Presidential elections are the Super Bowls of all elections, and for that reason, interest in predictions of who will emerge victorious are almost always of great interest.
Next year's election involves an incumbent, George W. Bush, seeking re-election, which to some degree helps simplify the process of gauging how the election is progressing. The Gallup Poll measures Bush's overall job approval rating on a frequent basis (107 times during his administration so far), and it has been established that by next year, Bush's approval rating will become an excellent predictor of his chances of winning a second term.
At this point, however, an incumbent's approval rating is not very useful in making predictions about Bush's re-election prospects. History indicates that too much can change between now and Election Day. Previous Gallup analysis has established that an incumbent president's job approval rating does not begin to be helpful in predicting the outcome of his re-election until about March or April of the election year. At that point, presidents with job approval ratings above 50% have generally been re-elected, while those with job approval ratings below that point have not.
The primary focus on the Democratic campaign at this point is centered on which candidate can win his or her party's nomination, rather than the candidate's ultimate electability. Unlike 2000, when Al Gore was the presumed nominee because of his position as a sitting vice president (even though he received a strong challenge from Bill Bradley), the field this year is essentially wide open. At this writing, there are nine Democratic candidates with the possibility that at least one other candidate, retired Army Gen. Wesley Clark, will jump into the fray.
The Gallup Poll frequently assesses the preferences of Democrats nationwide for their party's nomination, and has done so for the past 50 years. The purpose of this analysis is to look at the ability of these national trial heats to predict the ultimate outcome of the nomination process.
It is clear that national trial heats among Democratic candidates -- in the summer and early fall of the year before the election -- aren't necessarily related to which candidate will become the party's nominee. The voters for the most part have yet to tune in to the election, and many of the candidates are simply not well known enough for voters to establish strong opinions about them. And, in the past, many leading contenders made late decisions not to run (e.g., Mario Cuomo in 1992), which totally changed the complexion of the election.
There have been 10 races over the last 50 years in which there was a significant contest for the Democratic nomination: 1952, 1956, 1960, 1968, 1972, 1976, 1984, 1988, 1992, and 2000. The nature of these contests has changed over the years, of course. Prior to 1972, the Democratic nomination process was more focused on the conventions themselves, and the nominee was not always determined by who did the best in the primaries. For example, Hubert Humphrey, who ultimately won his party's nomination at a contentious convention in Chicago, was a late entrant in the nomination fight in 1968 (after Lyndon Johnson announced in late March that he would not run).
Here's what the historical Gallup Poll data say about the relative position of the potential nominees in Gallup trial heat polls among Democrats in the late summer or early fall of the year before each of these elections (i.e., about where we are now):
|
Democratic Nomination for President |
|||
|
Candidates |
Gallup Poll National Results |
Survey Date |
|
|
% |
|||
|
2000 |
|||
|
Democratic Primary |
Gore |
63 |
Early September |
|
Bradley |
30 |
||
|
1992 |
|||
|
Democratic Primary |
Cuomo |
30 |
Late August |
|
Bentsen |
29 |
||
|
Jackson |
29 |
||
|
Wilder |
8 |
||
|
Brown |
18 |
||
|
Clinton |
11 |
||
|
Tsongas |
7 |
||
|
1988 |
|||
|
Democratic Primary |
Jackson |
24 |
Early September |
|
Dukakis |
15 |
||
|
Hart |
-- |
||
|
1984 |
|||
|
Democratic Primary |
Mondale |
37 |
Early September |
|
Glenn |
26 |
||
|
McGovern |
9 |
||
|
Cranston |
5 |
||
|
1976 |
|||
|
Democratic Primary |
Humphrey |
23 |
Early October |
|
Wallace |
19 |
||
|
H. Jackson |
11 |
||
|
McGovern |
9 |
||
|
Muskie |
9 |
||
|
Shriver |
8 |
||
|
Carter |
<3 |
||
|
1972 |
|||
|
Democratic Primary |
T. Kennedy |
26 |
Mid-August |
|
Muskie |
22 |
||
|
Humphrey |
13 |
||
|
McGovern |
6 |
||
|
1968 |
|||
|
Democratic Primary |
R.F. Kennedy |
39 |
Mid-September |
|
Johnson |
37 |
||
|
Goldberg |
8 |
||
|
Humphrey |
6 |
||
|
1960 |
|||
|
Democratic Primary |
J.F. Kennedy |
30 |
Mid-September |
|
Stevenson |
26 |
||
|
Johnson |
10 |
||
|
1956 |
|||
|
Democratic Primary |
Stevenson |
55 |
Mid-August |
|
Kefauver |
16 |
||
|
1952 |
|||
|
Democratic Primary |
Stevenson |
-- |
No surveys at this time period |
|
Candidates labeled in bold and italics won the party's nomination for president. |
|||
As noted, these results may not be surprising given that the earnest campaigning really doesn't begin until Labor Day, according to many accounts. This year, for example, August polling shows that a majority of Democrats nationally said they were still not paying close attention to the race for their party's nomination.
