Wednesday, June 11, 2008

Summertime, and the polling is easy

Polls released in the past few days provide an early gauge of the upcoming elections in New Jersey. In the U.S. Senate contest, today’s Quinnipiac Poll gives Frank Lautenberg a healthy 9 point lead over Dick Zimmer, whereas Monday’s Rasmussen poll showed the incumbent with an insignificant 1 point advantage. The conventional wisdom says that New Jersey voters tend to flirt with Republican candidates early in the campaign, only to return to “blue state” form by mid-October. However, this is a relatively new development.

As recently as 1992, New Jersey was still in play for presidential elections, and both the 1993 and 1997 gubernatorial elections saw the GOP candidate win in a squeaker. And while New Jersey voters have not sent a Republican to the U.S. Senate since 1972, they came within 3 percentage points of doing so in both 1994 and 2000.

Summer polls make interesting news fodder, but what do they tell us about where a campaign will eventually end up? I took a look back at polls taken in seven different races in New Jersey since 2000. To be fair to individual pollsters, I averaged the Democratic candidates’ margin over/under the Republican candidate in each race for five time periods: June through August, all of September, early October, late October, and the final November polls. [As a side note, the final polls’ average margin was within 2 points of the election day victory margin in five of the seven contests – perhaps a compelling argument for averaging poll results!]

NJ Democratic Candidate Advantage   (poll averages) *

ElectionSummerSeptEarly OctLate OctNovActual
’06 Sen+3-2+4+5+7+9
’05 Gub+10+10+7+8+6+10
’04 Pres+12+3+4+8+5+7
’02 Sen++7-7+2+7+12+10
’01 Gub+17+15+12+14+13+14
’00 Pres+4+12+14+9+10+16
’00 Sen+16+9+14+7+4+3
*   Data for 2006 are from 15 polling organizations and for 2005 are from 11 pollsters. The 2000 to 2004 data are averaged from Quinnipiac and Eagleton polls.
+   Due to a candidate switch in 2002, the summer and September polls test a different Democratic nominee than the later polls. While this election is one of New Jersey’s wonderful little anomalies, it is included in the table for continuity and doesn’t necessarily disrupt any trend.

One question is whether the summer polls are predictive of the final outcome. When it comes to the final vote margin, they are not. The summer polls hit the nail on the head just once (in 2005), and were within 3 percentage points of the actual victory margin in 2001. They were also within 3 points in 2002, although the Democratic candidate was literally a different person in the summer and at the final bell. In 2004 and 2006, the summer polls were off the final margin by 5 to 6 points. And they were off by double digits (12 to 13 points) in both the presidential and senate races of 2000.

Interestingly, the differences between the summer polls and the actual victory margin show no partisan consistency. In two elections – 2000 president and 2006 senate – the Democratic candidate performed better on election day than he did back in the summer. [This was also true for the 2002 senate switcheroo.] However, for three elections – 2000 senate, 2001 governor, and 2004 president – the Democrat did relatively worse in the final vote margin than in the summer polls.

Another question is whether New Jersey Republican candidates show stronger support in summer polls than they do later in the campaign season. This does not seem to be the case either. Only in the 2000 presidential race did summer polling show the narrowest D minus R gap of the campaign season. For three races (2002, 2004, and 2006), the Republicans actually enjoyed their best showing in the September polls. The 2001 election showed the Democratic advantage at its slimmest, albeit nominally so, in the early October polling. Only in the 2000 senate and 2005 governor races did the Republican candidate narrow the gap throughout the final weeks of the campaign, although the movement in 2005 was neither substantive nor predictive of the final outcome.

Indeed, the only point on which all seven races were consistent is that the Democrat was ahead in the summer polls and went on to win each of those elections. Given the very different pictures painted by the two polls released this week, we will need to see a few more polls before we have a good sense of how this race is shaping up. However, there are a number of factors this year that will probably make it even tougher than usual for the GOP to do well in New Jersey. A Republican who stands a chance of winning statewide needs to start off with a broader base of support than the party appears to have now.

