A new Daily Kos/Research 2000 poll of the New Jersey gubernatorial race provides further corroboration that the race between Democratic Gov. Jon Corzine and his Republican challenger Chris Christie is getting closer — but another claim, that there is a potential latent minority vote for Corzine that is yet to surface, is not necessarily the case.
The top-line numbers: Christie 46%, Corzine 42%, independent Chris Daggett 7%. The margin of error is ±4%. Two months ago, Christie led by 48%-40%, and Daggett was not included in the poll. This in line with other polls that have shown Christie only leading by three or four points.
Kos makes an observation that his internals show a whopping 25% of black voters, 1% percent of Hispanics, and 13% of other groups are undecided — meaning that those voters could potentially break big for Corzine. However, a key issue here is that racial sub-samples have a higher margin of error, and minority sub-samples in polls have a very high margin of error. Let’s look at two other recent polls.
The Monmouth poll had a top-line of Christie 43%, Corzine 40%, Daggett 8%. A combined sample of black and Hispanic voters had only 7% undecided. These minority groups were given to me by Monmouth as a combined set because the sample size was too small. The margin of error on the minority group sub-sample is a whopping ±14%.
The Quinnpiac poll had a top-line of Christie 43%, Corzine 39%, Daggett 12%. Quinnpiac has informed me that they had only 9% of blacks as undecided. A breakdown for Hispanics was not given, because the sample was too small. The margin of error on the black voters group was ±13.2%.
Research 2000 president Del Ali told me that the margin of error for his black voters sub-sample was ±11.4%, for Hispanics ±15%, and for other ±20%. He agreed with me on my doubts about releasing such small samples or taking them too seriously, but also explained: “I understand why Markos does release them. It is for transparency reasons.”
