As you may recall from last week, we took a look at Bill Tancer and his high level predictions when he analyzed aggregated search string data. In the fifth week of CEDO 540 it is now our turn to mine Internet data and makes heads or tails of it. Specifically, we are using Google Trends for our database and it is up to us to determine an approach to a topic or problem, gather the data and analyze it. Sounds pretty easy, ay? Well, let's just say I have a new found appreciation for Bill Tancer's line of work.
A more useful site for gleaning Web data is Google's beta version of Insights for Search. Insights for Search which can be broken-out by country, states/provinces and metropolitan areas. Bill Tancer has been able to predict the winners of Dancing with the Stars by examining how often a contestant's name has been searched for on The Internet. In Wisconsin we have a heated Senate race between incumbent Russ Feingold and newcomer Ron Johnson. I did a search within Wisconsin on both candidates and Johnson is definitely on the rise. Wait, that didn't sound right. Just so there are no misunderstandings about that statement let me rephrase by saying 'Ron Johnson has had more searches in the past few weeks than Russ Feingold.'
But does the greater number of Ron Johnson searches over Russ Feingold mean that he will triumph in November? I looked at the numbers this past Tuesday and now on Friday it appears that Russ Feingold is trending downward on the number of searches. Below is the data showing Johnson in red and Feingold in blue. I was able to generate this graph and copy the embed code into my blog. Feel free to hover over each of the data points and see the changes.
So, does the sheer volume of searches for a candidate predict the ultimate victor? Bill Tancer believed this to be true until the Stacy Keibler scenario (see post from last week) took hold. Are some of these searches unrelated to the campaign? Am I analyzing the right search strings and discounting the ones that don't matter? Perhaps people doing their own research on the candidates and what does this tell us about their feelings towards either one? The Internet is a cyber superorganism that is evolving every second - changing from moment to moment - and this alters the language people use for interaction. In this case, search strings are the interactive language and, like Early Modern English, they will continue to morph into new forms - only much faster.
Scene from 2001: A Space Odyssey |
Hi Brian!
ReplyDeleteYou've got some good questions about the validity of the candidates research and trends. Our group was trying to do an analysis on gubernatorial candidates in much the same way, and you just don't know who trust! Where in the world did Ron Johnson come from...he's just causing a late election campaign flurry of interest, isn't he!?!
Oh well! Thanks,
Mary
Well, I hope the number of searches isn't the indicator of who will win! While I'm not thrilled with Feingold, I can't see myself liking a Republican ever. I wonder if people are just trying to find out more about Johnson since he's so new? Only time will tell...what's your prediction?
ReplyDeleteDidn't Tancer also say that the closer the election gets the more the opponents supporters are searching for the other candidate???
ReplyDeleteIt will be interesting to see if the Ron Johnson search activity is just because we know less about him than his opponent. Time will soon tell the correlation. Sampling does create an element of uncertainty. It will be interesting to see if this is minimized or quantified over time.
ReplyDeleteBrian,
ReplyDeleteI think that Ron Johnson is getting more searches because he appears to have more TV advertisements. I also agree with those above that there is not a lot known about him. Russ Feingold has been a prominent senator for a long time so his positions are mostly known.
It would be cool to see who ends up winning and then compare the searches at the end.