- The Washington Times - Tuesday, September 25, 2012

As President Obama and congressional Democrats have surged in polls, Republicans are increasingly arguing that the surveys are skewed — and some are going so far as to accuse pollsters of purposely trying to demoralize GOP voters.

Pollsters say there is no attempt at bias and that they are simply sampling people who answer their phones. But that has led to several high-profile polls that have shown Democrats substantially outnumbering Republicans, even though they usually hold only a slight lead on Election Day.

Polling firms downplay the effect of an imbalance, and many differ on whether to cancel it out by giving added weight to GOP responses, but they all acknowledge that they have a difficult job bound to receive criticism.

“There are judgments that have to be made regarding how you handle the data,” said John Zogby, senior analyst for JZ Analytics, which conducts polling for clients including The Washington Times. “We’re all taking pulses right now. If you look at them, we’re all pretty close together.”

Polls over the past week have shown Mr. Obama leading Republican challenger Mitt Romney by 1 percentage point to 8 percentage points in the popular vote.

Polls’ random samplings are designed at the outset to yield an accurate representation of voter demographics by party, gender, ethnicity and other categories, but that is easier said than done, especially when surveys usually sample fewer than 1,500 people.

The question for pollsters then becomes whether to tweak the data to give added weight to responses from groups that often go underrepresented in polls — such as Republicans, young voters and minorities — to come up with a sampling that better represents the electorate.

Charles Franklin, who runs the Marquette University Law School poll, which is tracking the presidential and U.S. Senate races in Wisconsin, said he adjusts weights for gender, age and ethnic balance but not for political party.

He argued that an increase in participation by a certain party can be an indicator that it is more energized than the other side.

Democrats historically hold a turnout edge in presidential elections, but their advantage often fluctuates.

In 2008, 39 percent of presidential voters were Democrats compared with 32 percent who were Republicans, according to exit polling.

The parties each represented 37 percent of the electorate in 2004, which was the Republicans’ largest share in decades.

“Forcing there to be a certain percentage of your electorate denies that change,” Mr. Franklin said. “There’s a legitimate debate as to which way to go, and I don’t think it is by any means a settled statistical question.”

But Scott Rasmussen, founder of Rasmussen Reports, which has Mr. Obama with a 1-point lead over Mr. Romney, said, “We absolutely know that party affiliation is the best single indicator of how someone is going to vote.”

In addition to cries of party bias, pollsters are also dealing with issues brought on by technology and changing demographics.

Many polls continue to collect their data from calls to land lines rather than cellphones, which could increase responses from older voters who rely on their home phones and minimize participation among younger voters.

Mr. Rasmussen predicts that polling might move entirely online in five to 10 years.

The different methods and varieties of approaches among the various pollsters mean poll watchers need to take individual surveys with a grain of salt.

“Look across all of the polling,” Mr. Franklin said. “Don’t just try to believe that one poll is always right or that one is always the best. If you see all of the polls pointing in the same direction, that’s certainly more reliable.”

Copyright © 2018 The Washington Times, LLC. Click here for reprint permission.

The Washington Times Comment Policy

The Washington Times welcomes your comments on Spot.im, our third-party provider. Please read our Comment Policy before commenting.


Click to Read More and View Comments

Click to Hide