Monday, January 24, 2005

The A-Bomb vs. the I-Bomb, or Why Time Series Frequency Matters

by Tom Bozzo

A funny thing happened to the traffic at this blog on the way back from Blogger dinner, which broke up just before midnight Saturday. Need I mentioned that the attendees vary widely in intensity of traffic-monitoring obsession? I'm not even sure I'm the most obsessed, though I shall give it a go.

Anyway, here are the Sunday hourly visits to Marginal Utility through 10 P.M., courtesy of Site Meter.

Traffic has been running around 20-25 visits a day, and as recently as last month the traffic patern above would not have been at all unusual for the daily take. So what happened?

A steady stream of curiosity-seekers arrived from Jeremy's and Tonya's simul-blogging throughout the dinner and subsequent festivities. Then the A-bomb dropped, as Ann weighed in with links to the other attendees' simulblogging posts. Since her traffic is roughly 200 times larger than mine, and unknown multiples of everyone else's, even lowish click-through rates have big effects on my traffic.

Then, something else happened in the fifteenth hour, betwen 3 and 4 P.M. (CST). Some sleuthing reveals that Glenn Reynolds linked Ann's blogger dinner post with an entry stamped 4:21 P.M (EST). So some of the incremental InstaPundit traffic is, even more remarkably, not satisfied with Ann's prose and clever Nina-as-blur photography and drives an additional traffic spike here. Go figure.

Could the Althouse and Instapundit effects have been identified from lower-frequency data? No. The next lower frequency Site Meter offers is daily; the graph is reproduced below.

Next week's topic: mean reversion.
Where can I to learn abt it in detail?
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?