<iframe width="640" height="480" src="http://www.youtube.com/embed/G-HALE2InRo" frameborder="0" allowfullscreen></iframe>
So a study that looks only at historical data that is not always pertinent to the current election says Romney will win. That's like saying a black guy can't become President because every President before him was white. The study falls a bit short. History is great and all, but it does not always repeat and the 2008 election is proof of that.
Not to say that history has to repeat, but this thing has been pretty darn accurate in the past. I'm just throwing this out there because frankly...I like it and what it says in support of the chances of my candidate.
Yeah, but like any good scientist would tell you, just because something has shown consistency of it working does not mean it will always work. I find it to be suspect and here is why: the algorithm has been tailored to fit the results of all the previous elections. By doing so, of course the algorithm is going to be darn good, it was built that way! However, when it comes to actually extrapolating a point outside of your current data set it may not fit. It's like this, give me a series of points that follow a supposed logical rhyme or reason. I can easily fit an nth order polynomial to it and say I've extrapolated the function behind your points. Does that mean the next point will fall on said polynomial? Probably not. In all the data I have taken as a scientist it is rarely the case where a function designed to fit previous data actually predicts future data unless the connections/mechanism behind the function is sound. Because this mechanism seems to be based on more definite numbers and not so much on the pulse of the current election I feel that the mechanism is lacking. With that said, if it predicted Obama was going to win, I might have looked at it and said 'meh instead of investigating the methods...
"You know those ratings systems are flawed. They don’t take into account houses that have more than two television sets and other things of that nature. Hey, Burgundy. You know those sample audiences aren’t big enough! Stop hiding behind those phony numbers, Burgundy! I’m coming after you! I hate you, Ron Burgundy. I hate you! You can’t say one word? Even the guy who can’t think says something!"
The Halloween mask indicator has accurately predicted the last many elections too. As has the Redskins indicator, since 1940: http://en.wikipedia.org/wiki/Redskins_Rule
I'm no scientist or expert in a related field, but the results have not been met with criticism from academic peers. Gonna go with the PhDs over some posters in D&D.
Yet again, you aren't a scientist. No criticism does not always mean it comes at rave reviews. A lot of times, low criticism means it's poor research to begin with and isn't worth someone's time to talk about it. You can take that from a poster and from someone who is getting his PhD.
In defense of bigbird CU study, economy and public opinion of the govt does tend to be pretty crucial to who wins the election. And many have pointed out how weird this race is in terms of going against conventional wisdom. That said, every election is different and quite frankly, 7 straight correct guesses don't really mean much. Especially since many results have been landslides. It would be like correctly guessing that the Heat has a great chance at beating the Bobcats. Not exactly impressive
Halloween mask indicator doesn't have a long enough trend. Redskins indicator failed in 2004, regardless of the rule change.
Fair enough - but that still gives it a 94.7% success rate, which is more definitive than this CU study's 77% prediction.
And the Redskin rule will hold true again this year..... The Sunday before election day, the Redskins play Carolina at home. Easy win for the Redskins, easy win for President Obama..... :grin:
