From Deseret News archives:

Experts say polls don't deserve a black eye

Published: Saturday, Nov. 6, 2004 11:50 p.m. MST
 |  E-MAIL | PRINT | FONT + - 
PROVO — Exit polling received another black eye this past week when some early data appeared to predict a landslide victory for John Kerry. But the bad rap is bogus, say two well-respected Utah pollsters.

"When you want a poll to tell you who wins a race, you have to remember that a poll is an estimate with a margin of error," said Brigham Young University political science professor Kelly Patterson, who helps run the KBYU/Utah Colleges Exit Poll. "In those very close races, you could get it wrong and still be within your margin of error."

The science took a massive hit in 2000 when exit polls showing Al Gore winning Florida turned out wrong. The problem was compounded in 2002 when the national exit poll computer system crashed.

On Tuesday, moods flipped and flopped as leaked exit poll data surfaced on the Internet and showed Kerry ahead — only to be proved wrong when the votes were actually counted. The drum beat of criticism picked up quickly.

"Somebody should reassess exit polling," pundit Tucker Carlson railed on CNN's "Crossfire." "It's useless."

"The exit polls got it flat wrong," Charles Gibson said Wednesday morning on ABC's "Good Morning America."

It wasn't that the information was wrong, it was more a case of it being mishandled, according to Patterson and professional pollster Dan Jones.

Story continues below
"Many times the early leaks are wrong, that's all I can tell you," said Jones, who regularly conducts polls for the Deseret Morning News. "And that's what happened (nationally)."

Patterson agreed. "Once the data are all called in, and once they've had the opportunity to weight the sample properly, those National Election Pool polls are quite accurate," he said.

Five networks — NBC, ABC, CBS, Fox News and CNN — combined with the Associated Press and other news organizations to form the National Election Pool. They also added the expertise of two polling groups, Edison Media Research and Mitofsky International.

Researchers randomly selected 13,047 voters Tuesday as they exited the polls. Snapshots of the results were reported to the news organizations several times during the day.

The problem began when snippets of national exit poll results, meant only for analysts at subscribing news organizations, were leaked Tuesday to Internet bloggers. The Web logs published the incomplete information, some with disclaimers.

Comments

You can be the first to comment on this story.

Image
Deseret Morning News graphic

previousnext

Latest comments

Nate: Scout Eater is not where the fatality occurred, and no, the passage is...

Deseret News Ms. Volleyball 2009

Hey Lori, Congratulations! We are so happy for you! We loved getting to...

kearns is terrible i watched them play granger and it was disgraceful to...

I have never done or said anything wrong. I am perfect. Obviously Max Hall...

Witness: Mitchell wanted attention

What the stake president did back in the late 1970s is one of the unclear...

Colorado, Thank you for the answer. I agree with all you say. I was just...

You are all losers. Go get a life. Help a neighbor in need. Serve the poor....

TCU dominates all-MWC honors

I'm a Utah fan and have never been "intoxicated". Stereotype?

TCU dominates all-MWC honors

TCU was a great team to watch this year. Andy Dalton is by far a better,...

I have to agree with "OC" reader. I know this is an "LDS Times" article, but...

Advertisements