The One-Offs: New Jersey

In polling, it is trends that matter.

In polling, it is trends that matter.

In polling, it is trends that matter.

I will keep saying it until the nation’s news organizations listen.

Today’s example of a “one-off” American presidential poll causing the collective heads of news organizations to become unglued comes from New Jersey.

In my baseline tally, the Garden State is classified as “Solid” Democratic. This is due to its history of voting in American presidential elections. In the past seven elections (since 1988), New Jersey has only voted for the nominee of the Republican Party once (1988). In the last six elections, New Jersey has given its votes in the Electoral College to the nominee from the Democratic Party. In the last five elections (since 1996), the Democratic Party presidential candidate has received a majority of the votes. In the last four elections (since 2000), the median margin of victory for the blue side has been 15.7%. While not quite as “Solid” as California or the District of Columbia, New Jersey is still quite solid for the Democratic Party.

Since the start of this year, there have been six polls taken documenting the presidential preferences of the citizens of New Jersey (full list here) for 2016. In five of those surveys, the nominee from the Democratic Party, Hillary Clinton, wins by margins of 16 percentage points, 14, 7, 11, and 15 for an average of 12.6 percentage points over her opponent, the nominee from the Republican Party, Donald Trump.

So what of that sixth poll?

Given that level of support for the Democratic Party presidential nominee, I guess it should not have been a surprise that trumpets would blare and click-bait would be written when a poll was released showing Clinton only defeating Trump by 4 percentage points. Stories were written of the Trump-Clinton race in New Jersey now being “tight“, “close“, and with Clinton “clinging” to a lead.

All this was started on May 31 when Monmouth University released a poll (703 registered voters, margin of error 3.7 percentage points, results here) showing Clinton defeating Trump in New Jersey by a tally of 38%-34%. Given the margin of error, the results could also show a Trump victory where the Republican earns 37.7% of the vote while his opponent nets only 34.3%.

While a possible Republican victory in New Jersey would indeed by newsworthy, what did the trends say?

The survey released prior to Monmouth’s results was done by Fairleigh Dickinson. Released on May 24, this survey (702 registered voters, margin of error 3.7 percentage points, results here) shows Clinton defeating Trump 48%-37%, a margin of 11 percentage points. Given that the Republican Party nominee’s numbers are fairly consistent from the Fairleigh Dickinson survey (37%) to the Monmouth University poll (34%), what can explain the sharp dip in Clinton’s numbers (48%: Fairleigh Dickinson; 38%: Monmouth)?

Clinton’s usual cadre of supporters (Democrats, females, non-whites) seem to disappear between the Fairleigh Dickinson and Monmouth polls. In over a week, Clinton’s support among respondents who identify as Democrats falls from 80% (Fairleigh Dickinson) to 72% (Monmouth). For comparison, Trump’s support among Republicans stays near constant from Fairleigh Dickinson’s number of 74% to Monmouth figure of 73%.

The female demographic supported Clinton by 56% in the Fairleigh Dickinson survey, but that number drops to 44% in the Monmouth poll. For comparison, Trump’s share of the male vote also drops from a level of 49% (Fairleigh Dickinson) to 40% (Monmouth).

As for the non-white vote, Clinton sees her share of that demographic go from 70% (Fairleigh Dickinson) to 54% (Monmouth), while Trump’s share of the white vote remains level from 46% (Fairleigh Dickinson) to 44% (Monmouth).

This loss of core support might seem to spell doom for the presidential nominee from the Democratic Party, except for the fact that less than a week after the Monmouth University results were released, another poll came out showing Clinton defeating Trump by double digits.

Released on June 5, a survey done by CBS News/YouGov (1,194 likely voters, margin of error 3.8 percentage points, results here) showed that a sample set of New Jersey voters preferred Clinton over Trump by a margin of 49%-34%.

In the CBS News/YouGov poll, Clinton’s core support comes back. Voters who call themselves members of the Democratic Party support Clinton by a margin of 84%-5% This 84% level of support even surpasses the 80% level from Fairleigh Dickinson. Female respondents in the CBS News/YouGov poll support Clinton by a margin of 54%-30%, up from Monmouth’s figure of 44% and level with Fairleigh Dickinson’s figure of 56%. The percentage of non-white voters who select Clinton as their presidential choice (73%) in the CBS News/YouGov poll is even higher than Fairleigh Dickinson’s figure of 70%.

Yet, despite the CBS News/YouGov survey showing a return to the average margin of victory for the presidential candidate from the Democratic Party, the writers of news headlines were rather quiet when it came to declaring that New Jersey is, once again, safe to be painted blue come Election Day.

If the Monmouth University poll is taken away, the trend lines that emerge from the Fairleigh Dickinson survey to the CBS News/YouGov results show continued support for Clinton from Democrats, females, and non-whites. It also shows a lengthening of her overall lead from 11 percentage points to 15 percentage points.

So how to account for the Monmouth University results? What makes it a one-off in this 2016 election cycle? The answer is “I don’t know.”

This answer of “I don’t know”is a variation of the answer given by poll respondents when they choose to vote for neither Trump nor Clinton. The number of these undecided, “I don’t know” voters is what helps skew the Monmouth figures. Let’s go back to comparing the Monmouth University poll to the Fairleigh Dickinson numbers.

In terms of overall numbers, the percentage of voters who selected neither Clinton nor Trump in the Fairleigh Dickinson survey is 15%. In the Monmouth University poll, that figure is 27%. Among Democratic voters, 21% of those voters did not choose between the candidates of the Democratic Party and Republican Party. In the Fairleigh Dickinson poll, that figure of “I don’t know” Democrats is 12%. As for women, 17% selected neither Trump nor Clinton in the Fairleigh Dickinson poll while that number jumps to 28% in the Monmouth University poll. With non-whites, Monmouth University found 31% of that demographic opted for neither of the major party candidates while Fairleigh Dickinson only tallied 11% for that group who chose neither R nor D.

So, rather than the Monmouth University survey being a document showing a tightening of the American presidential contest in New Jersey, I am arguing that the pollsters simply found a rather large data sample of undecided Garden State voters and that is what sapped Clinton’s support. The CBS News/YouGov poll showed Clinton back on track with New Jersey voters…but that makes for a boring headline.

In polling, it is trends that matter.

In polling, it is trends that matter.

In polling, it is trends that matter.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s