Could Google predict the next “not Romney” candidate? Maybe…but what about Ron Paul?

"Mu-ah ha, ha! I am the 'not Romney!'"

Could Google predict who’s next in the “not-Romney” race for the nomination?

Maybe, suggest political scientists. Says Erik Voeten at The Monkey Cage:

One of the driving forces behind the creation of Google Insights was the observation that Google searches can predict flu epidemics more quickly than other types of observations. Shauna Reilly, Sean Richey, and Benjamin Taylor have a forthcoming article (nongated), which suggests that political behavior too can be predicted from search terms. In particular they find that the level of Google searches on ballot questions taken one week before Election day correlates with actual participation on those ballot measures.

OK, so what if we apply this to the primary race? If we were to run searches through Google Insight, who might we see rising to be the next “not Romney?”

First, let’s look at Gingrich, the current “not Romney” candidate. I’ve included a screen shot of the search so that you can see what I did. Later images may feature just the graph, unless otherwise necessary.

Click on graph to enlarge.

Here, you see two major bumps in the searches for “Newt Gingrich.”  The first was back in May, with a slight trailing bump, right after his campaign imploded. You’ll recall, he went on a cruise to Greece and much of his campaign abandoned ship. From there, it’s fairly ho-hum until mid-October, when we see a gradual incline, increasing dramatically at the beginning of November.

Cain is represented by the red line, while Gingrich is in blue. Click on graph to enlarge.

This might be seen as result of  a combination of solid debate performances on October 18, November 5, 9 and 12, and Cain’s sexual harassment scandals as support moved from Cain to Gingrich. You will note that the increase seems much more dramatic than it actually is because the time frame is a full year. In reality, Google Insight doesn’t show any more support than 27% at its peak, and it’s actually decreased slightly since. However, the decrease seems to carry across all the candidates, probably a result of a lull between debates.

That’s history, the past. So Gingrich is up, and Cain’s bubble has popped (as has Bachmann’s, Perry’s…somewhere in here is a good argument for campaign season and it’s ability to winnow out bad candidates, but I’ll save that for another day). What about predictive power? What does Google Insight show us about other candidates and who will be the next “not Romney” candidate?

First, let’s look at some of the major candidates over the last twelve months so as to create a comparison. I’ve used Newt Gingrich (current “not Romney”), Herman Cain, Jon Huntsman, Rick Perry, and Mitt Romney (Google Insight limits search terms to five, so I’ve cut out Michelle Bachmann, Rick Santorum, and Ron Paul…not because I’m trying to ignore him. We’ll just come back to him later). With these five, then, we get this:

Click on graph to enlarge.

Admittedly, it’s a longer period than necessary, but it makes the surges look slightly more dramatic. For comparison, here is the Real Clear Politics graph of public opinion polls covering roughly a similar period:

Click on graph to enlarge.

What I see is, first, Perry, followed by Cain, and now Gingrich, peaking as interest in their candidacy rises. Each, except for Gingrich, probably because it’s too soon to tell, drops off dramatically. I think if we added Bachmann or Pawlenty (remember him?) we would see a similar rise and then dive, likely back in July and August.

Romney, in contrast, and Huntsman, to a lesser degree, have stayed relatively level, with slight dips and climbs. Actually, Huntsman’s biggest climb is currently happening. It’s slight, but his strong foreign policy performance at the AEI debate got him some attention. Another couple of performances like that, and he could earn his own spike. (HA! Right…)

Since we’re wondering who the next “not Romney” is going to be, maybe we should look at the most recent period. Let’s cut the time period from a year to 90 days. Further, since Iowa and New Hampshire carry more weight (right now) than later states in the primary campaign, let’s just look at those two states.

And that’s when I start running into problems with the data. On such a short span of time, Herman Cain distorts the data, presumably because his name is getting more print for scandals than for 9-9-9.  I removed him to make the graph readable.  That leaves four (and I’ve added News Headlines to this one) for this chart of New Hampshire:

Click on graph to enlarge.

Newt Gingrich (left) and Mitt Romney (right)

Suddenly, with the exception of Perry’s spike after his “oops” (which I am sure he would have gladly avoided), the candidates don’t look at that far from each other. Romney, who has continued to collect endorsements at a rate and caliber to be envied (Pawlenty, Daniels, Christi and now Thune, just to name a few), is currently on top, but not really in any way to signify that he’s beginning to break through to conservatives beyond his 25% base. In contrast, Perry’s at an all-time low. Huntsman continues to keep up some growth, if only gradual, while Gingrich may actually be starting to soften…or spike.

All this while it has been over a week since any debate. Without any specific events pitting the candidates against each other–and that really only means a debate, poll, caucus, or primary–it’s hard to see candidates make ground or lose it without something sensational happening (like appearing to be drunk at an event, or a sexual harassment scandal, for examples).

How about Iowa? Where’s that graph? How about this: either Iowan’s don’t use the internet as much as New Hampshire-ites or there isn’t enough data on some of the candidates to make meaningful graph, but I can’t get the data to work on a 90 day scale. It just doesn’t have enough data on the candidates to show anything meaningful. If that changes, I’ll update this post.

But back to the point: does Google Insight tell us anything? I would venture that it might, but right now there’s not enough data. As the contests in Iowa and New Hampshire approach, I think we will see an uptick in interest. Further, another couple debates will strengthen, or weaken, candidates relative positions, giving Google more data with which to work. As yet, it does seem that interest shown by the searches does follow increase in polling. Whether it will actually predict the winner I think will depend on how close to election day we are when we look at the data.

"Why won't anyone listen to me?!" Ron Paul

Check out Google Insights. Insert some terms and have some fun. And please share your insights here.

_______________________________________

Now, as I promised: Ron Paul.

First, a national chart over the last twelve months:

Click on graph to enlarge.

I wonder what it would the chart would look like if we compared his interest here on Google with the amount of attention he gets from traditional media…

Related articles

• Is Newt Gingrich ahead of Mitt Romney in the Republican race? (100gf.wordpress.com)
• Chris Christie Slams Newt Gingrich (blogs.wsj.com)
• Neutering Newt Gingrich (jeffwinbush.com)
• Google Unveils Think Insights (seochat.com)

[The Monkey Cage] [Google Insight]