Sun pollster to take questions tomorrow
Got questions about The Baltimore Sun's new poll? OpinionWorks President Steve Raabe will participate in a live chat at noon Tuesday on this blog.
If you can't make it but would like to submit a question, please comment on this post. You can also leave a question tomorrow morning, in advance of the chat. Raabe has been a pollster for 25 years and founded Annapolis-based OpinionWorks in 2001.
Sun poll methodology
The Baltimore Sun telephone survey of 798 likely voters was conducted Oct. 15-20. The Sun's pollster, OpinionWorks of Annapolis, used a Maryland Board of Elections database to identify registered voters with a history of voting in gubernatorial elections or who had registered to vote since the last election, and obtained survey results from those who ranked themselves seven or higher on 1-to-10 scale of their likelihood to vote.
The Sun's sample was designed to approximate the racial, gender, geographic, partisan and age breakdown of the state's voting population as a whole, based on turnout patterns averaged over the last four Maryland general elections. Results were weighted to reflect a higher-than-average Republican turnout this year, and slightly lower African-American participation than in recent elections. The margin of error for questions that reflect the entire sample is 3.5 percentage points, which means that in 95 times out of 100, the actual answer obtained by surveying every Maryland voter would be within 3.5 percentage points of the answer obtained by using the sample. For questions about Anne Arundel voters and slots, the sample size was 422 voters, with an error rate of 4.8 percentage points.
Comments
The first methodological description stated that a sample drawn from registered voter lists was matched by a commercial firm to obtain telephone numbers. What was the actual pecentage of the list that was successfully matched? How were voters who have only cell phones, not land lines, handled?
Did the sample of 798 contain the 422 Anne Arundel voters? In other words was Anne Arundel overweighted at the expense of other counties?
Some of the weighting adjustments seem rather subjective. Why not just rely on the likely voter screen and only weight by factors that have real data, such as voter turnout by county?
Posted by: C. Welsh | October 25, 2010 3:49 PM
I just finished voting early at this site:Honeygo Community Center
9033 Honeygo Blvd
Perry Hall, MD 21128
I am very disturbed with there not being open available parking for voters due to "poll workers" or "Tea Party Rally" attendees. Poor planning by the Baltimore election committee ... or whomever polices this issue. All spaces except 4 were full and other vehicles were parked up on the grass close to the curb. The 4 spaces vacant were taken up by a few people in chairs with orange cones in the spaces. It doesn't matter what your party, having available parking for "VOTERS" is a must.
Though the voting lines were moving well without problems ... I saw more than one car come in and leave without finding a place park.
It'd be nice to have a reporter go out and ask "some questions" to the poll workers. do you think?
Posted by: Perry Evans | October 25, 2010 4:35 PM
The last two polls showed Harris by 11 (less than a week ago) and Harris by 3 (a month ago). Has there really been a groundswell shift to Kratovil?
Also, what voter model did you use to indicate increased or decreased turnout among partisans (Ds/Is/Rs), African-Americans, and off-presidential year? Aka, were you sampling more voters with a history of voting in gubernatorial years (roughly 20% less of the total electorate in most cases) or were you using some sort of hybrid? It would appear to me that the big difference is in the sampling methodology choices that leads to the varied outcomes.
Posted by: John | October 25, 2010 5:46 PM