In this section, I will maintain up-to-date reports and analyses of current polls on the presidential campaign. I will apply a method of analysis which I believe more accurately reflects the information contained in such polls than do the methods used by the traditional mass media outlets upon which we usually rely. And if you wish to learn the methodology I am using for my analyses, see below.
NATIONAL
The Current Picture As Of October 17th:
Bush Trending Down In 2-Way "Ticket" Surveys Of Registered Voters, A Showing Not Contradicted In Other Surveys. However Bush Either Leads Or Ties With Kerry In All Polling Universes. Undecided Trending Down In Most Polling Universes.
There are eight windows through which one can look at this presidential race. One can look at registered voters or likely voters. Within those two broad categories one can ask people's preference between the tickets of Bush/Cheney, Kerry/Edwards and/or Nader/Camejo or one can ask their preferences among the presidential candidacies of Bush, Kerry, and/or Nader. Finally one can choose to include or not include either the ticket of Nader/Camejo or the presidential candidacy of Ralph Nader.
In the below analysis I have subdivided my analysis between registered and likely voters.
The bottom line however seems to be that Bush is trending down while Kerry is holding his own. But Bush continues to lead Kerry narrowly. In addition, Undecided appears also to be trending down.
REGISTERED VOTERS: NATIONAL
The Current Picture As Of October 17th:
In 2-Way "Ticket" Polls Kerry Now Ties And Clearly Has Momentum As Bush Trends Down. 2-Way "Presidential Candidate" Surveys Still Show A Tie, With Undecided Trending Down
Bush/Cheney Now Lead Kerry/Edwards In Current 3-Way "Ticket" Surveys With Undecided Moving Up And Clearly In Third Place With Nader In Last Place. 3-Way "Presidential Candidate" Surveys Show It All Tied Up, With Undecided Trending Down
Cutting to the chase, it appears, when looking at all four flavors combined of registered voter surveys, that Kerry/Edwards are running in place, Bush/Cheney are continuing to trend down but probably maintain a slim lead over the challenger, and Undecided is probably trending down, with the one exception of the 3-way "ticket" surveys where Undecided is trending up.
Among 2-way surveys, a very peculiar and contradictory story emerges. Among "ticket" surveys of registered voters there appears to be rather encouraging news for Kerry supporters. The Kerry/Edwards ticket is still tied with the Bush/Cheney ticket, but the Bush/Cheney ticket has trended sharply down between the 24th and the 15th, somewhere between 1 and 11.6 points, according to the latest polls of the current polling universe between 9/24 and 10/15 provided by ICR, the LA Times, 2 from Newsweek, and the latest from Time. However 2-way "presidential candidate" surveys of registered voters are showing a sharp DOWNWARD trend among undecided voters. Bush and Kerry are still tied, but here neither seems to be showing any statistically significant trend, either up or down. Could it be that Cheney did some damage to Bush in his debate with Edwards, but unless people are reminded of Cheney Kerry reaps no benefits?
