The other day I posted about a remarkable surge of people using Google to search for ISIS at a rate that dwarfs the previous 11 years of searches for other terrorist organizations. I discovered this using Google Trends. The visual representation of people’s lopsided interest in ISIS is below:

When discussing the possible cause of this phenomenon, fellow blogger siriusbizinus commented:

It would be most interesting to find the source of the volume of searches. I know that there are some other atypical things about ISIS, like westerners trying to join its ranks. Could this also be a product of their media outreach? I mean, despite the fact that they publicly behead people and commit other atrocities, you have people trying to join that organization.

I responded by amending my post to reflect an almost identical surge in Google searches for “join ISIS,” which appears to support this idea. This is shown below:

Initially I noted how similar the peaks and valleys are between “ISIS” (represented by green in the first image) and “join ISIS” (represented by blue in the second image). A side-by-side comparison should lead anyone to draw similar conclusions. But then I remembered that little administrative headache in Google Trends.

Rather than show frequency (or hard numbers), Google Trends shows us a value on a linear scale, from 0 to 100. No score can exceed 100, meaning that a comparison between the two becomes problematic. If the frequency of searches for ISIS is sufficiently greater than the frequency of searches for “join ISIS” then the y-axis will be skewed in a manner sufficient to make “join ISIS” no greater than 0. Indeed, this is exactly what happened when I tested this out today:

Because interest in ISIS is so great, it is impossible to compare them along a linear scale. For example, even if searches for “join ISIS” were at an actual frequency of 10,000 in any given day, merely searching for “ISIS” at a rate of 1,000,000 during any other day would make the results for “join ISIS” register as 0. Any value of one result 100 times greater than even the highest value of the other result would nullify our ability to compare the two results.

In other words, I can find no evidence using the method above to support a hypothesis that increased interest in ISIS is partly due to increased interest in joining ISIS. (The independent and dependent variables can also be swapped, and the effect would still be nil.) Of course, this is not exactly what siriusbizinus meant.

Siriusbizinus appears to suggest that the effects of ISIS’s public recruiting campaign draws attention to the organization in ways Hezbollah, Boko Haram, and others haven’t done. This leads to increased Google searches for the organization. Reversed, siriusbiznus also appears to suggest the fact that people want to join ISIS encourages others to research the group. Either hypothesis appears to be reasonable.

[NOTE: None of this addresses the public beheadings part of the comment.]

Of course all of this is meaningless. Why people are searching for ISIS on Google has no discernible predictive qualities. I might be wrong, but I can’t think of anything significant we can learn from understanding why people are interested in the group. To me this is merely a very interesting phenomenon—something to talk about over drinks.