I've been using Pollster's averages for my electoral map, in which I'm attempting to paint a clearer picture of which candidate would be the stronger nominee against McCain.
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=132&topic_id=5544447&mesg_id=5544447But I just hit a glitch today. I've been taking their numbers at face value, which sped things up but now I see it was a mistake. I noticed this when suddenly, in one day's time, Obama went from having a 2.8% lead in New Mexico to a 5.8%
deficit vs McCain.
I thought to myself, what?? One single poll did this to him?
I like Pollster's graphs. They're cool, and the more data points they have the more likely it is that they're right. But even when there are fewer polls, they should be able to do better than what they did in the above graph. It's almost like someone's just connecting dots wherever they feel like it. I mean, I guess the pink dot below the final two blue dots is completely irrelevant.
:wtf:
In reality, the graph should look more like this.
It's not THAT big a difference when you're just looking at graphs (although it IS a clear difference), but when I'm inputting numbers into a big electoral chart it's ESPECIALLY a big difference. It's the difference between Obama being ahead by one percentage point, or behind by 5.8%!
(The above graph was created using a rolling average of the three most recent polls for each data point except, of course, the initial two data points. That's not the only way I could do it, but it serves as a good enough illustration of the shortcomings of Pollster.)
So what I'm basically saying is, I have to start over and do my own analysis of the polls in each state to see what the true percentage of support is for each candidate.
Dang it.