Friday, October 19, 2012

Polling Discrepancies

Gallup General Election Polls
No doubt everyone heard the news by no of the general election poll from gallup that reads Romney by 7 among likely voters. Three things to consider about this poll...
1.) They have a very suspect track record in the past and of greater importance is their reputation for having findings away from the norm. As easy as it would be to look at this poll and think that it must surely mean Romney is winning, there is just not enough evidence in one isolated poll to draw those conclusions. One must consider this poll as an outlier and wait patiently to see if other polls start to follow there trend. My guess based upon toady's Rasmussen tracking poll is that this poll means very little.

2.) Ordinarily, likely voter polls are far more accurate then registered voter polls. However, in this case gallup has a very sketchy model or screen they employ to discern likely voters.

3.) This is merely a general election tracking poll that compiles information aver a seven day does not reflect substantial information on any given swing state and is mainly used for trending purposes. Generally these tracking polls that stretch there findings over this long of period tends to be very volatile and subject to dramatic swings. My best guess is that it will begun to retreat closer to the norm of an approximately dead even race nationally in the days to come.

4.) Those of you that love Rasmussen tracking needs to look no further then today's poll which shows it a tie. This is a two point swing overnight and can potentially be a sign that Obama is going to see a small jump from the past debate. Tomorrows poll will be the first one that is compiled entirely post debate. However, Romney did jump one point yesterday so it may be a little early to project much from this. It is a bad sign for a positive poll for a republican when Rasmussen (a reputatble poll with a slight republican bias as a whole) is 5-6 points different

Susquehanna Poll Showing Romney by 4 in Pennslyvania
Democrats step away from the cliff! Republicans stop your Romney inaugeration plans! This poll that is taken from the republican party is also an outlier that has not yet seen any verification from other polls that the race is truly dead locked in the Keystone state. Yes, other recent polls has shown some pretty dramatic movement towards Romney, but nothing like this. This poll has been fair in years past, but has also shown a 2-3 point republican lean. I know in this case that would still be tremendous news for the Romney columnm, except this is just one poll.

Using my formula...this effectively makes this about a 4 point race in Obama's favor still. Generally, I would take a state out of the safe column and move it to the Leaning Obama column when the margin hits 4 points. However, in this case I am going to withhold judgment until more data arrives. Since the over all adjusted average of polls show about a 4 1/2 point race without the outlers and gernally bad ones this adjusts this poll back to a razor thin margin for Obama. The next couple days will be crucial in making a decision here. Since polls like these are placed at the median point between the poll and the adjusted average of other polls. Any additional evidence that this is even a 2-3 point race will adjust this Susquehanna poll back towards Romney and can dramatically change the picture. Yet, the same is also true if a reputable firm shows this a 6-7 point race for Obama. I will not change the category a state is in just to make people feel good about the race.

Yesterday's Good Day for Obama
There was some fairly good news for Obama yeseterday, but when adding in these numbers it still does not alter the final projected numbers. The two NBC/WSJ/Marist polls has a couple thing against it. First, it showed outlier numbers that is not reflective by other reputable firms. Plus this poll has a pretty clear democrat bias in past elections. When reverting these numbers and plugging them into the states it was still not enough to change Wisconsin out of a toss-up that is projected to go towards Obama and Iowa is still in the leaning Obama column. The PPP democratic poll from Colorado does change things a little in that a very suspect lead for Romney became a literal tie in my projections.

No comments:

Post a Comment