Hard to say but, whatever is the case, here are the stats for the latest 2-way "ticket" surveys among registered voters:
Final 2-Way 9/24-10/15 (ICR, LA Times, Newsweek, Newsweek, Time)
Bush/Cheney 46.4-49 -1-11.6
Kerry/Edwards 45-46.6
DK 4.4-9
And here are the latest stats for 2-way "presidential candidate" surveys of registered voters, these being releases from Winston and ARG:
Final 2-way 9/22-10/4 (Winston, ARG)
Bush 46-48
Kerry 45-49.1
DK 4-7.1 -.6-10.3
Among 3-way "ticket" surveys, Bush/Cheney now enjoy a narrow lead over Kerry/Edwards, while Undecided is clearly in third place, with Nader bringing up the rear in fourth. The only clear trend-line is among Undecideds, who have climbed somewhere between .4 and 5 percentage points between the 24th and the 15th. Here are the combined stats for the current universe of 3-way "ticket" surveys of registered voters conducted between September 24th and October 15th, consisting of the latest from ICR, the LA Times, Newsweek, CNN, Pew, ABC/Washington Post, CBS/NYT, Marist, and the AP:
Final Composite 9/24-10/15 3-Way (ICR, LA Times, Newsweek, CNN,
Pew, ABC/Post, CBS, Marist, AP, ABC/Post, CNN, Newsweek)
Bush/Cheney 47-49
Kerry/Edwards 44.5
DK 5.4-6 +.4-5
Nader/Camejo 1-3.5
Among 3-way "presidential candidate" surveys of registered voters Bush and Kerry have it all tied up, with the only statistically significant trend apparent among Undecideds, who have trended down between .7 and 9.3 points between the 27th and the 15th. The combined stats for the current universe of 3-way "presidential candidate" surveys conducted between October 2nd and October 15th, released by ARG, and Time (2 from the latter), are as follows:
Final 3-Way 10/2-10/15 (ARG, Time, Time)
Bush 43-47
Kerry 44-46
Nader 1-5
DK 4-6 -.7-9.3
Remember that if you're interested in looking more closely at the methods I'm using to arrive at these conclusions you are cordially invited to visit The Poll Corner feature on my personal home page at http://criggs.home.pipeline.com/pollcorn.htm since I also post these reports at those locations.
LIKELY VOTERS: NATIONAL
The Current Picture As Of October 17th:
3-Way "Ticket" and "Candidate" Surveys Show Kerry/Edwards And Bush/Cheney Tied With Undecided Trending Down
2-Way Surveys, Both Of The "Ticket" and "Candidate" Flavor, Show Bush With A Narrow Lead
The story with likely voters appears slightly different from registered voters. Where registered voters seem to treat Kerry/Edwards most kindly in TWO-WAY surveys, likely voters seem to treat them most kindly in THREE-WAY surveys. There are no differences in the picture from "candidate" surveys versus "ticket surveys". In these 2-way surveys of likely voters there appear to be general agreement that Bush/Cheney continue to lead Kerry/Edwards slightly, and with no clearly discernible statistical trends to give one or the other an edge. There is one possibly significant internal number which is different: Bush/Cheney appear to be trending down among registered voters but among likely voters they appear merely to be stalled.
Among 3-way "ticket" surveys, the current polling universe appears to have emerged some time between the 6th and the 7th, with a discernible statistically significant change downward on the part of Undecideds from then through the 16th. Is it possible that the vice-presidential debate on the 5th played a role in that? To summarize, in the current polling universe of 3-way "ticket" surveys among likely voters, Bush/Cheney and Kerry/Edwards are tied with neither showing a clear upward or downward trend, while Undecided is clearly trending down. Four polls appear to be in agreement on this universe, the two latest ABC/Washington Post surveys, and the latest from CNN and Newsweek. Combining their numbers where they are all in agreement shows the following picture:
Final Composite 10/6-10/16 3-way (ABC/Post, CNN, Newsweek, ABC)
Bush/Cheney 47.5-52
Kerry/Edwards 45-48
Nader/Camejo 0-3.5
DK 1-4 -1-5
Among 3-way "candidate" surveys, the picture is virtually identical. The two candidates have not moved significantly at all lately, but Undecided has trended down between 1.8 and 2 points between the 2nd and the 15th.
