Despite open contempt from party elders and a major fundraising disadvantage, Ed Case has the Democratic nomination for U.S. Senate within reach.

A snapshot taken 10 days before the election shows Case with the narrowest of leads and the momentum, but other data from The Civil Beat Poll indicates rival Mazie Hirono might have the inside track. Case’s 47-46 edge is within the margin of error of +/- 2.8 percent, making the race a statistical dead heat. Case has more support from those who decided recently, but Hirono banked more votes in the first days of early voting. Seven percent remain undecided.

Civil Beat surveyed 1,227 likely Democratic Primary voters statewide between July 31 and Aug. 2.1 Hirono led among those who said they’d already voted, the poll found, while Case had the advantage among those who said they definitely planned to vote. The winner will advance to the November general election to face likely Republican nominee Linda Lingle, who led John Carroll 84-11.2

The poll’s neck-and-neck results for the leading Democrats are consistent with two previous Civil Beat surveys taken in June and January. Those polls found a 46-46 tie and 41-39 Case lead, respectively. But all three of the Civil Beat results3 are far different from other independent polling of the race.

Just weeks ago, the Honolulu Star-Advertiser- and Hawaii News Now-sponsored Hawaii Poll had Hirono up by a large margin, 55 percent to 37 percent. In February, the Hawaii Poll pegged the race at 56-36 in Hirono’s favor.

Why the Difference?

There are key differences between The Civil Beat Poll and the Hawaii Poll that could explain some the divergent results:

Methodology: Civil Beat uses interactive voice response technology, also known as IVR, touch-tone polling or, to critics, “robo-polling.” The Star-Advertiser uses live telephone interviews. IVR has some downsides — you don’t know who’s actually answering the phone, for example — but the technology allows for faster surveys of larger samples closer to election day. More on that in a minute.

Neighbor Islands: The Star-Advertiser/Hawaii News Now poll, according to its cross-tabs, included 408 Oahu voters and 199 neighbor island voters for a 67-33 ratio. The Civil Beat Poll had the split at 70-30. This might sound insignificant, but the geographic strengths and weaknesses of the two candidates make it significant. Case led 50-42 on Oahu in The Civil Beat Poll, while Hirono led 59-41 on the neighbor islands that make up much of the 2nd Congressional District both Case and Hirono have represented in Congress. The Hawaii Poll also showed Hirono significantly stronger on the neighbor islands. In the 2010 primary election, 71 percent of votes cast came from Oahu. Will the Honolulu mayoral race create an enthusiasm gap?

Survey Timing: Every survey is a snapshot in time. The Star-Advertiser’s poll was conducted from July 12 to July 21, while Civil Beat’s calls were made July 31 to Aug. 2. That two- to three-week difference is significant because a full 24 percent of respondents in The Civil Beat Poll said they chose their candidate in the two weeks before they answered the survey. Another 18 percent said they decided between two weeks and one month before the poll. Case had the momentum with late deciders, leading 61-39 among that first group and 55-45 among the second group. Hirono led 54-46 among the 57 percent of voters who decided more than a month ago.

Sample Size: The Star-Advertiser/Hawaii News Now poll surveyed 606 people over 10 nights for a margin of error of +/- 4.0 percent. Civil Beat surveyed more than twice as many people over just three nights for a margin of error of +/- 2.8 percent. The margins of error are both reasonable, but Civil Beat’s larger sample means we can do cross-tabulation analysis on subsets of the data — things like county-by-county support or a when-did-you-decide question — without introducing high levels of uncertainty.

Age: The Star-Advertiser/Hawaii News Now poll had 51 percent of its sample aged 55 or older. Civil Beat had 81 percent of its sample aged 50 or older. This is a large difference, and would have a large impact on the outcome because Case outperforms Hirono among older voters (49 percent to 44 percent among both the 50-64 and 65-plus age brackets) while she beats him among younger voters (54-40 percent among those in their 40s, 41-34 percent among those in their 30s and 55-38 percent among those between 18 and 29). Primary election voters tend to be older than the general population, registered voters and general election voters. If younger voters turn out Saturday, it would help Hirono.

Republicans and Independents: This might be the most important variable, and it’s a tough one to predict. In the Star-Advertiser/Hawaii News Now poll of likely Democratic primary voters, 72 percent usually vote Democratic versus 9 percent Republican and 18 percent independent. The split in The Civil Beat Poll is 59 percent Democrats, 8 percent Republicans and 27 percent independents.

