PRINCETON NJ -- The Gallup Organization has been tracking presidential job approval for more than 60 years. We ask a straightforward question -- " Do you approve or disapprove of the way [president’s name] is handling his job as president?" We have probably asked this question more than any other, and it has given us an unparalleled measure of Americans’ reactions to their presidents since the days of FDR.
The job approval measure has, in fact, acquired such an importance that it regularly turns up in movies and television shows about the presidency. The 1995 movie The American President, starring Michael Douglas and Annette Bening, contained a number of references to the fictional president Andrew Shepherd’s job approval numbers -- and even included a role of a pollster played by David Paymer. The hit TV series The West Wing contains frequent references to President Bartlet’s job approval numbers.
One of the biggest challenges in dealing with presidential job approval is the temptation to pay too much attention to the short-term change between any two measures. Presidential job approval numbers will fluctuate some from measure to measure based on standard measurement error -- as suggested in the margin of error estimates you see reported with most polls. The usual margin of error for a national poll of about 1000 people is about 3%. This margin of error refers to the differences that occur between any sample and the population from which it is drawn and meant to represent. What this implies, of course, is that if we take two job approval measures a week or two apart, it wouldn’t be unusual to find a few points difference between them, even if there had been no change in the underlying population parameter.
We also need to consider that presidential job approval as a measure is quite volatile by its very nature. There are several reasons for this. For one, we find that Americans tend to use their rating of the president as a repository for their feelings about what is going on in the country in general, and those feelings are subject to rapid change. Like the quarterback of a football team, the president tends to get too much credit when things go well, and to get too much of the blame when things go badly. So, almost anything that happens in the country can theoretically cause presidential job approval to go up or down.
I also think Americans feel comfortable changing the ratings they give the president on a short-term basis. The public is probably used to the idea that all of our elected officials can be and should be "graded"daily, weekly, and monthly. Like judges in an Olympics competition, the public can easily view itself as sitting in judgment of the president, marking him up or down based on the latest information they have received. The president in the United States isn’t subject to repeated votes of confidence as he or she would be in a parliamentary system, but job approval ratings from the public in a presidential system may provide the public some of this same thumbs up/thumbs down function.
It is also important to note that the president is almost constantly in the news. There is ample information flow from which the American public can make its judgments on the president’s performance on a day-to-day and week-to-week basis. There are very few days that go by in which the actions of the president are not covered in the news media in one way or the other. Some of these reports are neutral, but the American political and media system have institutionalized the idea that elected officials are fair game for criticism. The interested observer is quite frequently subjected to all manner of negative comments on the behavior, performance and motives of the president as well as spirited defenses of his actions and behavior.
For all of these reasons, then, it is not surprising that there should be real world changes from measurement period to measurement period in presidential job approval. Couple this real-world change with measurement error, and we have a situation in which we fully expect to see job approval movement from measure to measure. For the measure to stay absolutely still is probably the exception; to move is the norm.
Our challenge in interpreting these job approval changes is fairly straightforward. We need to pay attention to short-term movement, but use caution in asserting what these movements might tell us about long term, more lasting changes in the public’s perception of their president.
Let’s look back over the job approval ratings we have for George W. Bush. We have 16 measures stretching from our first rating on February 1-4, just a week or two after his inauguration, through July 19-22.
|
"Do you approve or disapprove of the way George W. Bush is handling his job as president?" |
|||
|
Approve |
Disapprove |
No opinion |
|
|
% |
% |
% |
|
|
2001 Jul 19-22 |
56 |
33 |
11 |
|
2001 Jul 10-11 |
57 |
35 |
8 |
|
2001 Jun 28-Jul 1 |
52 |
34 |
14 |
|
2001 Jun 11-17 |
55 |
33 |
12 |
|
2001 Jun 8-10 |
55 |
35 |
10 |
|
2001 May 18-20 |
56 |
36 |
8 |
|
2001 May 10-14 |
56 |
31 |
13 |
|
2001 May 7-9 |
53 |
33 |
14 |
|
2001 Apr 20-22 |
62 |
29 |
9 |
|
2001 Apr 6-8 |
59 |
30 |
11 |
|
2001 Mar 26-28 |
53 |
29 |
18 |
|
2001 Mar 9-11 |
58 |
29 |
13 |
|
2001 Mar 5-7 |
63 |
22 |
15 |
|
2001 Feb 19-21 |
62 |
21 |
17 |
|
2001 Feb 9-11 |
57 |
25 |
18 |
|
2001 Feb 1-4 |
57 |
25 |
18 |
If we had reported on Bush’s job approval change between the first June survey and the middle June survey, we would have said "no change." The same interpretation would be appropriate if we had selected a number of other adjacent points in the measurement time line -- between early and mid February, between late February and early March, between mid May and late May, and most recently between July 10-11 and July 19-22.
But, let’s see what else might have been possible. Between the March 9-11 poll and the April 6-8 poll, Bush’s job approval fell five points, and then went back up six points. Had we made a big deal about either of these changes we probably would have been guilty of making too much out of a short term blip that did not work out to be a longer term trend.
Perhaps more importantly, look what would happen if we picked two points further separated in time. By selecting the March 5-7 poll as point A and the May 7-9 poll as point B, and ignoring the other polls in between, we could have reported that Bush had lost 10 points in job approval in just two months. We could have made a similar interpretation by using just the April 20-22 and the June 28-July 1 polls. But a choice of comparison between March 26-28 and July 10-11 would have lead to a different interpretation.
All these point A and point B numbers are legitimate. The ups and downs they reflect are most probably, as we have noted above, a combination of some normal measurement error and some short-term real world change. In particular, the higher numbers in March could have reflected the American public’s reaction to the downed spy plane in China.
But, in looking at these numbers, we recognize that any two statistics in isolation can be less than representative of a broader pattern. A lack of intermediate points in time does not allow the reader to know if there has been fluctuation between measures, and without combining several measures in a row, we run the risk of paying too much attention to what may turn out to be a short-term blip.
For that reason, it is usually best to consolidate more than one survey when ascribing significance to change. The survey-to-survey numbers should be reported, but interpretation of the significance of the change can better be made using averages of longer periods of time.
One of the most convenient and useful units of consolidation is the quarter. Averaging out the polls done in a quarter smoothes out differences and allows us to search for more meaningful longer-term change.
With that in mind, what have we found for George W. Bush? We have just finished the second quarter of the Bush presidency, based on three-month periods from his inauguration on January 20. Gallup measured Bush’s job approval seven times in the first quarter of his presidency, from January 20 to April 19, and eight times in the second quarter, from April 20 to July 19. Bush’s first quarter job approval average was 58.4%, and his second quarter job approval average was 55.8%. This is about a 2.5% difference, representing general stability in job approval, despite some of the short-term fluctuations we have discussed above.
By way of comparison, here are some quarter-to-quarter changes from past presidents that show what type of change can occur in job approval:
- Bush the elder dropped 16% points from quarter three 1991 to quarter four 1991, from 69% to 53%.
- President Bill Clinton dropped 11% from quarter one of his first year in office in 1993, from 55% to 44%.
- President Jimmy Carter dropped 7% from quarter four of his first year, 1977 to quarter one of his second year, from 54% to 47%.
- President Ford dropped 15% points from his first quarter in office after taking over from Richard Nixon (the third quarter of 1974) to his second quarter in office (the fourth quarter of 1974, from 59% to 44%).
- Richard Nixon himself dropped 17% points from 61% in the first quarter of 1973, at the beginning of his second term, to 44% in the second quarter of 1973, as Watergate revelations began to move to the front page of the newspapers.
It is also worth noting, that where a president begins in his first year can bear little relationship to where he ends up in later years. Note these examples:
- Kennedy still has the record for the highest job approval average of all time -- 70%. Still, after enjoying a first and second yearly average of 76% and 70%, his job approval numbers were beginning to slip, down through the 60% range and to an average of just 59% in the third quarter of 1963. This drop, probably due to civil rights unease among whites in the South, was in part the reason he agreed to go to Texas in November of that year.
- One of the highest quarterly averages was the first quarter of 1991, when President George Bush (the elder, of course) had an average of 83% job approval. That was that period of time when the United States and its allies won the Persian Gulf War. Bush’s problem was that by the third quarter of 1992, when he was running for re-election against Bill Clinton, his job approval averaged 35% -- an amazing and highly significant drop of 48% points in just seven quarters.
- Despite the drop in his first year, which ultimately bottomed out at 41% for the third quarter of 1994, Bill Clinton climbed back up and was averaging 57% by the third quarter of 1996, enough to push him to re-election over Bob Dole.
- Jimmy Carter’s job approval averages in his first two quarters lingered in the 60% range. By his third year in office in 1979, his average numbers were in the 30% range, and after a rally in the last quarter of 1979 and first quarter of 1980, he fell back down again into the 30% range and, of course, was defeated by Ronald Reagan.
All in all, there are two major points here regarding presidential job approval. First, we need to be quite cautious in attributing too much significance to short term fluctuations in job approval, and to comparisons between just two points in time. Quarterly averages seem to be an excellent way to approach this issue where possible. Second, job approval can change, and can change rapidly, so what you see today may bear little relationship to what you will see tomorrow. George W. Bush seems to be fairly stable now with a job approval in the mid-50s, but whether or not he will drop down ala Bill Clinton in the months ahead, or soar up like his father is still very much in the air. And, from a longer-term perspective, Bush’s success in claiming a second term in office will be highly related to his job approval ratings in the year 2004, which in turn may bear little resemblance to what we find today.