For instance, both the February 5th and June 3rd primaries registered higher turnout among Democrats than Republicans. This in itself is not unusual, but in both instances the Democratic turnout was nearly double that of the Republican vote. That is definitely unusual. Also, the Monmouth University/Gannett New Jersey Poll in April indicated that 6-in-10 Republican-leaning voters had no real preference as to who their party’s U.S. Senate nominee should be. This is certainly not evidence of tremendous enthusiasm among the GOP base.

Another major factor this year is that the economy has become the predominant voter concern – an 800 pound gorilla shoving all other issues to the background. A commonly accepted rule of politics is that certain issues naturally benefit one party over the other. Campaigns are designed to raise “their” issue as the primary consideration for voters. Democrats typically have an advantage when economic concerns dominate the debate. The GOP will need to move their issues – such as national security or taxes – to the top of voters’ minds if they want to grab the advantage here in New Jersey.

Couple all this with the fact that New Jersey voters rarely tune into statewide races until a couple of weeks before the election, and you have a hefty grain of salt with which to read these early poll results. Pollsters, Monmouth/Gannett included, will continue to track these races through the summer and fall. My advice is to look for the numbers behind the “horse race” to get a better understanding of both the mood of the electorate and the dynamics that will drive this year’s election. What are the top issues on voters minds? Are voters tuning in early this year? Are there partisan differences in enthusiasm for the campaign? In the early polls, this information will be more useful, and indeed more interesting, than the horse race numbers in giving us a picture of the electorate, both here in New Jersey and throughout the country.

Monday, June 2, 2008

A Numbers-Based Primary Primer

The candidates have staked out their issues and have sought to bring undecided voters into their camps. But when it comes down to it, primary elections are mainly a numbers game. There are generally few, if any, issue positions that separate the contenders – that is if there is even more than one candidate seeking the nomination. So, the following information is to provide a numbers-based context for the June 3rd primary.


Hey, didn’t we already vote in a primary this year? Yes, and in record numbers. For the first time, New Jersey separated its presidential contest from the primary election for all other offices. On February 5th, 1,141,000 voters came out to vote in the Democrat primary and 566,000 voted in the GOP contest. Is this a harbinger for higher turnout Tuesday? Probably not. One of the reasons given for New Jersey’s generally low turnout in most elections is that we hold too many of them. Including tomorrow’s primary, some towns in the state will have held FIVE elections in the past four months! Yes, you read that right – in addition to the February 5 and June 3 primaries, there were fire district elections on February 16, school elections on April 15, and non-partisan municipal contests on May 13.

Throw in November’s general election, and somewhere in New Jersey, probably in Delran, there lives a voter who will have gone to the polls six times this year. Unfortunately, he or she is probably the only one. Most voters are fatigued by all these elections, and so most observers expect that tomorrow’s turnout will be even lower than usual.

A typical primary election headlined by a contested U.S. Senate seat would draw about 275-300,000 Republican voters and maybe 350-400,000 Democrats based on turnout figures from the past 15 years. The Dem number is less certain since the only recently contested primary occurred during the 2000 presidential year. If turnout looks more like the 187,000 Republicans and 219,000 Democrats who voted in the 2006 Senate primaries (where the front-runners faced only token opposition), then it’s likely that primary fatigue has been a factor this year.

Democratic Senate Race

This U.S. Senate race is a story of party lines. Not surprisingly, Rob Andrews has lined up solid support in his native south (as far as getting the party line on the ballot), but this will only account for about 22% of the total Democratic turnout statewide. In the 2000 Senate contest – a similar North versus South battle between Jon Corzine and Jim Florio – Florio won the seven southernmost counties by a 2-to-1 margin. Even with his poor showing in South Jersey, Corzine only needed 55% of the Democratic vote from the rest of the state to win the nomination. He actually got 65% – including 77% in Essex and 78% in Hudson.

