Pollster.com

Articles and Analysis

 

Florida Republican Poll Errors

Topics: 2008 , The 2008 Race

1FLPollErrorRep12131.png

Polling had a pretty good night in Florida on Tuesday. While early polls understated both McCain and Romney's vote, the polls got better as election day closed in, reflecting a (measured) upward trend in both candidates support. By the last day or two of polling most polls were within the ten-ring, and a number were close to the five-ring.

There was more disagreement among the polls as to who was ahead, and this included some late polls that put Romney ahead of McCain. However, most of these polls reflected a race "too-close-to-call", rather than egregious errors about the leader, with one large exception towards the lower right in the plot.

As we've seen all year, the polls got the vote for third and fourth place finishers very accurately, almost all within the five-ring.

2FLPollErrorRep34131.png

 

Comments
Sean:

Just like earlier primaries, there seems to be a last minute shift towards the winner. Here the shift was small compared to the Dems in South Carolina or New Hampshire, but every poll underestimated McCain's final tally. In all of the exit polls I have seen, McCain and Romney split voters that made up their minds at the last minute. So where are these late swing votes coming from?

____________________

jb:

A similar comment from me as with the SC chart.

These graphs would be more meaningful if undecided were not counted. Otherwise there will always be a systematic undercount of polls vs actually votes in every election.

For example if a poll has 32%Mccain 30%Romney but has 9% undecided: I would prefer to see that graphed as 32%/(1-9%) for Mccain and 30%/(1-9%) for Romney.

This would yield 36% Mccain 33% Romney.

What value is there to repeatedly showing "error" that only occurs because of the undecided percentage.

In fact I'd argue that with significant undecideds an accurate pollster won't be at the exact center of your chart. Mathematically, the only way a pollster would achieve the bull's eye on Romney-Mccain (assuming some undecideds) would be to underestimate one of the other candidates,

Thus these charts are flawed as they imply that being at the center is an achievement for the pollsters.

By ignoring undecideds every pollster seems more accurate but some improve much more than others and move much closer to the center

Florida Example:
Zogby
35M (actual 36)
31R (actual 31)
13G (actual 15)
13H (actual 14)
2P (actual 3)
6UNDEC

Zogby appears to be much more accurate than Suffolk (see below). Zogby would appear much closer to the center on your chart than Suffolk.

Suffolk
M30 (actual 36)
R27 (actual 31)
G13 (actual 15)
H11 (actual 14)
P4 (actual 3)
UNDEC 15


However I would argue that Suffolk is really almost as the good as Zogby as they are very close on the proportions of final vote of those who expressed a preference (just as close as Zogby in almost every case).


If you ignore undecideds, Zogby becomes
37M (actual 36)
33R (actual 31)
14G (actual 15)
14H (actual 14)
2P (actual 3)


If you ignore undecideds Suffolk becomes

Suffolk
M35 (actual 36)
R32 (actual 31)
G15 (actual 15)
H13 (actual 14)
P5 (actual 3)

Thus I'd like to see Suffolk near the center of your chart instead of out in the "inaccurate" zone

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR