SurveyUSA loses his little credibility . But it's not only in NC, in every poll, SUSA underperforming Obama. in WA for exemple, they have Obama leading only by 4. It's not possible, Seattle decide every election in Washington state and Seattle is Obama county.
SUSA was good for FInal Primary POlls. That´s it. Their wild swings by Party ID make their pre-election polls useless. However, I think Rasmussen is pretty accurate.
Unfortunately Rasmussen is too wedded to historical turnout ratios in determining how it weighs its polls by party id; this can be good when the shares of the electorate between the parties are fairly stable but this year Dems are making huge strides (and Reps in contrast are having trouble just to tread water) in a number of battleground states, be it FLA, NV, IA, CO, PA, and yes NC.
" Unfortunately Rasmussen is too wedded to historical turnout ratios in determining how it weighs its polls by party id;"
As far as I know, they take their Daily Tracking samples to create a big PartyID poll updated monthly (with a sampling size of 30,000 per month or so). That should be enough to get sufficient data, at least for the larger states.
rasmus you don't understand. after rasmussen gets its results back from the field it weighs or calibrates them using a preset number of how much of the electorate should come from each party. Thus, if rasmussen, as it does, look to historical turnout in the states it is polling this weighing can miss or undercount a candidate's numbers if the candidate's party is showing a larger than normal surge in voter registration. That is EXACTLY what has happened in ALL the states I mentioned. Accordingly, I stand by my earlier statement --- Rasmussen is undercounting Obama's support this year because its weighing system is too wedded to past turnout in the states it is polling.
Are we paying too much attention to massive turnouts? If historical number of people are registered, the random sample in surveys should show the enormous lead for Obama? We are suddenly in the business of downgrading all the polls and we are assuming this massive turnout will offset all our calculations.
More over, more people are giving their cell phone numbers as the primary means of contact. So, the 2004 argument of cell phone Vs Home phone needs to be out of the window.
SurveyUSA loses his little credibility .
ReplyDeleteBut it's not only in NC, in every poll, SUSA underperforming Obama. in WA for exemple, they have Obama leading only by 4. It's not possible, Seattle decide every election in Washington state and Seattle is Obama county.
Seriously, SurveyUsa is ridiculous.
If you think their bad,rasmussen is a republican schill.They play for one team, the republicans.
ReplyDeleteSUSA was good for FInal Primary POlls. That´s it. Their wild swings by Party ID make their pre-election polls useless. However, I think Rasmussen is pretty accurate.
ReplyDeleteUnfortunately Rasmussen is too wedded to historical turnout ratios in determining how it weighs its polls by party id; this can be good when the shares of the electorate between the parties are fairly stable but this year Dems are making huge strides (and Reps in contrast are having trouble just to tread water) in a number of battleground states, be it FLA, NV, IA, CO, PA, and yes NC.
ReplyDelete" Unfortunately Rasmussen is too wedded to historical turnout ratios in determining how it weighs its polls by party id;"
ReplyDeleteAs far as I know, they take their Daily Tracking samples to create a big PartyID poll updated monthly (with a sampling size of 30,000 per month or so). That should be enough to get sufficient data, at least for the larger states.
rasmus you don't understand. after rasmussen gets its results back from the field it weighs or calibrates them using a preset number of how much of the electorate should come from each party. Thus, if rasmussen, as it does, look to historical turnout in the states it is polling this weighing can miss or undercount a candidate's numbers if the candidate's party is showing a larger than normal surge in voter registration. That is EXACTLY what has happened in ALL the states I mentioned. Accordingly, I stand by my earlier statement --- Rasmussen is undercounting Obama's support this year because its weighing system is too wedded to past turnout in the states it is polling.
ReplyDeleteFolks,
ReplyDeleteAre we paying too much attention to massive turnouts? If historical number of people are registered, the random sample in surveys should show the enormous lead for Obama? We are suddenly in the business of downgrading all the polls and we are assuming this massive turnout will offset all our calculations.
More over, more people are giving their cell phone numbers as the primary means of contact. So, the 2004 argument of cell phone Vs Home phone needs to be out of the window.