I am a big believer in polling averages, which over the 2004 and 2006 elections were excellent predictors of outcomes in statewide elections. However, in the Democratic primary season, the emergence of three distinct types of national primary polls are rendering polling averages more problematic. Since Obama's fundraising announcement, we have seen a) three polls that include the entire field but exclude Gore, b) three polls that include both Gore and the entire Democratic field, and c) three polls that include only Clinton, Obama and Edwards (plus another poll that might be type a or type c, but it isn't clear). Averaging polls across these three types produces questionable results, since they are all quite different from one another. Figuring out the current state of public opinion within the national Democratic primary electorate is further complicated by the lack of information on candidate standings in early states, by
the different (typically inaccurate) ways polling firms define the Democratic primary electorate, and by
the greatly different degrees to which these polls push undecideds. For an election watcher such as myself, this is all very frustrating. I just want to know where the campaign currently stands, and polling firms are not making that task very easy.
To try and sort this all out, here are the current averages for each of the three different types of national polls conducted entirely from April 5-15:
National Polls With the Entire Field But Without Gore
| Poll |
Gallup, 4/15 |
Rasmussen, 4/12 |
ARG, 4/12 |
Mean |
| Clinton |
37% |
32% |
36% |
35% |
| Obama |
29% |
30% |
24% |
27.67% |
| Edwards |
19% |
16% |
19% |
18% |
National Polls With the Entire Field Plus Gore
| Poll |
Gallup, 4/15 |
CNN, 4/12 |
LA Times, 4/9 |
Mean |
| Clinton |
31% |
30% |
33% |
31.33% |
| Obama |
26% |
26% |
23% |
25% |
| Edwards |
16% |
12% |
14% |
14% |
National Polls With Only Clinton, Edwards and Obama Included
| Poll |
CBS, 4/12 |
LA Times, 4/9 |
Time, 4/9 |
Mean |
| Clinton |
39% |
42% |
32% |
37.67% |
| Obama |
24% |
32% |
26% |
27.33% |
| Edwards |
21% |
20% |
25% |
22% |
There is also a CNN poll showing Clinton 36%, Obama 28%, and Edwards 15%, but it is unclear whether that poll only includes those three candidates, or whether it includes the entire field without Gore. I think it is the latter, but I don't want to include it in the polling averages until I know for certain.
These three charts produce three different margins between Clinton and Obama: 7.33%, 6.33%, and 10.33%. It would not be accurate to average these three gaps, since some polls are included in more than one of these averages. Further, I still maintain that many of these polls are pushing undecideds too hard (which probably inflates Clinton's lead), and that all of them are further inflating Clinton's advantage through the inclusion of too many low-information, unlikely primary voters. I would love it if every firm just included the entire field without Gore, did not push undecideds, and screened (at the very least) for likely general election voters. I would love it even more if there were at least a few more polls of early states, which of course also followed those methodological guidelines.
Using the data available, it is very difficult to figure out where the race stands. Clinton probably has a very slight edge nationally, but with Obama's money and donors, Edwards's strength in Iowa, and the current situations in New Hampshire and South Carolina (which are unclear and probably tied), I can't emphasize enough that any lead by Clinton is extremely slight. Anyway, we still have nearly nine months before the start of the primary / caucus season, so figuring out whether or not Clinton or Obama leads right now may not really be all that important. However, if I was doing a cattle call, I would think long and hard about putting Obama at #1. We will find out more when the next round of early state polls are released.
