In two earlier posts (here and here), I looked at the relationship between weblogs' traffic and the associated number of inbound links. Tonight, I look at the same relationship, but exclude from the analysis the top 25 most-visited weblogs in N.Z. Bear's sample with open Sitemeter statistics as of July 7, 2003. (These 25 websites account for almost exactly 2/3rd of all traffic for this sample of blogs.)
Once again, I find a highly statistically significant relationship between inbound links and traffic. However, for this sample of blogs each inbound link is correlated with an additional 2.3 visits per day, as compared to the 12 visits per day for the entire sample of blogs. This discrepancy may provide some evidence that, as Professor Volokh noted, traffic may cause links, or some other factor may cause both traffic and links.
For this smaller sample of 626 weblogs, controlling for the number of inbound links, I find that the number of months a blog has been in existence is positively correlated with the number of daily visits. Each additional month of blog-age is correlated with an additional 3.4 daily visits. This supports CGHill's suggested inertial effect.
Finally, after removing the most-visited sites and controlling for the number of inbound links and blog-age, I find that sites on BlogSpot are correlated with statistically significantly fewer daily visits (25 fewer visits per day) than sites hosted elsewhere. (The BlogSpot sites in the top 25 are: Eschaton (Atrios), Dave Barry's Blog, Howard Dean 2004, TBogg and Merde in France.)
Note that, for the full sample of blogs, these three variables "explain" 62% of the
variation in traffic; for this smaller sample, the corresponding number is only 43%.
For what it's worth...
Posted by oscarjr at July 31, 2003 09:00 PM | TrackBackInteresting. You should post the spreadsheet (assuming that's what you used).
Posted by: Kevin at 11:06 AM