Another Look at Excluding Non-Voters from Polls: There’s Plenty Evidence Views Differ from Voters; A Problem-Free Survey on Rail Is Long Overdue

"Class-Based and Value-Based Issue Attitudes of Voters versus Non-Voters" (see below).

It doesn’t take an advanced degree to appreciate the differences between voters and non-voters. Just a few clicks on your computer will find numerous resources that cite differences in education, wealth, attitudes on politics, vehicle ownership, employment, public spending and other characteristics.

The more we’ve clicked, the more we’ve become convinced Civil Beat made a significant error by excluding non-voters in its March poll on Honolulu rail.

Editor John Temple strenuously defended the survey when we first posted that criticism, but now that he’s left to take a position with The Washington Post, maybe Civil Beat will reassess the exclusion of non-voters – a huge misstep – and conduct a survey of a representative sample among all residents on rail transit.

Horse Race vs. Policy

Restricting a poll to voters makes sense when the intent is to learn which candidate is likely to win an election, but since government policymakers presumably serve all citizens, non-voters should not have been excluded in determining levels of support for Honolulu rail.

We believe doing so produced an erroneous conclusion – that support for rail had slipped. Taking the survey at face value, it showed less support for rail than previous surveys among voters. The views of the non-voting residents – one-half or more of total registered voters – were irrelevant to the pollsters, an astounding starting point.

A cursory Internet search finds plenty of support for the conclusion that voters are not necessarily representative of the entire population. Non-voters are more likely to be less educated, less wealthy, more likely to work multiple jobs, and – the most significant difference regarding the rail project – more likely to be dependent on public transit than voters.

The Research

Our ‘net search was just large enough to determine that many scholars have found significant differences between voters and non-voters. “Who Votes Now? And Does It Matter?” by Leighley and Nagler (2007) is one such source.

The researchers  didn’t examine opinion differences on public transit policy, but it’s significant that they did find differences on other issues – government guaranteeing of jobs and health insurance, among them. Among their conclusions:

“…the electorate seems to have become less representative of non-voters in its opinions. In 2004, there is a twelve percentage point difference between non-voters and voters believing that it is the government’s responsibility to guarantee jobs, and a nine-percentage point difference between voters and non-voters who believe that people should ‘get by on their own.’ These patterns both result in an electorate that reflects more conservative views than non-voters”

It isn’t a huge leap to expect similar differences between voters and non-voters on their preference to improve Honolulu’s public transit infrastructure – i.e., supporting construction of Honolulu rail to provide better service for transit-dependent citizens, or opposing rail and requiring citizens to “get by on their own” using their cars.

Throughout their paper, Leighley and Nagler refer to “significant differences between voters and non-voters…” on the issues they and others researched. They reference the 2004 Annenberg National Election Study, which supplied the table at the top of today’s post. The differences between non-voters and voters were obvious in that table, and there’s every reason to believe there would be similar differences between the two categories on the issue of whether Honolulu should dedicate billions of dollars to improving transit services on Oahu.

Conclusions

Leighley and Nagler make the following observations in their Conclusion:

We offer in this paper a more extended analysis of the extent to which voters represent non-voters. Importantly, we take issue with the assumption that voters are indeed representative of non-voters… (A)fter 1972, voters and non-voters differ significantly on most issues relating to the role of government in redistributive policies. In addition to these differences being evident in nearly election since 1972, we also note that the nature of the electoral bias is clear as well: voters are substantially more conservative than non-voters on class-based issues.”

Our own conclusion is that Civil Beat did not understand the inevitable consequence of excluding non-voters from its rail survey – a false reading on how much support rail enjoys among the population at large.

Voting patterns on Oahu show far less voter participation in Leeward Oahu than in East Honolulu, whose residents (generally) don’t support rail because the line won’t reach their neighborhoods. By restricting the survey to only voters no matter where they live, including the Leeward side, Civil Beat can’t truly say it knows what half the population thinks about rail – the half more likely to use transit and therefore potentially more supportive of building Honolulu Rail.

Why This Matters

Civil Beat’s survey, along with the Hawaii News Now/Star-Advertiser poll taken immediately after Ben Cayetano announced his candidacy for mayor, continue to be referenced in the media as if they were the Gospel Truth – repeatedly mentioned until those results have become part of the daily discourse on rail.

From where we sit, those polls were bogus. We’re still waiting for one that’s fairly timed with appropriate questions and an inclusive population sample to learn what Oahu residents really think about rail.

When that finally happens, the results are highly likely to show continuing majority support for the project – mirroring the results of three earlier polls that were problem-free compared to what we’ve seen so far this year from the media.

This post has been added to our "aggregation site" under the Public Opinion heading