This polling universe is represented by polls conducted by ARG, Fox, Zogby, Democracy Corps, and Time between 10/2 and 10/15. Combining their numbers where they are all in agreement shows the following picture:
Final 3-Way Composite 10/2-15 (ARG, Fox, Democracy Corps, Time, Zogby, Zogby,
Time)
Bush 44.9-46.9
Kerry 45.9-48.9
Nader 0-3.9
DK 5.1 -1.8-2
In 2-way "ticket" surveys among likely voters there has been no statistically significant movement on the part of either ticket since the start of my analysis period, which begins with surveys conducted exclusely on or after the 22nd (remember that I ignore polling conducted before Kerry's NYU Iraq speech). Bush/Cheney continue to narrowly lead Kerry/Edwards in this universe, and with no statistically significant movement on the part of Undecided either. This polling universe is represented by polls conducted by CNN, ICR, the LA Times, Zogby, Newsweek and 2 from Time, all conducted between 9/24 and 10/15. Since all of these polls are in agreement, once one takes into account each poll's margin of error, here are the results of combining their numbers over the ranges in which they all agree:
Final Composite 2-way 9/24-10/15 (CNN, ICR, LA Times, Zogby, Time, Newsweek,
Time)
Bush/Cheney 48-49.1
Kerry/Edwards 44-46.6
Dk 5.9-6
2-way "candidate" surveys among likely voters show precisely the same picture, with Bush narrowly leading Kerry and no statistically discernible difference among the undecideds. This polling universe is represented by polls conducted by TIPP, GWU, Rasmussen, ARG, Fox between 9/22 and 10/4. Since all of these polls are in agreement, once one takes into account each poll's margin of error, here are the results of combining their numbers over the ranges in which they all agree:
Final 2-Way Composite 9/22-10/4 (TIPP, GWU, Rasmussen, ARG, Fox)
Bush 48.9-49
Kerry 45-47.1
DK 5-7
The bottom line is that among likely voters our guy does best in 3-way surveys (Bush/Cheney versus Kerry/Edwards versus Nader/Camejo), where he ties Bush/Cheney with no clearly discernible or statistically significant trends up or down from either of them, but with Undecided clearly trending down. He clearly trails Bush with no upward momentum among 2-way surveys that don't mention Nader. It is almost as if the presence of Nader acts as a goad to likely voters to support Kerry! Rather odd.
Remember if you're interested in looking more closely at the methods I'm using to arrive at these conclusions you are cordially invited to visit The Poll Corner feature on my personal home page at http://criggs.home.pipeline.com/pollcorn.htm since I also post these reports at those locations.
REGISTERED VOTERS: SWING STATES
The Current Picture As Of October 3rd:
Florida, Ohio, And Pennsylvania Show It All Tied Up In Single 3-Way "Candidate" Surveys. Michigan 3-Way "Ticket" Survey Shows Similar Tie There
2-Way "Candidate" Surveys in Arizona and Ohio Show Bush Edge. Pennsylvania "Candidate" Survey Shows Tie
Among both registered and likely voters in statewide polls released since the NYU speech, only Arizona, Georgia, and North Carolina show a clear leader across the board, and that is Bush or Bush/Cheney. All other states polled come out tied in one way or another.
Florida's 3-way survey comes from CNN, and here are those numbers:
Florida 9/24-27 CNN +/-4 Final 3-way
Bush 49, 45-53
Kerry 44, 40-48
DK 5, 1-9
Nader 2, 0-6
Ohio's 3-way survey also comes from CNN, and here are those results:
Ohio CNN 9/25-8 +/-4 Final 3-way
Kerry 49, 45-53
Bush 46, 42-50
DK 3, 0-7
Nader 2, 0-6
Pennsylvania's 3-way "candidate" survey comes from Quinnipiac, and here are their numbers:
Pennsylvania (Quinnipiac) 9/22-6 3 pts. Final 3-way
Kerry 46, 43-49
Bush 42, 39-45
DK 8, 5-11
Nader 4, 1-7
The "ticket" survey of Michigan comes from the Detroit Free Press, and here are their results:
Michigan Detroit FP 9/22-28 +/-3.5 Final 3-way
Kerry/Edwards 48, 44.5-51.5
Bush/Cheney 46, 42.5-49.5
DK 4, 0.5-7.5
Nader/Camejo 2, 0-5.5
And finally here are the results in the 2-way "candidate" surveys for Arizona, Ohio, and Pennsylvania:
Arizona ASU 9/23-26 4 pts. Final 2-way
Bush 49, 45-53
Kerry 38, 34-42
DK 12, 8-16
Dispatch 9/22-10/1 +/-2
Bush 51, 49-53
Kerry 44, 42-46
DK 5, 3-7
Pennsylvania (Quinnipiac) 9/22-6 3 pts. Final 2-way
Kerry 46, 43-49
Bush 41, 38-44
DK 13, 10-16
Remember that if you're interested in looking more closely at the methods I'm
using to arrive at these conclusions you are cordially invited to visit The
Poll Corner feature on my personal home page at http://criggs.home.pipeline.com/pollcorn.htm
since I also post these reports at those locations.
