Most polls are based on 'likely voters'

[Spinzonner]

Given that there seems to be a lot of people getting motivated to vote against Bush, how accurate are these methods and measures, especially if they are based upon assumptions associated with past elections ?

Are there any polls or studies addressing this issue ?

[NMDemDist2]

1. likely voters with traditional land line telephones

pretty much misses the whole 18-29 year old demo IMO

[Florida_Geek]

2. And all the new voters being added to the roles

mainly by Dems.

And the college "free" screenings of F9-11 with voter reg. after it will IMHO add maybe a million new voters to the roles.....

[chuck555]

3. No,but you have good question.

i read an article about pollsters doubting their own methods.

Caller Id

Cell phones

Answering machines/voice mail

People in general not wanting badgering calls.

These were the problems.

Food for thought. I leave it up to you.


I have thought that those that are most likely to see F911 are the least likely to vote,until now.

TIA seems to know and I see a landslide. GM.

[KharmaTrain]

4. Polls Are A Snapshot Of The Past, Not The Future

By the time a poll is conducted, gathered and analyized, the data is several days old. The best data one gathers from this type of research is where people were at at a particular date & time...and place. By putting this data with past surveys with identical or very similar methods, then you see trends from where you can make determinations about the future. One poll alone means little without being taken into a bigger picture.

Thinking Kerry's lead means victory in November, at this point is still extrememly premature, but the tracking of recent polls looks good. It seems like we've never had as many polls as before, and each one tauts a unique sampling or methodology (weighting of sampling)...we even have our own armchair pollsters here on DU. It just amplifies that polls and research are tools that show you were you were, not where you're going.

Likely voters are your "best bet" in getting a sample of who will show up on election day as opposed to "all voting age" (where over 50% still stay home) or "registered voters" (as those numbers are generally party partisans and miss out on the fluid "temporary" voters...ones who will vote this year and may not vote again until 2012).

While it's important to know who is sampled, I still haven't really seen the questions these people are being sampled with. I participate in Zogby's online polling and the qusetions this week (variable testers for certain) asked if I was on a "low carb" diet...to get a better profile of the electorate. But you won't see those questions, just the numbers...and that's all that count.