Friday Strategists Sextet
Janet Elder's "On Polling "column at the Grey Lady has a cautionary comment on the pitfalls of recent horse-race poll reporting that should be of interest to readers and reporters alike, "When a Poll Changes the Way People Think About the Race."
Also at The Times, Paul Krugman's "Mandates and Mudslinging" column makes a persuasive case that Senator Obama needs to tweak his health care plan to make everyone pay into it and cover everyone and resist the temptation to use Republican arguments against his Democratic rivals.
Bob Moses has an Alternet post addressing a critical question for Democratic strategy "Have Democrats Already Lost Florida?"
The purpose-driven Hillary courts moderate evangelicals, and apparently shows how it's done, according to Carla Marinucci's San Francisco Chronicle report.
Chris Bowers has a pair of interesting articles up at Open Left, a report on a new poll of the Lesbian, Gay and Bisexual constituency's political views and a thoughtful look at endangered Democratic Senate seats.
Comments
Janet Elder's column has a glaring statistical error. She says that when a poll says 40 percent and has a margin of error of +-3% that means that 19 times out of 20 the number will be between 37 and 43 if you poll all Americans. This is wrong, wrong, wrong. Go to wikipedia and read about confidence intervals. We don't know the actual level of support. If the actual level of support is 40 percent then 19 times out of 20 the number will be between 37 and 43. But like I said we don't know the actual level of support. Moreover you can do statistics to determine how likely it is that a candidate is ahead of the other. If Obama has 30 and Clinton has 26 and you ignore sample bias then in fact it is more likely that he is ahead then she is. I wouldn't rant here but the damn NYTIMES has no comments section.
Posted by: Craig![[TypeKey Profile Page]](http://www.thedemocraticstrategist.org/strategist/nav-commenters.gif)
|
November 30, 2007 11:23 PM