LIKELY VOTERS: SWING STATES
The Current Picture As Of October 3rd:
3-Way "Ticket" Polls Show Tie In Iowa, Bush/Cheney On The Decline In Michigan, A Tie In Minnesota, New Jersey, A Narrow Lead For Bush/Cheney in Ohio, And A Tie, Yes, A Tie(!) In Virginia. 3-Way "Candidate" Surveys Show A Tie In Connecticut, A Slight Edge For Bush In Florida, A Clear Lead For Him In Georgia, And A Tie In Maine And Ohio
2-Way "Ticket" Polls Show A Tie In Iowa, Michigan, Minnesota, New Jersey, An Edge For Bush/Cheney In North Carolina And Ohio. 2-Way "Candidate" Polls Show Bush Leading In Arizona, Georgia, A Tie In New Jersey, Bush With An Edge In North Carolina, And All Tied Up In Pennsylvania
Among both registered and likely voters in statewide polls released since the NYU speech, only Arizona, Georgia, and North Carolina show a clear leader across the board, and that is Bush or Bush/Cheney. All other states polled come out tied in one way or another.
Here are the 3-way "ticket" polls for Iowa, Michigan, Minnesota, New Jersey, Ohio, and Virginia:
Iowa Strategic Vision 9/27-29 +/-3 Final 2-way
Bush/Cheney 49, 46-52 Final 3-way
Kerry/Edwards 44, 41-47
DK 5, 2-8
Nader/Camejo 2, 0-5
Michigan Strategic Vision 9/26-28 +/-3 Final 3-way
Kerry/Edwards 48, 45-51
Bush/Cheney 42, 39-45 -1.5-14.5
DK 9, 6-12
Nader/Camejo 1, 0-4
Minnesota Strategic Vision 9/26-8 +/-3 Final 2-way
Kerry/Edwards 48, 45-51
Bush/Cheney 45, 42-48 Final 3-way
DK 5, 2-8
Nader/Camejo 2, 0-5
New Jersey Strategic Vision 9/24-6 3pts. Final 3-way
Bush/Cheney 44, 41-47
Kerry/Edwards 44, 41-47
DK 9, 6-12
Nader/Camejo 3, 0-6
Ohio Strategic Vision 9/24-26 3 pts. Final 3-way
Bush/Cheney 52, 49-55
Kerry/Edwards 42, 39-45
DK 5, 2-8
Nader/Camejo 1, 0-4
Virginia Mason/Dixon 9/24-27 +/- 4 Final 3-way
Bush/Cheney 49, 45-53
Kerry/Edwards 43, 39-47
DK 7, 3-11
Nader/Camejo 1, 0-5
In the above list note particularly Bush/Cheney's number in the Michigan poll. Bush/Cheney's Michigan support is now down somewhere 1.5 and 14.5 points as compared to the previous polling universe of 9/22-8 reported in the Detroit Free Press.