Assuming that Morristown mayor Donald Cresitello takes about 8% of the statewide vote and Andrews does as well in the south as Florio did, Frank Lautenberg will only need about 49% of the vote in the rest of the state to take the nomination. There are some rumors that party operatives in Essex, Union, and Middlesex may be secretly working for Andrews. It’ll be worth watching how well he does in those counties.

GOP Senate Race

The Republican race is in greater flux. Dick Zimmer has better statewide name recognition, but Joe Pennacchio may have support where it counts most. Basically, Zimmer has the party line in counties that represent about 48% of a typical GOP primary vote, while Pennacchio has the line in counties which account for 49% of the projected vote. Importantly, Pennacchio has the line in the two biggest counties – Bergen and Morris – which together represent about one-quarter of the GOP vote. Since Morris is his home county, it’s worth paying attention to both his margin and the total turnout in that county. There’s also a contested freeholder race in Bergen that could have an up-ballot effect.

Furthermore, Pennacchio has the party line in the four counties that fall in the 7th congressional district – one of two hotly contested open GOP primaries. This means he is running with Leonard Lance in Hunterdon and Somerset. (He’s also on the line with Kelly Hatfield in Union and Kate Whitman in the Middlesex portions of that district). Look for a potential reverse coattails effect in those races (see below for the congressional outlook). It should be noted that Zimmer has the party line in the counties comprising the 3rd congressional district, but it’s not clear whether this race is generating the kind of voter enthusiasm that will help Zimmer.

Also of note is that Murray Sabrin, who has only one county line (Gloucester), is running bracketed slates in CD#s 1, 2, 6, 7, and 11. It’ll be interesting to see if he does better in those areas where he is running a slate – and if that hurts Pennacchio.

For historical perspective, in the 2000 U.S. Senate GOP primary, only three counties were lost by the party organization’s designated choice. On the other hand, Doug Forrester won the 2002 Senate nomination despite having only one line. And if we throw recent gubernatorial contests into the mix, it becomes anybody’s game. So the larger question is whether party lines will carry the day, or if GOP primary voters will reject the county party organizations’ choices, as they have done many times in the past.

GOP Congressional Races

All eyes are on the 3rd and 7th districts, where retiring GOP incumbents have led to heavily contested fights for the party’s nomination. The 7th district is a free-for-all among seven contenders, led by state senator Leonard Lance and Kate Annis, nee Whitman (the daughter of the former governor, who is reverting to her parents’ last name for purposes of the ballot). Lance has the party line in his home county of Hunterdon, along with Somerset – which together make up two-thirds of the Republican primary vote in that district. (Note: I’m averaging turnout share from the last three congressional races.) Despite residing in Somerset, Whitman has only the Middlesex line, which comprises just 6% of the GOP vote. Kelly Hatfield took the line in her home county of Union, making up the remaining 27% of the vote.

It’s worth watching Hunterdon County. About 3-in-10 voters in the 7th district have already seen Lance’s name on the ballot as a state legislator. Most, if not all, of these Hunterdon voters have become accustomed to voting for him. It’ll be worth watching to see if Lance can break the 50% barrier in his home county among this crowded field. If so, he’ll only need a nominal win in Somerset to take the nomination.

The 3rd congressional district race has been acrimonious, to say the least. Of the three contenders there, the two main candidates are Medford mayor Chris Myers, who has the Burlington and Camden lines, and Ocean freeholder Jack Kelly, who has the line in his home county. The Ocean County portion of GOP primary turnout in this district has ranged from 39% to 49% in the last three elections. However, none of those congressional nominations have been contested and the turnout has been largely driven by other offices. Kelly, as a freeholder, is known to the voters of Ocean, whereas Myers’s base is more local. Furthermore the Burlington party organization is seen as more fractured than the Ocean party. This race may come down to which county – Ocean or Burlington (including the small part of Camden in the district) – can turn out more than half of the total vote in this primary.