In political polling, how you call is becoming just as important who you call.

National estimates put cellphone-only households at about one-third of the country. Here in Hawaii, an estimated 35 percent to 50 percent use cellphones exclusively.

The importance is not whether you are packing an iPhone or a push-button landline; the importance is if pollsters can reach you.

Increasingly, pollsters are finding the world divided between those with a cellphone and those with a landline — and they are also discovering that the results are different. Polls that don’t include cellphones report somewhat different results.

Nate Silver, who writes the well-respected blog FiveThirtyEight, last week cautioned that polls that don’t include cellphones are showing a much tighter race for president.

"In the polls that use an automated dialing method ("robopolls") or which exclude cellphones, Mr. Obama’s bounce has been much harder to discern, and the race looks considerably closer," Silver wrote.

The "robopoll" uses a computer to randomly dial phone numbers of registered voters, but by law cannot randomly dial cellphones. The robopoll is generally considered to be less expensive and there are enough concerns about its validity that news organizations such as The Associated Press tell reporters: "Avoid polls in which computers conduct telephone interviews using a recorded voice."

The Honolulu Star-Advertiser uses Ward Research for its political polling and those polls are done with live people calling voters both with landlines and those with just cellphones. Another major Honolulu polling firm, QMark Research, does polling for political candidates and issue groups. It also includes cellphones in its research.

But, a local online news site, Civil Beat, uses robopolls that do not include cellphones.

"The ‘likely voter’ model here is about the oldest in the country, our pollsters tell me, and remains so heavily landline accessible that cellphones (for now) are not a big deal in Hawaii," Patti Epler, the editor of Civil Beat wrote in an email response to an emailed question on polling methodology.

Becki Ward, president of Ward Research, cautions that no political poll can or is trying to predict the outcome of an election, but polls can show you what is happening in an election.

Ward says pollsters get more information if they include cellphone users.

"I definitely believe it is important; we have seen differences in voting intentions of cellphone samples.

"If it were a tight race, it would make a difference," Ward said in an interview.

Ward Research compared the differences between cellphone and landline respondents’ results of the last Hawaii Poll, taken for this paper and Hawaii News Now.

There are a lot of major differences between the two groups.

First, for those between 25 and 34 years old, just 5 percent have only a landline and 18 percent have only a cellphone. The numbers reverse for those more than 65 years old, with 33 percent of seniors having only a landline and 17 percent having just a cellphone.

According to ethnicity, Filipino voters are more likely to have a landline and Hawaiians much more likely to have a cellphone.

Also, men are much more likely than women to have cellphone exclusively.

Ward explains that it is possible to "weigh" a poll to correct for population groups that your poll results may not include, such as young people. But if you are including young people reached by landline, they are most likely young people living at home, as opposed to independent young voters.

"The problem is that you’re taking a skewed sample and weighting up the problem," Ward said.

Cellphone results will not make or break a poll, but repeatedly ignoring a specific portion of the population because you can’t dial their cellphone is bound to cause confusion.

———

Richard Borreca writes on politics on Sundays, Tuesdays and Fridays. Reach him at rborreca@staradvertiser.com.