Here are the 3-way "candidate" surveys for Connecticut, Florida, Georgia, Maine and Ohio:
Connecticut Quinnipiac 9/26-28 +/-3.5 Final 3-way
Kerry 50, 46.5-53.5
Bush 44, 40.5-47.5
DK 4, 0.5-7.5
Nader 2, 0-5.5
Florida 9/24-27 CNN +/-4 Final 3-way
Bush 52, 48-56
Kerry 43, 39-47
DK 4, 0-8
Nader 1, 0-5
Georgia Strategic Vision 9/2527 +/-3.5 Final 3-way
Bush 58, 54.5-61.5 Final 3-way
Kerry 33, 29.5-36.5
DK 8, 4.5-11.5
Nader 1, 0-4.5
Maine Strategic Marketing 9/23-7 +/-5 Final 3-way
Kerry 42, 37-47
Bush 39, 34-44
DK 15, 10-20
Nader 4, 0-9
Ohio CNN 9/25-8 +/-4 Final 3-way
Bush 49, 45-53
Kerry 47, 43-51
DK 3, 0-7
Nader 1, 0-5
Here are the results of the 2-way "ticket" surveys for Iowa, Michigan, Minnesota, New Jersey, North Carolina, and Ohio:
Iowa Strategic Vision 9/27-29 +/-3 Final 2-way
Bush/Cheney 49, 46-52
Kerry/Edwards 45, 42-48
DK 6, 3-9
Michigan Strategic Vision 9/26-28 +/-3 Final 2-way
Kerry/Edwards 49, 46-52
Bush/Cheney 46, 43-49
DK 9, 6-12
Minnesota Strategic Vision 9/26-8 +/-3 Final 2-way
Kerry/Edwards 49, 46-52
Bush/Cheney 45, 42-48
DK 6, 3-9
New Jersey Strategic Vision 9/24-26 3 pts.Final 2-way
Kerry/Edwards 45, 42-48
Bush/Cheney 44, 41-47
DK 11, 8-14
North Carolina Public Opinion Strategies 9/26-7 4 pts. Final 2-way
Bush/Cheney 53, 49-57
Kerry/Edwards 41, 37-45
DK 6, 2-10
Ohio Strategic Vision 9/24-6 3 pts. Final 2-way
Bush/Cheney 52, 49-55
Kerry/Edwards 43, 40-46
DK 5, 2-8
And finally here are the two-way "candidate" surveys for Arizona, Georgia, New Jersey, North Carolina, and Pennsylvania:
Arizona ASU 9/23-26 4 pts. Final 2-way
Bush 53, 49-57
Kerry 38, 34-42
DK 8, 4-12
Georgia Strategic Vision 9/25-27 +/-3.5 Final 2-way
Bush 58, 54.5-61.5
Kerry 34, 30.5-37.5
DK 8, 4.5-11.5
New Jersey Fairleigh 9/23-8 +/-4.5 Final 2-way
Kerry 45, 40.5-49.5
Bush 44, 39.5-48.5
DK 12, 7.5-16.5
North Carolina Mason-Dixon 9/26-28 +/-4 Final 2-way
Bush 52, 48-56
Kerry 43, 39-47
DK 4, 0-8
Pennsylvania Final Composite 9/22-29
Kerry 45.5-49
Bush 45-49.5
DK 3-8.5
Note that the last poll figures, from Pennsylvania are a composite/narrowing/triangulation of the numbers from two polls that agree with each other, namely Quinnipiac and Strategic Vision.
Remember that if you're interested in looking more closely at the methods I'm using to arrive at these conclusions you are cordially invited to visit The Poll Corner feature on my personal home page at http://criggs.home.pipeline.com/pollcorn.htm since I also post these reports at those locations.
I have had occasion before in these Musings to remark that most reports in the media do not accurately report the information that can be gleaned from opinion polls. The problem lies in the way the margin of error is reported. In reality, the margin of error is not some extra trivial statistical fact accompanying the poll. IT IS CENTRAL TO THE INFORMATION BEING UNCOVERED IN THE SURVEY.
