In an earlier post, I provided some evidence that the number of outgoing links on a site is correlated with the number of incoming links. What, then, is the value of an incoming link?
To attempt to answer this question, I gathered data on the number of average daily visits for 91 sites on my blogroll which have publicly-available Sitemeter statistics. For these 91, I also obtained the number of incoming links from the TTLB Blogosphere Ecosystem (for sites that do not participate in TTLBBE, I obtained data from Technorati).
The answer, for this sample of blogs, is that an incoming link is correlated with an average of 11 additional visits per day (this result is highly statistically significant). There you have it.
Back in February, CGHill looked at Blogstreet data and formulated a metric he called Blog Overachievement Factor calculated for a given site as Blogstreet's Blog Rank divided by its Blog Importance Quotient. Dr. Weevil recently posted on the same metric, calling sites that are linked by relatively more important bloggers "bloggers' bloggers". (I think they might better be called "'important' bloggers' bloggers".)
However, stealing Dr. W's terminology, these new data may allow one to better identify "bloggers' bloggers" and "readers' bloggers".
Of the readers' bloggers, the foremost is famed puppy blender and freak of bloggy nature Glenn Reynolds who receives about 57,000 more daily visits than would be predicted by the number of links to his site.
Other sites with much greater readership than might be expected given the number of incoming links include: The Volokh Conspiracy, Daniel W. Drezner, Rachel Lucas, ScrappleFace, Dave Barry's Blog, L.T. Smash and The Command Post.
On the other hand, the bloggers' bloggers, with relatively high link-to-readership ratios, include: Silflay Hraka, Sine Qua Non Pundit, Eject! Eject! Eject!, Ken Layne, a small victory, Amish Tech Support and VodkaPundit.
(For what it's worth, this site receives about 23 fewer daily visits than would be predicted given the number of incoming links.)
If anyone else sends me their stats, I'll gladly add them to the analysis.
Update: Professor Volokh kindly and gently points out that my careless choice of words implied causation not in evidence:
Interesting correllation, but which way does the causation go? You say "The answer, for this sample of blogs, is that an incoming link is worth an average of 11 addition[]al visits per day (this result is highly statistically significant)," and the "worth" suggests that incoming links cause the visits. But perhaps the visits cause the links -- the more people visit a site, the more likely they are to link to it. Or maybe a third cause (e.g., how attractive the site is to people) causes both incoming links and visits, no?
(New Poll Shows Correlation Is Causation.)
Hence, I've replaced "worth" in the third paragraph with "correlated with". Testing causation is a project for another day.
Posted by oscarjr at May 18, 2003 04:56 PM | TrackBackJay,
Quite interesting; I'll link this to OTB momentarily. I think you may want to make a distinction in "link" that will almost certainly improve your correlation: permanent links (i.e., blogrolling of the site itself) vs. vanishing links (a mention of a specific post that disappears into the archives after a few days).
My thesis is that permanent links are generally much more valuable than post links--with the notable exception of Instalanches.
This will almost certainly explain a lot of the variation you see. For example, almost everyone links to InstaPundit and, during the war at least, Command Post. They visit those sites a lot but they don't link that often.
Posted by: James Joyner at 09:43 AMJames:
Thanks for your comments and the kind link. It would have been nice to distinguish between blogroll links and post links, but none of the data sources easily allow such distinctions. I suppose I could have done so using Technorati's data, but it would have been even more laborious.
I would like to test your thesis, though. My own experience, for what it's worth, has been the opposite: most of my admittedly few visits have been via the occasional IMAOlanche, Hrakalanche, Michelelanche; etc. But that may not be indicative.
Note, also, that the data indicate that InstaPundit and Command Post are relatively underlinked given their number of daily visitors.
Thanks again.
Maripat:
If OneStat's statistics are directly comparable to SiteMeter's, you and Lori are also bloggers' bloggers.
I'm wondering if some of this effect may be due to a blog's age. Ah, more research...
OJr.
Posted by: Oscar Jr. at 11:21 PMComing from the "older than dirt" division of the blogeoisie, I think most of my traffic is motivated by inertia: "Aw, hell, we always go there, let's see what he's got."
Mr. Hill:
I hope to test that theory soon (send me your stats!).
Speaking for myself, I visit for, among other reasons, the clever post-titles.
OJr.
Posted by: Oscar Jr. at 08:59 AMDo you have a mechanism for capturing SiteMeter data, as done by TTLB?
Posted by: Douglas at 12:51 AM