL.COM is one of the big ISPs where their user’s IP address address is constantly changing as they browse your site.

Posted in: Uncategorized

Cancel Culture Hypocrites on Left and Right

November 20, 2019 06:55 AM by Michelle Malkin

Three cheers for refugee reduction!

November 6, 2019 07:31 AM by Michelle Malkin

Impeach Amnesty Ana: TV’s Foulest Open Borders Windbag

September 25, 2019 05:28 AM by Michelle Malkin

America Takes an Antifa Beating

July 3, 2019 07:40 AM by Michelle Malkin

Malkin’s Handy History of Fake Noose

February 20, 2019 06:12 AM by Michelle Malkin

Categories: Uncategorized