Let me give you a hypothetical example of why this is so. Let us examine two hypothetical polls, one with a margin of error of +/-2 points, the other with a margin of error of +/-5 points. And here are the figures for those polls as they typically would be reported by the average newspaper or TV broadcast:
Poll A +/-2 Poll B +/-5
Forbes 25 Forbes 25
Dole 20 Dole 20
Buchanan 15 Buchanan 15
At first glance, it would appear that these two polls are telling precisely the same story. But in fact nothing could be further from the truth. This is because when one reports on the figures from a poll in the preceding manner, ONE IS NOT REPORTING THEM ACCURATELY. Figures in a poll should accurately be reported as follows, WITH THE MARGIN OF ERROR ABSORBED INTO THE HARD NUMBERS THEMSELVES, THROUGH A POINT-SPREAD METHOD OF REPORTING:
Poll A +/-2 Poll B +/-5
Forbes 25,23-27 Forbes 25,20-30
Dole 20,18-22 Dole 20,15-25
Buchanan 15,13-17 Buchanan 15,10-20
When we finally have a chance to read the figures reported in this manner, it is apparent that these two polls are telling a very different story. Poll A is telling us that Forbes is clearly in the lead, Dole is clearly in second place, and Buchanan clearly brings up the rear. Poll B, on the other hand, is telling us that ALL THREE CANDIDATES MIGHT BE LOCATED AT OR NEAR 20%; in other words, Poll B is reporting a very tight, three-way race between the three candidates.
Similarly, one should be very cautious when one sees or reads a report claiming that 'recent opinion polls are all over the map and contradict each other.' Bear in mind that most statewide primary polls have a margin of error of +/- 5, which means that the sample size is around 400 people. This means that all such polls are in fact stating in the instance of any given candidate that that candidate currently enjoys the support of somewhere between X-5% and X+5% of the electorate. If one takes this factor into account, it quickly becomes apparent in lots of cases that two polls which appear on the surface to be contradicting each other in fact do no such thing. Yet too often in such cases the news media makes much of these insignificant differences and jumps through hoops in vain attempts to account for such apparent contradictions. Much is made of the fact, for example, that one poll may only conduct polling every other Tuesday, while the other poll may make a point of including a representative sample of left-handed Eskimos, that one poll may only ask an undecided voter once how he or she may be leaning, while the other one may threaten their credit rating in order to get a decided response, etc. etc. But of course if one is dealing with a distinction without a difference between two polls, the need for an explanation of that non-existent difference evaporates. In other words, if two polls agree with each other, once the margin of error is taken into account, that fact is more an indication of the probable accuracy of BOTH polls, than of the relative significance of the differing methodologies which two such given polls may have utilized. In fact, such agreement, within a poll's margin of error, serves to minimize the IMPORTANCE of such differences in methodologies. In many cases, in fact, individual pollsters may erroneously CLAIM for such reasons that their polls cannot be compared to other polls taken at the same time; but those reasons emerge as entirely bogus when the supportive figures from other polls are laid side by side with that pollster's efforts, and shown to be easily duplicated, even when slightly different methodologies may be employed.
Now let's take a look at two sample opinion polls, which do not differ in their margin of error, but differ in the time when they were taken:
Poll C +/-5, Jan. 1-4 Poll D +/-5, Jan. 15-19
Forbes 20 Forbes 30
Dole 25 Dole 15
Buchanan 10 Buchanan 20
I am confident that in the average newspaper or TV report, we would hear that Poll D showed clear and striking movement as compared to the earlier Poll C. Much would be made of Dole's precipitate collapse from first to third position, of Forbes' strong first position, and of Buchanan's meteoric rise to second place. IN FACT, THESE TWO POLLS ARE TELLING US NONE OF THESE THINGS. This becomes apparent when one utilizes once again the point-spread method to report the figures for these polls:
Poll C +/-5, Jan. 1-4 Poll D +/-5, Jan. 15-19
Forbes 20,15-25 Forbes 30,25-35
Dole 25,20-30 Dole 15,10-20
Buchanan 10,5-15 Buchanan 20,15-25
Looked at in this light, it no longer appears that these two polls DISAGREE. It becomes apparent that they AGREE. Taken together, they tell us that Forbes is probably right at 25%, Dole is right at 20%, and Buchanan is right at 15%. Most important to bear in mind is this: the rather accidental single figures that each poll came up with are not as important as the probability range which those single figures give us when combined with the margin of error. Equally important to bear in mind is the fact that if two polls differ only within their margin of error, it is more likely that the variations are caused by random fluctuations within the polling sample than that there has been any change IN THE REALITY OF THEIR RELATIVE SUPPORT WITHIN THE ELECTORATE. Consequently, as in the example given above, where two polls disagree only WITHIN the margin of error, it is reasonable to TRIANGULATE a candidate's level of support. Hence the conclusion that Forbes is at 25%, Dole is at 20%, and Buchanan is at 15%.
