New Jersey has no precedent for modeling a likely electorate in a special statewide election. Especially one that occurs just three weeks before the regular general election. And doubly so for one that occurs on a Wednesday rather than a Tuesday.
This alone may account for the variations in U.S. Senate polls released over the past two days. The Monmouth University sample is different from other public pollsters because we decided to use a list sample of registered voters that includes information about gender, age, past voting history and party registration.
[Reminder: party registration is how voters are listed on the voter rolls and is a very stable characteristic. Party identification is how people answer the question: “In politics today, do you consider yourself…” – it is not as stable and is subject to change based on the external political environment.]
Here’s an overview of some benchmarks from official New Jersey election records and voter list providers.
The last two general elections in New Jersey where the U.S. Senate topped the ballot saw turnouts of 46% (2002) and 48% (2006) of registered voters. By comparison, elections where Governor was the top office saw turnouts of 47% (2009) and 49% (2001 and 2005). Turnouts where the House of Representatives tops the ballot has been 42% (1998 and 2010) and off-year elections where the legislature tops the ballots have seen turnouts ranging from 27% to 34% of registered voters.
[Note: Turnout figures for elections prior to 1998 are not comparable, because of the Motor Voter law which registered a lot of “unlikely” voters. This increased the denominator of registered voters, but did not change the numerator of people who actually show up to vote.]
Of course, these turnout levels are all for regularly scheduled elections. There is no precedent for a special election in New Jersey.
We can turn to Massachusetts, though, which held a special election this past June to fill the Senate seat of John Kerry after he was appointed Secretary of State. That election saw a 27% turnout. Although the Democrat won handily, this is the low turnout level that the Lonegan camp hopes to see tomorrow.
On the other hand, Massachusetts held another special election in January 2010 to fill the late Ted Kennedy’s seat. This race, which was eventually won by a Republican, garnered a lot of national attention as a referendum on Obama – an issue the Lonegan camp hopes will factor into the New Jersey election. This election had a high turnout of 54% of registered voters. [Both Massachusetts special elections were held on Tuesdays.]
A review of voter list data indicates that the demographic composition of New Jersey’s electorate in non-presidential years is fairly stable, regardless of turnout.
For example, young voters age 18 to 34 made up just 8% of the electorate in 2009’s gubernatorial election, 9% in 2010 when the U.S. House topped the ballot, and 7% in 2011 when the state legislature was the marquee event. The proportion of voters age 65 and older was similarly stable at 34% in 2009 and 2010 and 38% in 2011. [By contrast, young voters comprised 18% of New Jersey’s 2012 presidential electorate, while older voters accounted for just 24%.]
Gender is also relatively stable, with women voters making up 52% (2011) to 53% (2009 and 2010) of the electorate. Gender by party [i.e. actual partisan registration status, not self-reported identification] is similarly stable. For instance, among registered Democrats, 57% to 58% of those who voted in 2009, 2010, or 2011 were women. The gender split is basically even among registered Republicans and unaffiliated voters. Women made up 49% to 50% of Republican voters and 49% to 50% of unaffiliated voters in those same three election years.
Finally, the Democrat-Republican party registration spread for non-presidential years has been a very stable 9 to 11 points regardless of turnout or election type. The only major variation is the proportion of unaffiliated (“U”) voters in the electorate, which decreases as overall turnout drops. In 2009, the D-R-U split was 42-32-26. In 2010, it was 43-32-25. In 2011, it was 44-35-21.
[Note: The D-R-U registration split is significantly different in presidential election years. In 2012’s electorate, for instance, it was 37-24-39.]
The historical turnout data presented above is based on New Jersey voters who show up to vote on the first Tuesday after the first Monday in November. We don’t know whether these patterns will hold for a Wednesday in October. Monmouth's internal likely voter model suggests that tomorrow's turnout will be between 35% and 40% of registered voters. But here’s one intriguing finding from the Monmouth University Poll to keep in mind. Just last week, we talked to voters who are known to have voted in at least two of the last four general elections. Fully 10% of these “regular voters” told us they had no idea that a special election was being held on October 16th!