Once again, as in 2006, Jewish Democrats are trumpeting the results of the national exit poll, which this year seems to show a very solid win for Barack Obama over John McCain among Jewish voters. And once again, as in 2006, anyone who has a clue about the exit polls and how they are conducted, should be much more cautious than the NJDC about interpreting the results. In 2006, the national exit poll surveyed about 200 Jews out of a national sample of 13,000, barely 1.5% of the voters interviewed. This year, about 2% of the combined 20,000 exit poll interviews conducted on the street or by phone were Jews, or about 400 in total.
Jews make up about 3% of the electorate, about 50% higher than their share of the population, and in the 2004 national exit poll, 3% of those surveyed were Jewish. Why has the national exit poll under-represented Jewish voters in the last two elections? Could that matter in terms of whether the national exit poll data accurately represented how Jews voted in the last two elections?
The national exit poll survey does not purport to be a random sample of voters, either nationally, in any state, or among small subgroups such as Jewish voters. If a pollster were trying to conduct such a survey for the Jewish vote , then Jewish voters in various parts of the country, and of various synagogue groupings -- Orthodox, Conservative, Reform , and unaffiliated Jews, would all be included in such a survey, if their numbers were sizable as a share of the total Jewish vote.
The people who conduct the national exit poll survey admit that their interviewers tend to be young, and more likely to conduct interviews with certain voters than others (over- representing Democrats and liberals). This is why on Tuesday night last week, many networks did not call some states for Obama despite big leads for him in the exit polls for these states. When the analysts compared the exit poll data in certain precincts in a few battleground states with actual tabulated vote totals, the exit poll data consistently overstated the Obama share of the vote.
Might Orthodox Jewish voters, who are far more likely to vote Republican, be underrepresented in an exit poll survey? How likely is it that Orthodox Jews would walk up to be interviewed by an exit pollster, or have an exit pollster show up at their polling place in Crown Heights or Williamsburg, or Skokie, compared to a pollster appearing to interview more liberal Jewish voters on the upper West Side of Manhattan or some affluent Jewish suburb?
According to the national exit polls, Jews voted for Obama over McCain 78% to 21%. The NJDC hailed this victory, since it was by a bigger margin than the margin by which Kerry beat Bush among Jewish voters in 2004 according to the exit poll data from that year (74%-25%). All things being equal, why should it be a surprise that Obama did better among Jews than Kerry did, since Kerry lost the popular vote by 2.4% nationally, and Obama won the popular vote by 6.5% nationally, a 9% shift between 2004 and 2008. By this measure the Jewish margin for Obama was 8% greater than it was for Kerry, while the national margin among all voters for Obama was 9% higher than it was for Kerry. This would suggest Obama slightly under- performed among Jewish voters, according to exit poll data.
Remember too that in 2006, the NJDC was boasting of the exit poll data from the Congressional elections which showed Democrats winning 87% of the Jewish vote. If those numbers were real, then Obama lost ground among Jewish voters the last two years, only getting 78% of the Jewish vote this time.
The NJDC is also happy that Obama did so much better among Jews in the exit poll data than he seemed to be doing in polls of Jewish voters conducted earlier in the summer and fall by other groups (e.g., Gallup, the AJC, J-Street, Quinnipiac polls in various states) Some of those earlier surveys showed Obama winning by a 2 to 1 margin among Jewish voters, and later in the election cycle, by a 3 to 1 margin, similar to the 2004 exit poll results. Obama moved up considerably in national polls of all voters from the time the first Jewish specific surveys were conducted, so one would expect his Jewish voter numbers to have moved as well. The latest Jewish voter surveys conducted by professional pollsters close to the time of the election suggesting a similar percentage of the Jewish vote for Obama as for John Kerry in 2004, are more reliable than the national exit poll as a measure of how Jews voted in 2008. These were professional surveys, with a stated margin of error, and random sampling, neither of which is associated with the national exit poll subgroup of Jewish voters. To show the problem with putting too much weight in the national exit poll, look at the results for gay and lesbian voters, who were 4% of those interviewed both in 2004 and 2008, a larger share of the national exit poll survey than Jews were in either year. In the 2004 national exit poll, gays and lesbians backed John Kerry over George Bush by 77-23%, a 54% margin. But in the 2008 exit poll, gays and lesbians voted for Obama over McCain by 70% to 27%, a 43% margin. Why would the margin for the Democratic candidate decline by 11% among gays and lesbians when the national margin for Obama was 9% greater than it was for Kerry? Did Obama have any weakness among gay and lesbian voters that Kerry did not have? I cannot think of one.
It is possible that the gay and lesbian vote recorded in the national exit poll may have been off by 20% or more in either 2004 or 2008 or both surveys were off by that amount combined. By the same token, taking the 78% to 21% exit poll victory for Obama over McCain among Jewish voters as gospel is equally suspect. If you do not know the margin of error for a survey sub-sample, how certain can you be that the survey results are a good snapshot of the community being surveyed? If the survey group was not a random sample of voters in that subgroup, what weight can you give the survey result?
I think it is highly likely that Obama defeated McCain by a decisive margin among Jewish voters. Beyond that, I can not venture any opinion based on the national exit poll data. Anyone who reads too much into that exit poll data, is really projecting what they want the results to mean, rather than what they reveal.
Richard Baehr is chief political correspondent of American Thinker.