A couple of days ago, I wrote a diary explaining RBS style polling, and how I thought the high level of seniors in the recent Datamar Florida poll might explain Edwards' strength in that poll.
I polled Florida last night, using the voter file to identify previous voters. I had FloridaVoterFile.com pull two samples, one of Democratic primary voters and a second of registered Democrats who vote in general elections. Note that Florida holds a primary for other offices separately from the presidential preference primary and that the normal primary has double the turnout of the presidential primary, which is normally held too late to have much impact. The early presidential preference primary will likely see heavier turnout than usual, but the increased turnout will likely come from past voters. First time voters are typically a very small percentage of the vote, and they generally do not differ from other voters by enough to have significant impact unless the race is a real squeaker.
Datamar's demographics were very close to past elections, and with the high senior vote and other polling showing Edwards with strength among seniors, their results seemed plausible. Calling randomly generated numbers is unlikely to get such a high percentage of seniors, so maybe RDD pollsters were underweighting this group.
My polling did not confirm Datamar's results. In particular, Florida seniors went strongly for Clinton and did not give Edwards significantly different results than other age groups. In trying to reconcile the two polls, I couldn't find any way to recreate their numbers. Edwards did not lead in any region, age group, sex, ethnic group or anything else I analyzed. Clinton had a healthy lead in most categories, though Obama had a large lead among voters under 40.
There were differences between the regular primary voters and the general voters. Clinton was over 50% in regular primary voters, with Edwards 6 points ahead of Obama. Among general voters, Clinton was under 40%. Obama was 12 points back in second, but still 15 points ahead of Edwards. African-American primary voters went for Clinton three to one over Obama, but African-American general voters went for Obama three to one over Clinton. Edwards was tied for second in each case.
The first followup question was "Assuming the candidate you indicated hasn't dropped out before the primary, are you definitely going to vote for them, or are you still considering other candidates?" Clinton and Obama both received 70% 'definite' while Edwards received 55%.
The second followup was "If Al Gore entered the race, would you definitely change your vote, possibly change your vote or definitely stay with your current choice?" Gore would take second place based on 20% saying they would definitely vote for him. Another 34% would 'possibly' change their vote. One in four Edwards supporters and one in five Obama supporters would definitely switch, but only about one in fifteen Clinton supporters would definitely move to Gore. About one in three of each top candidate's supporters would 'possibly' switch.
487 surveyed, 4.4% MoE for the blended population, subgroups have higher MoE. Weighted by sex, age group and ethnicity to match historical results. Graphic at IVRPolls.com with additional analysis when I get around to it.
Primary population - Clinton 51%, Obama 11%, Edwards 17%
General population - Clinton 38%, Obama 26%, Edwards 11%
Blended population - Clinton 45%, Obama 18%, Edwards 14%
I think there is a good chance that the actual results will be closer to the pure primary results as that already has significantly higher turnout than past presidential preference primaries.
Other May polls for comparison:
Datamar - Edwards 26%, Clinton 24%, Obama 19%
Strategic Vision - Clinton 37%, Obama 20%, Edwards 19%
Schroth, Eldon & Assoc - Clinton 42%, Obama 19%, Edwards 12%
ARG - Clinton 45%, Obama 17%, Edwards 15%
|
|
|
Permalink :: 76 Comments :: Post a Comment
|
In order to post a comment, you must be logged in. If you have a member account, please log in to comment.
If not, you can make an account right here. It's quick and free.
