CBS News has published a new poll showing President Bush’s job approval at 34%. I was going to spend the next 678 words denying that number but, as I think this is an aberration anyway, perhaps our time will be better spent talking about polling in general.
I present this as an informed amateur when it comes to polling. Real pollsters – Republicans and Democrats – are to statistics as Leonard Bernstein was to music: They see notes and numbers in a dimension the rest of us simply do not.
This column is the result of conversations with two very, very smart guys: Ed Goeas of The Tarrance Group and Neil Newhouse of Public Opinion Strategies. Any errors in what follows, however, are mine.
When a survey is “in the field” is important. The days that phone calls from a call center are being made to households are called being “in the field.”
The CBS poll was in the field from February 22 – 26. Two things: (a) The nation’s hair was on fire over the Dubai ports deal and (b) at least half of the survey was taken over a weekend.
Political pollsters know that surveys taken between Friday night and Sunday afternoon tend to veer sharply to the Left.
Another point to be aware of is whether you are seeing the results of a survey which sampled adults in the US or a survey of likely voters.
When political pollsters are doing their thing, they use a variety of techniques to determine whether or not someone is a “likely voter” and will, therefore, have an effect on the outcome. Non-voters are of no moment if you are in the final two weeks of a campaign.
CBS had a sample of 1,018 respondents which they weighted to reflect 28 percent Republicans; 37% Democrats; and 34% Independents. Not likely voters, but adults in the American population.
In the general population, those who claim to be Democrats outweigh those who claim to be Republicans by 7 to 9 percentage points.
However, the ratio of those who actually vote is much closer. In 2004 those who went to the polls and voted for President were about evenly split between Republicans and Democrats at 37 percent.
More important to this discussion, in 2002 – the last mid-term election – the percentage of participation among the voting age population was 36% - just over a third.
Political pollsters will press to get a respondent to admit they tend toward the Democratic or Republican party. These are called “leaners.” If you include “leaners” the percentages are in the low 40s for Republicans and Democrats, and only about 20% will end up as ture independents.
The order in which survey questions are asked can make a significant difference in the answers. A “push poll” (which the CBS was not) will start by asking you who are going to vote for. Then the phoner will give you some “information,” such as:
Would you be more or less likely to vote for Rich Galen if you knew he bragged about driving a vehicle which gets only 15 miles to the gallon.
After a series of such questions (“pushing” negative information), the phoner would again ask you for whom the respondent intend to vote. Assuming the information was presented in a damaging enough manner, the result would be significantly different than the first time the “ballot test” was asked.
The CBS poll doesn’t claim to demonstrate voter intention (although it will be portrayed that way by the popular press and the Democratic National Committee). But, you can look at the CBS poll and get a clear sense that the public at large is not happy with the current performance of the Administration of George W. Bush.
What you cannot do – or at least should not do – is to draw any conclusions as to whether these numbers will have any major impact on election outcomes next November.
It’s a long way from March 1 to November 7.
To answer your first question, Richard Nixon was at 27% job approval in an August, 1994 Gallup poll – just before resigning in the wake of the Watergate scandal.
On a the Secret Decoder Ring page today: A link to the CBS News poll; a Mullfoto which is quickly becoming boring; and, a Catchy Caption of the Day.