As the Voter Churns

The conventional wisdom in New Jersey’s gubernatorial race is the better that Chris Daggett does, the worse Chris Christie does. Certainly, Daggett’s rise in the polls over the past four weeks is a nearly point for point match with Christie’s drop in support.

That certainly is true at the aggregate level, as I have noted elsewhere. Specifically, if you compare this week’s Quinnipiac Poll to the one they released September 1, you will find that Christie’s support dropped by 6 points, Corzine’s increased by 3, Daggett’s increased by 5 and Undecided decreased by 1. While Corzine made some gains, it seems the big switch was from Christie to Daggett, with Undecided remaining stable.

I stand by this analysis, but there may be more to this phenomenon than the naked eye can see. Research conducted by the Monmouth University Polling Institute with a panel of New Jersey voters indicates that this “net” effect may actually be masking a lot more individual-level churning in the electorate.

The first round of our online panel was interviewed September 23-28 (Wave 1). A total of 340 of these respondents then participated in a second round of interviews on October 9-14 (Wave 2). [Note: the intention of this panel study is to track individual level change over time. As such, it is not necessarily designed to be representative of candidate choice for the full electorate. That is why I refrain from reporting “horse race” percentages here. We’ll leave that for our standard telephone polling.]

The survey analysis divided the vote choice question into 14 separate categories. Those who make a candidate choice (Christie, Corzine, Daggett, Other) were asked if they are either “very sure” about their choice or “might change” their mind before election day – leading to a total of 8 categories. Those who initially indicate they are Undecided were then asked if they “lean” toward a candidate – producing 5 categories (Lean to Christie, Corzine, Daggett, Other or do not lean to any candidate). The final category is for those who say they will not cast a vote for governor on the ballot.

In the Wave 2 interviews, fully 66% of participating voters stayed in exactly the same place in the 14 vote choice categories where they started in Wave 1. Another 16% stayed with the same candidate, but shifted their strength of support (e.g. from lean to sure, etc.).

In the first wave of interviews, a total of 7-in-10 respondents said they were “very sure” about their vote choice. Two weeks later, though, 15% of these “firm” voters had changed their minds – including 9% who softened the level of support for their chosen candidate and another 6% who actually switched their preference to another candidate.

Grouping all levels of support (sure, might change, and lean) together, both Jon Corzine and Chris Christie held onto about 9-in-10 of their voters from Wave 1 to Wave 2. Chris Daggett held onto about 8-in-10 of his Wave 1 voters. [Make no mistake – the sample sizes are relatively small, especially for Daggett. Grains of salt should be large and copious.]

So how did Daggett increase his overall margin? Apparently, by pinching voters from both major party candidates and other independents as well as picking up a bigger share of the undecided vote.

It seems that part of Chris Christie’s previous support is also leaching into the Undecided column. Thus, the aggregate Undecided vote share remains stable, while Daggett’s net support grows and Christie’s drops. The inference here is that the 5% of voters who are Undecided in this week’s Quinnipiac poll (or any poll for that matter), are not the same individual voters who were Undecided a month or so ago.

Another preliminary finding from this study is that all candidates saw some of their voters stay with them, but shift the strength of their support. Both Corzine and Daggett saw more of their supporters shift to a stronger rather than weaker position. Christie, on the other hand, had more supporters weaken their level of support than strengthen it.

One implication is that many of the Undecided and “leaning” voters today have flirted with being a Christie supporter in the past. Can the Republican win them back or has he lost them for good? That’s a question will be looking at over the next few weeks.

On a final note, please view these findings with caution. This analysis represents the results of a relatively small, self-selected panel study. However, the results suggest that there is a lot more individual-level churning involved in the recent Christie to Daggett swing than the top-line poll numbers indicate.