PDA

View Full Version : Poll Response Rates Under 20% - New Article




DrZ
12-10-2007, 01:34 PM
Over 80% of people that pollsters call don't pick up the phone:
http://zogby.com/Soundbites/ReadClips.dbm?ID=16193

I never realized polling had become this biased and unreliable! The response rate is far less than 20% when you include the cell phone factor. I for one will never get polled because I don't list my phone number on my voter registration, I don't answer toll-free and unknown callers, and I will likely hang up on any junk caller that does get through. And people take these polls as gospel and let the polls influence them on who to vote for and who has a "realistic" chance of being elected?

I'd like to see the following statement attached to the next major media poll:

"We surveyed a bunch of old farts who had nothing better to do than sit around home and answer our phone calls. Unfortunately, 80% of the people we wanted to survey never answered their phone. Maybe they had a life? Oh, and another 10% of the people we wanted to survey didn't even have regular phones. But here's the scientific result of what the losers we polled had to say [margin of error +=100%]..."

Cleaner44
12-10-2007, 01:41 PM
I would guess that a large percentage of those responding to the polls are the same little old ladies that get their life savings scammed away. Then we all wonder "who falls for that crap".

Ben Elliott
12-10-2007, 01:44 PM
That is why the truth will come with the polls.

Jwaksman
12-10-2007, 01:47 PM
Remember, these pollsters don't just count the first 1000 people who answer the phone. They all try to determine just how many voters will be repeat voters, how many will be new, how many will be Republicans, how many Independents, how many men, women, old people, young people, union members, etc.

So, if 50% of the people who answer the phone are little old ladies and 5% are teenagers, and they think that there will only be 5 times as many little old ladies voting as teenagers, then they will discount their results from the old ladies 50%.



Polling isn't as bad as you make it out to be. The two sides of this chasm are both wrong. The mainstream media who tout a single poll as exactly describing the situation without explaining about randomness and the margins of error are wrong (i.e. Huck supposedly being 22 points up in Iowa, which he most certainly is not - maybe 10 points). But meanwhile, people who think that polls aren't even close to reality aren't right either. Ron Paul isn't at 30% in Iowa, he just isn't. I do believe he is being undercounted in the voting, but it's not as extreme as some people make it out to be. Simply because he attracts a lot of people who tend to be unreliable voters, but who are now no longer apathetic and will vote. But since they demographically fit in the "unreliable voters", what they say is somewhat discounted by the pollsters.

RPinSEAZ
12-10-2007, 01:49 PM
Polling companies don't just take the answers of the 20% and call it good. They account for over-representation or under-representation by giving less or more weight to certain demographic's responses. They know who they're calling and are able to use that info to correct for a true representative sample.

For examples. Let's say they contact 1000 people, but they are only able to contact 20 people in the 18-22 college male group. They'll take that 20 people and correct for the actual size of their demographic by multiplying their answers by whatever it needs to be.

raystone
12-12-2007, 04:48 PM
interesting !

rollingpig
12-12-2007, 04:58 PM
According to hucklebee, it's a divine intervention

Ronin
12-12-2007, 05:13 PM
I'd like to see a poll of with voters likelyhood of voting compared to support. Kinda like this...

Out of a thousand...

Candidate Support Likelyhood of voting Votes
------------ -------- ---------------------- ------
Romney 30% 25% 300 * .25 = 75

Paul 10% 100% 100 * 1 = 100

Who's the winner now :)