As a further example, let us apply these reporting principles to some recent New Hampshire opinion polls.
Let's look at Dole's and Forbes' standings in two polls:
WMUR 1/29-2/4+/-4.7 NY POST/WNYN 2/1-4 +/-4.5
Forbes 29,24.3-33.7 23,18.5-27.5
Dole 25,20.3-29.7 22,17.5-26.5
These two polls agree with each other, once the margin of error is taken into account. Therefore I make the assumption that no candidate has moved appreciably between 1/29, the starting date of WMUR, and 2/1-4, the dates for WNYN. Therefore, I go on to assume that BOTH polls are describing the same set of circumstances as best as they are able, given the limiting effect of their respective margins of error. Therefore, I conclude that triangulation is possible on this basis in order to narrow further the probable range of each candidate's support. The result is as follows:
Reps/Inds 1/31-2/5 Final Composite
Forbes 24.3-27.5
Dole 20.3-26.5
In addition, I have gone back in time to the last poll which DISAGREES with the preceding collection of polls (that is to say, gives us numbers which differ from the above crop sufficiently to be OUTSIDE the margin of error) and displayed the range of possible change from that old poll to the above polls. Here's an example:
WMUR 1/7-10 +/-4.7 Reps/Inds 1/31-2/5 Final Composite
Forbes 18,13.3-22.7 24.3-27.5+1.6-14.2
In other words, I have concluded that between 1/7-10 and 1/31-2/5 Forbes has moved up anywhere from 1.6 to 14.2 points.
Hopefully, my ground rules are now clear. So, applying those ground rules, here follow the results of my analysis. First I display the last poll of the combined Republican/Independent universe which DISAGREES with the later crop of those polls, followed by my final composite of the figures for that later crop.
WMUR 1/7-10 +/-4.7 Reps/Inds 1/31-2/6 Final Composite
Forbes 18,13.3-22.7 24.3-27.5+1.6-14.2 1.
Dole 37,32.3-41.7 20.3-26.5-5.8-21.4 2.
DK 12,7.3-16.7 18.5-16.7+1.8-9.4 3.
Buchanan 10,5.3-14.7 7.3-13.5 4.
Alexander 7,2.3-11.7 5.5-11 4.
Gramm 5,0.3-9.7 2.5-9.5 4.
Lugar 2,0-6.7 0-6.5 5.
Taylor 0,0-4.7 0-5.5 5.
Keyes 3,0-7.7 0-4.7 5.
Dornan 0,0-4.7 0-4.5 5.
On the basis of the preceding figures, I conclude that, while Forbes and Dole may overlap each other between 24.3 and 26.5, Forbes nevertheless has the edge because he has trended upward OUTSIDE OF THE MARGIN OF ERROR since WMUR's 1/7-10 survey, whereas Dole has trended downward OUTSIDE OF THE MARGIN OF ERROR since that survey. The Don't Know category seems clearly positioned by itself statistically in third place, while Buchanan, Alexander, and Gramm are all tied for fourth place within their combined overlapping point spreads.
As the campaign proceeds, I will maintain this up-to-date poll section on the presidential race utlizing the methodology described above. It is, I believe, not only a more accurate mode of survey interpretation, but also a more interesting and even dramatic one in its interpretational results than the traditional, tired, and innacurate reporting method.
As always, your comments and reactions are welcome. Please write me at
criggs@pipeline.com