August 8, 2014 by Thomas Hauber
As Hurricane Iselle hovers over Hawaii Volcanoes National Park, the Hawaii Democratic primary also sizzles in its own way. Hurricane habits and movements are capricious and so are voters, but we will know tomorrow which way both have gone. What is most interesting in the final days before tomorrow’s Democrat primary is the striking difference in two recent polls each taken at the same time from supposedly the same voters showing completely opposite results?
The August 1st poll published by Honolulu Civil Beat gave interim US Senator Brian Schatz a 49-42 lead over US Representative Colleen Hanabusa. Civil Beat, Hawaii’s largest online news service, used Merriman River Group who has polled the race on previous occasions. Merriman’s executive Director, Matt Fitch, has stated the most recent poll shows a clear trend for Schatz, up from its own May poll of 44-39.
Merriman River Group provides election management services for private organizations and political consulting services for public candidates and campaigns. Its poll for Civil Beat surveyed 1,240 registered Hawaii voters statewide from July 24-28. It uses IVR technology that is totally automated.
Contrast this with a poll released August 4 conducted by Hawaii research firm Ward Research for Hawaii’s biggest print media the Honolulu Star-Advertiser. Their poll shows US Representative Colleen Hanabusa with a surprising lead over Schatz 50-42.
Since 1980, Ward Research also provides research services and consultation to companies in Hawaii and across the nation. The Ward poll is a live-caller poll of 692 likely voters conducted from July 21-29. It uses live techniques including cell phone users.
Both Ward and Merriman are private companies that don’t work for free. Political polls when done right can cost money, from $9,000 for a single House district poll to $25,000 for multiple races, according to President of Ward Research Becki Ward in a recent Hawaii Public Radio piece. With a declining number of public media outlets, Ward says, “The expense of a public poll is difficult for media to justify.”
So who paid for these polls? Neither candidate. Hats off to Hawaii media for footing the bill. Both pollsters were upfront with their affiliations. On-line Civil Beat paid for its own poll, and print Star-Honolulu-Advertiser paid for Ward. So much for claims of candidate/campaign bias. So why the difference?
There is some debate about IVR vs live-calling. Interactive Voice Response allows a computer to interact with humans through tones and voice recognition. It can reach more callers cheaper and claims to avoid errors introduced by live caller voice and question inconsistencies. Live-calling from a local phone-number claims to achieve a better quality response by putting a voter at ease using, for example, a local voice inflection.
That is a topic for another day. Suffice it to say, the real guts of any poll is in the sample, i.e. who of the “likely voters” is really going to vote? What does the sample pie look like? What percentage of likely White voters or Asian voters or native Hawaiian responses do we ultimately count? The sample is always somewhat subjective. The pollster’s art is reflected when the sample responses match the actual vote.
There are always wild cards. Will weather affect the turnout, what about mail-in ballots and so on? So who got it right? We’ll know in a day or so.