The New Republic

Mike, How accurate has polling in Iowa ever been for predicting caucus results? How on earth can you determine how the caucuses will turn out just by calling people and asking who they support, (let alone how can you know in September)? I am amazed and disappointed at how you have so often and thoroughly joined the pack of lazy writers and chatters who over-interpret poll numbers. Early candidate preference polls tell us mostly about name recognition and a bit about general feelings for/against a candidate. And they do a poor job at screening for "likely voters" or caucus-goers. Except in the week before a vote (as opposed to a caucus), they are usually poor predictors of electoral outcomes, whether in the primary or general.

- stgla

September 16, 2007 at 9:36pm

Give the guy a break, stgla; he has to write about something. Unless you want four months of "it's still too far away to know" plank posts. That said: yeah, this means practically nothing.

- ratnerstar

September 17, 2007 at 9:55am

Interestingly enough, if you look at "Iowa Poll" from the Des Moines Register in 2003/2004, they did a pretty good job keeping track of the race: http://miva.dmregister.com/miva/cgi-bin/miva?extra s/iowapoll/poll.mv+file=prez0401 they captured the last minute surge by Kerry/Edwards. I don't put too much stock into these early polls though. Pollster has a good feature on why it's so hard to tell who caucuses (every year, about 50% of Caucus goers did not attend 4 years before)

- virginiacentrist

September 17, 2007 at 10:11am

I like the TNR coverage of campaign staff, who's advising who, organizational strength (yeah, it's process), campaign profiles, interest group alignments, fundraising, message building. But the messages that ballot preference polls tell us this early in the game are much more subtle than so-and-so is leading.

- stgla

September 17, 2007 at 10:40am