This blog is still alive, just in semi-hibernation.
When I want to write something longer than a tweet about something other than math or sci-fi, here is where I'll write it.

Tuesday, April 3, 2012

Prediction for April 3:
The Wisconsin and Maryland primaries

There are three primaries today, Wisconsin, Maryland and Washington D.C. There were no polls from D.C., so neither Nate Silver or I have a prediction from there. There were multiple fresh polls from Maryland and Wisconsin, so there is enough data for two prediction battles with Nate Silver's method. Here are the numbers, mine given first and Nate's second.

Wisconsin
========
Romney    43.5%     40.9%
Santorum    36.0%     33.1%
Paul        10.5%    13.8%
Gingrich      8.5%    12.3%
none of the above    1.5%    -0.1%

Maryland
========
 
Romney    50.0%     52.8%
Santorum    28.5%     28%

Gingrich    11.5%    10.4% 
Paul        8.5%    8.8% 
none of the above    1.5%    0.0%
There are much bigger gaps in our predictions in Wisconsin compared to Maryland, and the reason is one outlier poll at the very end from We Ask America, one of the worst polling companies around this season.  I gave some thought to ignoring that poll completely, but I didn't and Nate obviously didn't, his numbers changing dramatically from yesterday to today.

An advantage to my system - at least I think it's an advantage - is that I take both averages and medians for the candidates, and then give a weighted average of those results.  This is EXACTLY the situation my method is built to exploit, damping down the effect of a late last poll that disagrees significantly with the other fresh polls.

The proof will be in the pudding tonight's battle in Wisconsin.  The things I like this time around are being on the high side of the Santorum numbers and the low side of the Paul numbers, since the polls often underestimate Senator Rick's support and overestimate Dr. Paul's.  Moreover, I have the high side for None of the Above, which has been getting over 1% nearly everyplace and is sometimes over 2%.

To have predictions this far apart as they are in Wisconsin is a little disconcerting, and even if I lose tonight I don't want to fiddle much with the system at this late date.  I've got an 8-6 lead right now. I'll report back on the numbers when the results are finalized.

 

2 comments:

Karlacita! said...

Just checking: Your numbers are first, and Nate's are second, right?

Matty Boy said...

Yes, that is the case. I forgot to make that explicit this time. Thanks for letting me know it should be corrected.