On Election Day 2008, CNN’s exit poll showed a Maryland electorate that consisted of 51 percent self-identified Democrats, 28 percent Republicans, and 21 percent independents.
Today, Public Policy Polling unveiled a poll showing Democratic governor Martin O’Malley leading Republican Robert Ehrlich by 3 percentage points. The poll’s sample splits 60 percent Democrat, 28 percent Republican, 12 percent independents.
As I noted in yesterday’s post, African-Americans made up 25 percent of the electorate in 2008, and 23 percent in 2006 and 2004. The PPP sample presumes that proportionally, African-Americans make up the same share of the 2010 electorate as they did in 2008.
Either this sort of thing bugs you or it doesn’t. Clearly, using really heavily Democratic samples doesn’t bother the guys at PPP.
While the assumptions may bug you, give PPP credit for putting thier numbers out there for others to crunch.
Playing with said numbers, if the percentages of Democrats, Republicans and independents are set equal to the numbers from 2008 (an assumption most would consider optimistic for Democrats at this point) and the preferences within groups are kept the same, the 45-42 O'Malley lead turns into a 44.5% to 41.4.% O'Malley _deficit_. I'd say those numbers look pretty favorable for Ehrlich.
Reply to this commentLinkReport AbuseYou should be able to get a pretty exact figure for 2008 Maryland voters given that Maryland has voters register by party. So you should have access to data giving you the exact split between registered Democrats, Republicans, and Independents.
Reply to this commentLinkReport AbuseJim -
You're conflating self-identifying Democrats in a 2008 exit poll with registered Democrats on today's voter roll.
In a state like Maryland where many people register Democratic simply to participate in Democratic primaries, it's not inconceivable that a 60 percent registered Democratic sample would self-identify 56 percent Democratic or even lower.
I didn't see demographics on PPP's poll report, but I'll go back and look.
- Lebowitz
Reply to this commentLinkReport AbuseEr, no, Lebowitz, I am not "conflating self-identifying Democrats in a 2008 exit poll with registered Democrats on today's voter roll." The PPP poll does not ask voters what party they are registered as; it asks, "If you are a Democrat, press 1. If you are a Republican, press 2. If you are an independent or identify with another party, press 3."
The question says nothing about party registration; it specifically says, "identify with".
I think the fact that you didn't bother looking it up before declaring that I had it wrong reveals a great deal.
Reply to this commentLinkReport AbuseJim - The 60 percent figure that you quoted was for self-identifying Democrats among the sample. You did not conflate self-identifying Democrats with registered Democrats as I said. I apologize.
Since 2007, the FAQ on PPP's blog has said about sampling,
"our samples are usually based on the voter registration database of a given state."
To find out more I called PPP's Tom Jensen, whose phone number was on the press release. He told me PPP's samples are of registered voters who participated in one or more of the past three general elections, which I consider a more accurate way to screen likely voters than self-reporting (ie; "press one if you are very likely to vote in the upcoming election, press two if you are somewhat likely....")
According to their FAQ, PPP pre-screens for demographics--age, gender, party affiliation, etc.--and for likely voters--from voter rolls.
Magellan, on the other hand, says they pre-screen from voter rolls only for registered voters and then post-screens or weighs for demographics and likely voters based on responses (ie; self-identifying).
We can debate over which method is more accurate straight through the first Tuesday after the first Monday in November, at which point one of us will be wrong and the other will be right.
- Lebowitz
Reply to this commentLinkReport Abuse