Cross-posted at PolitickerNJ
An article in the Star-Ledger today reports on a poll that gives President Barack Obama and Governor Chris Christie the same job approval rating among New Jersey women. This would be big news if true. But I’m not so sure I buy it.
The poll, only conducted among New Jersey women voters, reports that 59% approve of Obama and 57% approve of Christie. The article claims, “The poll confirms a recent trend for Christie who has, for months, been closing the gender gap. In October, a Monmouth University/NJ Press Media poll found women approved of the governor 53 percent to 40 percent.”
One problem with that statement is that since today’s poll only surveyed women, there is no way to assess whether there is a gender gap in the current data.
But a much bigger problem with that statement is that it is patently false – the result of selective, or just plain bad, research on the part of this reporter. Yes, the gender gap was closing in October, but it has since opened up again, as the more recent Monmouth poll in February showed.
In fact, every New Jersey poll released since last month showed a significant gender gap for both Governor Christie and President Obama.
Three recent Garden State polls conducted by Monmouth, Quinnipiac, and FDU show President Obama’s marginal approval rating at 54% to 58% among female voters in New Jersey. This is similar to the 59% result in the poll reported today. However, those same three polls set Governor Christie’s approval rating among women at 46% to 50%, lower than the 57% in today’s poll.
|Monmouth 2/7 *|
* The Monmouth University Poll releases provide gender breakdowns for all residents. The numbers in this table are for registered voters, to be comparable with the other polls.
For background, among all New Jersey voters, all three polls found Governor Christie had higher net job approval ratings than President Obama – between +17 and +20 for Christie and between +6 and +9 for Obama.
On the gender gap, all three polls showed Christie with a whopping positive net rating among male voters – from +23 to +35 – and a smaller net positive rating among women – from +5 to +10. For Obama, his rating among male voters was in negative territory – from -1 to -6 – while it was decidedly positive among women – from +13 to +24.
And the trend for the three polls suggests that the gender gap for both politicians may have actually widened rather than narrowed over the past six weeks.
Today’s poll was conducted for Kean University. Kean started publishing polls last year, but the methodology (sample design, weighting and analysis) is farmed out to a private polling firm. In the past, they have used a Republican polling firm to conduct their surveys. It‘s unclear whether this was true of the current poll, because the article did not report this key methodological detail.
Unlike the three polls cited in the table above, Kean does not subscribe to the National Council on Public Polls principles of disclosure. In other words, it’s impossible from their press release – which is not available online – to assess how the poll was actually conducted. [Note: I emailed the poll director for methodological information, but have not yet received a response.] Aside from the sampling and weighting issues, it’s unknown whether this poll asked the same job rating question as the other three polls.
I am a strong proponent of having a variety of sound public opinion polls covering the same populations and topics. No one poll can be comprehensive. Having a number of pollsters attack different angles of the same policy issue gives us a richer picture of the state of public opinion on that issue.
And as we have seen with election polling, having a plethora of polls enables us to calculate an aggregate projection which tends to be pretty much on target. In terms of office holder job ratings, multiple polls provide an important validity check.
In this case, that validity check does not pan out. A combination of unknown polling techniques and poor reporting has given us a tantalizing front page story line, regardless of its veracity.