Thursday, Oct 28, 2004
Fabrizio-McLaughlin massaged some poll numbers in an interesting way this week.
The polling firm surveyed 800 likely voters in 12 battleground states earlier this week. The respondent selection, “was at random within predetermined geographic units to reflect actual electoral vote allotment to each state.”
The 12 states selected were: CO, FL, IA, ME, MI, MN, NH, NM, NV, OH, PA, WI.
When they totaled everything up, President Bush edged Sen. Kerry by two-tenths of a percentage point.
However, when the firm weighted the results to reflect 2000 presidential exit polls Kerry was ahead by 3.5 points. The firm also weighted to reflect Census results and Kerry was ahead by 5.2 percent. The turnout result is the more important set of numbers, I think, because it reflects past voting patterns and not just simple demographic makeup. To the charts…
First, the “raw” trend lines since the summer. (Also notice how Ralph Nader’s support is dropping):
Now, the latest battleground poll with the turnout and Census weighting:
And here’s the chart that shows the difference between the original, raw data and the exit poll and Census modified numbers.
I talked with Tony Fabrizio this afternoon. I wondered whether the exit poll and Census numbers were for those 12 states, and he said they were. Fabrizio also agrees that the exit poll numbers are the more valid guage of what may happen next week.
Does this mean that Bush is cooked? Not according to Fabrizio. What’s really needed is a state-by-state survey with the FabMac massage. Fabrizio thinks that Bush might pick up some of the states listed above that Al Gore won four years ago, specifically, those with low numbers of minorities.