Past experience suggests that the preferences of Democrats nationally begin to coalesce around the ultimate winner as the caucus and primary season gets underway in January of election year. When a candidate wins early primaries, voters in other states begin to shift their support to that candidate, especially as it becomes increasingly inevitable that he or she will win the nomination.
But what about that period of time just before the first caucus and the primaries begin? What does the historical evidence suggest is the case in late December and early January? Do Democrats nationally begin to move toward the eventual winner at that point, and will the national numbers this year at that point give us a good indication of who the eventual winner will be?
Here's what a Gallup analysis shows about the standing of the candidates in the last poll before the January caucus and primary season got underway:
|
Democratic Nomination for President |
|||
|
Candidates |
Gallup Poll National Results |
Survey Date |
|
|
% |
|||
|
2000 |
|||
|
Democratic Primary |
|||
|
Gore |
52 |
Late December |
|
|
Bradley |
38 |
||
|
1992 |
|||
|
Democratic Primary |
Brown |
21 |
Early January |
|
Clinton |
17 |
||
|
Tsongas |
6 |
||
|
1988 |
|||
|
Democratic Primary |
Hart |
25 |
Mid-January |
|
Jackson |
19 |
||
|
Dukakis |
10 |
||
|
1984 |
|||
|
Democratic Primary |
Mondale |
37 |
Mid-January |
|
Glenn |
20 |
||
|
1976 |
|||
|
Democratic Primary |
Humphrey |
29 |
Early January |
|
Wallace |
20 |
||
|
Carter |
4 |
||
|
1972 |
|||
|
Democratic Primary |
T. Kennedy |
32 |
Late December |
|
Muskie |
25 |
||
|
Humphrey |
19 |
||
|
McGovern |
5 |
||
|
1968 |
|||
|
Democratic Primary |
Humphrey |
-- |
Early January |
|
1960 |
|||
|
Democratic Primary |
J.F. Kennedy |
32 |
Mid-January |
|
Stevenson |
28 |
||
|
1956 |
|||
|
Democratic Primary |
Stevenson |
49 |
Mid-January |
|
Kefauver |
17 |
||
|
1952 |
|||
|
Democratic Primary |
Truman |
36 |
Early January |
|
Kefauver |
21 |
||
|
Barkley |
17 |
||
|
Stevenson |
-- |
||
|
Candidates labeled in bold and italics won the party's nomination for president. |
|||
One obvious conclusion: Even as late as the beginning of the election year, there is no necessary relationship between the candidate leading in Gallup's national trial heat surveys among Democrats and the eventual winner of the party nomination.
This was particularly true in the "olden" days when the primary season was not nearly as front-end loaded as it is now, and when there was occasionally still some suspense as the convention rolled around in the summer.
Note that 1952 Democratic nominee Illinois Sen. Adlai Stevenson wasn't even included in the Gallup surveys early in 1952, a time when there were still questions about whether or not incumbent president Harry Truman would run again. Even when Stevenson was included later in the spring of 1952, he received very low vote percentages. However, he went on to secure his party's nomination.
Similarly, there was very little indication in the national vote percentages in the early months of the 1972 Democratic campaign that anti-war candidate South Dakota Sen. George McGovern would eventually take his party's nomination. And, in early January 1976, Georgia Gov. Jimmy Carter was receiving only 4% of the vote of Democrats nationwide.
A similar phenomenon held true in the more recent Democratic races in 1988 and 1992. In both instances there was no strong clue in the national trial heats conducted before the primaries began in 1988 and 1992 that the two eventual Democratic nominees -- Massachusetts Gov. Michael Dukakis in 1988, and Arkansas Gov. Bill Clinton in 1992 -- would go on to take their party's nomination.
What all of this suggests, of course, is that much can and does change during the course of the campaign leading up to a national presidential election. Trial heat polls are valuable because they provide a window into the election process, and because they enable interested observers to monitor the impact of events and campaign tactics through the months leading up to the beginning of the actual primaries.
For example, national trial heats this year have shown a rise in the percentage of Democrats supporting Howard Dean, the failure of North Carolina Sen. John Edwards to catch on early with voters nationwide, and the strong showing for Clark even before an announcement that he is running for the nomination.
But, all in all, what we are seeing in the national polls now is much like the first quarter score in a football game. It certainly gives us clues as to what is going to happen as the game continues to unfold, but quite often bears little relationship to the final score when the game is ultimately over.