That’s potentially a huge difference because Hirono dominates among Democrats, 58-37, while Case dominates among the other groups, 74-18 among Republicans and 63-30 among independents. If Linda Lingle‘s advertising campaign convinces voters to stick with her in the Republican primary, that would help Hirono, who lost to Lingle in the race for governor in 2002. If voters on the fence (or on the other side of it) pull a Democratic Party ballot, that would help Case.

The difference in the split can be traced, at least in part, to the wording of what’s known as the “screen” question in The Civil Beat Poll. We asked everyone who picked up the phone which ballot they would choose, and tried to replicate the experience of making that choice. For example, here’s what a neighbor island voter heard:

As you know, every Hawaii voter can choose between voting in the Democratic or Republican primary. The Democratic Senate primary is expected to be a close race between Mazie Hirono and Ed Case. The Democratic Congressional primary includes Tulsi Gabbard and Mufi Hannemann, and is also expected to be close. On the Republican ballot, there is a Senate contest between Linda Lingle and John Carroll thatʼs not expected to be as close. With this in mind, are you voting in the Democratic primary, the Republican primary, or neither primary?

Does that wording — particularly the “not expected to be as close” section — steer too many independents and Republicans to the Democratic primary? The split on Election Day could be the difference between victory and defeat.

What to Look for Saturday

Even if the first printout of election results Saturday night shows Hirono ahead, if her lead isn’t sizable — more than a few percent — she might be in trouble when the rest of the votes are counted at the end of the night.

That’s because the first results, coming out around 6:30 p.m. or whenever the final election precinct has closed, are going to be exclusively absentee voting — the entirety of the early walk-ins and a significant chunk of mail-in votes. The Civil Beat Poll found Hirono was leading, 51-44, among the 30 percent of respondents who had already voted. Case led, 49-42, among the 68 percent who said they would definitely be voting, but hadn’t yet.

That’s why turnout will be a decisive factor in the race. If everyone who said they would vote actually shows up, that works in Case’s favor.

In particular, a larger-than-expected turnout on Oahu by anti-rail, conservative Ben Cayetano supporters would be a boon for Case.

Though the mayor’s race is nonpartisan and some Cayetano voters will pick up a Republican ballot, 60 percent of Cayetano voters who said they’ll participate in the Democratic primary indicated they’ll vote for Case. Hirono got only 35 percent support from Cayetano voters even though she served as Cayetano’s lieutenant governor for eight years. That tie is severed; Cayetano even gave Case $500 last month.

Advertisements, Debates and Endorsements

In an effort to understand how voters made up their minds across major races this year, Civil Beat asked respondents how much valuable information they received from a variety of sources. What they said and who they said they’re supporting offers a glimpse into where candidates shine and where they don’t.

For Democratic primary voters, the most important outside influences on voters were, in order, news coverage, candidate debates, political advertising and endorsements. (Nearly all respondents said they conducted their own independent research of the candidates.) News coverage didn’t push voters one way or the other, but the other variables did.

  • Of the 42 percent who said debates provided a lot of information, Case led 50-44. Of the 30 percent who said the debates offered a little information, he led 54-41. And of the 23 percent who said they got no information from debates, Hirono dominated 57-36. This shows Case was perceived by voters to be a stronger debater. Only one Case-Hirono debate was widely televised, and that was after Hirono had avoided it for months.
  • Of the 25 percent who said they got a lot of information from advertisements, Hirono led 50-43. Of the 39 percent who said they got a little information from ads, Hirono led 51-43. And of the 32 percent who said they got no information from ads, Case led 55-39. Voters are reluctant to admit they’re influenced by commercials, but this split shows that Hirono’s heavy fundraising advantage worked in her favor, as she was able to buy time on television, on the radio and in print to get her message out.
  • Of the 12 percent who said endorsements were very important as they made up their minds, Hirono led 64-30. Of the 30 percent who said endorsements provided a little information, she led 53-39. And of the 54 percent who said they got no information from endorsements, Case was ahead 57-38. Hirono’s endorsers included the AFL-CIO, the Hawaii Government Employees Association, the United Public Workers, the Sierra Club and the Hawaii State Teachers Association, among others. Case’s only major endorsement came from the University of Hawaii Professional Assembly. Hirono led 51-41 among union households; Case led 50-43 among everyone else.

Follow Civil Beat on Facebook and Twitter. You can also sign up for Civil Beat’s free daily newsletter.