Wed Oct 24, 2012, 03:54 PM
Texas Lawyer (350 posts)
Question about polling - why are we doing so much better on poll averages when we exclude automated
phone polls from the averages?
This seems significant, but it doesn't get much analysis.
The Huffington Post's Pollster (http://www.huffingtonpost.com/news/pollster/) has us narrowly ahead in
New Hampshire (http://elections.huffingtonpost.com/pollster/2012-new-hampshire-president-romney-vs-obama?gem)
Ohio (http://elections.huffingtonpost.com/pollster/2012-ohio-president-romney-vs-obama?gem), and
while narrowly behind in
Florida (http://elections.huffingtonpost.com/pollster/2012-florida-president-romney-vs-obama?gec), and
North Carolina (http://elections.huffingtonpost.com/pollster/2012-north-carolina-president-romney-vs-obama?gec).
Pollster also allows you to make your own aggregate polling graphs.
If you re-do all the polls excluding those polls which are based exclusively on automated phone calls, we do consistently better across the board in the various different state-by-state races (the same holds true for the national horserace polling). For example, in Florida and North Carolina, by dropping the automated phone call polls, we are either tied or slightly ahead (depending on how much smoothing you apply to the graph line's connection of the data points).
Why is this?
It does not appear to be a deliberately ideological factor (right-leaning Rasmussen and left-leaning PPP both follow the automated-phone-call methodology), but it is consistent across the map (in states where have have bigger leads, our lead increases when you drop out the automated phone call polls and in states where we are further behind, these is some narrowing when those polls are omitted from the graph).
If there is a fundamental flaw in the automated-phone-call methodology, then the election may not be as close as is widely reported. On the other hand, if there is a fundamental flaw in the other polling (and only the automated phone call polls are getting it right), the race might be tighter than we imagine.
3 replies, 683 views
Always highlight: 10 newest replies | Replies posted after I mark a forum
Replies to this discussion thread
Question about polling - why are we doing so much better on poll averages when we exclude automated (Original post)
|Texas Lawyer||Oct 2012||OP|
Response to Texas Lawyer (Original post)
Wed Oct 24, 2012, 04:14 PM
JDPriestly (48,543 posts)
1. I hate automated polls and robocalls.
I hang up the minute they call me.
Maybe a lot of other Democrats feel that way. I like speaking to a real person, not a machine.
Response to naviman (Reply #2)
Wed Oct 24, 2012, 04:27 PM
DonViejo (14,621 posts)
3. Yesterday, someone posted the info that a call to a cell phone
must be "live;" it's against Fed Regs to do otherwise. That is, NOT a recording or a robo-call. Alot of polling firms rely on automated calls and cannot use cell phone owners in their research for that reason.