|
The latest Gallup poll which shows a 22 point Bush lead is probably fairly accurate. It is pretty well in line with other recent polls. The Harris poll (conducted for CNN/Time) that had Bush leading Dean by only 5 points is the oddball and likely not reflective of reality. I have been calling the Harris poll bogus since I first saw it. I can explain why I have thought so, but you need to understand a couple of things about how polls work:
First, the poll showing that Dean is close to Bush has a margin of error of + or - 4%. This means that even according to that poll Dean's support may be as high as 50% or as low as 42%, and Bush's support may be as high as 55% or as low as 47%. So, even by that (questionable) poll, Bush could have a lead as large as 13%.
Second, even the rather broad spectrum of possible results in the Harris poll (like all scientifically conducted polls) has a 95% likelihood of being accurate. To understand what this means, think about a basketball player who has a 95% probability of making a freethrow--he misses one out of every twenty times. So, while 19 out of 20 such polls do capture real public attitudes somewhere within their margins of error, 1 out 20 polls simply miss the mark. If the Harris poll that shows Dean only 5 points behind Bush is one the few polls that is just wrong (i.e. which fails to capture reality within its margin of error), then Dean must in actuality have less than 42% support (I am assuming he can't possibly have more than 50%). To know whether the Harris poll is the 1 out of 20 polls that is just plain wrong, we need to compare it to similar polls. Other polls asking the same question in the last few weeks have shown Dean with support in the low 40s or high 30s, and Bush with a 14%-22% lead (+ or - some margin of error--usually between 3% and 5%, depending on the particular poll's sample size). The recent Gallup poll is just the latest example of polling consensus that Bush has a big lead. The Harris poll that makes Dean look so "electable" therefore is likely incorrect.
Third, we then are left with two possibilities. (1) Something happened in the last week of 2003 to give Dean a huge boost that no other polls detected except the Harris poll; or (2) the Harris poll is simply one of the few wrong ones. Since I can't think of any reason to think (1) is true, my guess is that (2) is right. In fact, there is a good reason to believe that (2) is true. Scientific phone polling is becoming increasingly difficult and inaccurate as more and more people are not home, screen calls, refuse to answer polls, etc. These factors mean that the people who do answer the polls are not truly representative of the overall population. And all these problems become more extreme during a holiday. Holiday polls always yield questionable results--particularly since Republicans are more likely to be on vacation and less likely to be home than are Democrats. The Harris poll was conducted Dec 30th - Jan 1st--that alone makes its results questionable.
The latest Gallup poll, which seems to be fairly in line with other recent polls, is certainly much more accurate than the Harris poll done for CNN/Time. Bush's lead over Dean is much more likely to be 20+ points rather than only 5.
But this should not dishearten Dean supporters. Several times in recent presidential elections candidates have trailed there opponents by 20+ points and come back to win. What Dean supporters should fear is U.S success in Iraq. If troops are coming home and democracy looks to be taking root in Iraq in November, Dean will be steamrolled by Bush. Kind of sad, isn't it?
|
Printer-friendly format
Email this thread to a friend
Bookmark this thread
(1000+ posts)
Send PM |
Profile |
Ignore
