Why waste mental energy on “what ifs”. Decisions are made when they have to me made. Let’s get behind Fred now so he wins the nomination and we don’t have to worry about what we’ll do if does not win.
Seriously, if you want something, work to make it happen. It’s not too late and now is the time.
If you want to help, you can make calls into Iowa from your own home. [see fred08.com]
Send whatever money you can. Campaigns are expensive.
Talk up Fred to everyone you know.
>>
Below are the results of a three-day poll in the state of Iowa. Results are based on telephone interviews with 600 likely Republican cacus goers and 600 likely Democratic cacus goers, aged 18+, and conducted December 16-18, 2007. The margin of sampling error is ±4.5 percentage points for each party.
>>
This suggests yesterday’s ARG poll, taken the same days with the same sample size, but quoting a bizarre 8.7% “highly likely” GOP voters with grossly different results was an outlier.
We are starting to see the same questions about polling as existed in 2004. How do two polls of the same days and the same criteria yield such divergent results?
We had partial answers in 2004. Different pollsters use the different criteria about partisan mixture. Some allow it to vary with a random sample. Some aim at a fixed mixture.
I am coming to believe that the fluid nature of American relocation patterns means that making 600 random calls and allowing the Democrat/GOP/Ind mixture to be whatever it turns out to be is wrong. It is too dependent on zip code and calling area prefixes yielding the same nature of neighborhoods as existed 4 years prior.
A better approach is Rasmussen’s, where he has a targetted mix derived from the last national election and he keeps making phone calls until he has the target mix.
For primaries, even the variable mix pollsters are going some distance towards “targetted mix” in that they know what calling areas have GOP voters and they call only those. The problem for primary polling is the “highly likely” criteria.
The vast majority of pollsters are sampling that via self selection. “Did you vote in the last primary?” “Did you participate in the last caucuses?” And because those questions would exclude newly of age voters, or newly moved into that region, sometimes the question is just “On a scale of 1-10 with 10 being utterly certain, how likely are you to appear at the caucus?”