Wide lead?

The Star Tribune is talking about how every Democrat for major office is now in a wide lead for state office. Hardly a surprise, but consider that since 1987 the final Minnesota Poll results for Republican candidates have, on average, been 5.2 percent under the actual election results.

Since 1998, the final Minnesota Polls have underestimated Republican results by an average of 7.26 percent, while understating Democratic results by only .054 percent. But then in 1998, the poll showed that Skip Humphrey would win the governorship with 35 percent. But how much did he actually get? Mr. Humphrey received a whopping 28 percent, enough to put him in a solid third place.

Now put yourself in the position of those in charge of this consistently skewed poll. What would you do if your results were so wrong, so predictably – well – predictable? Perhaps you'd look to your pollster and ask a few questions. Maybe you'd even look for a new pollster. Not the Star Tribune.

If the Star Tribune Minnesota Poll is so accurate and if the Star Tribune has favored accuracy over politics; then why isn't Fritz Mondale our senator? Why wasn't Skip Humphrey our governor? Why didn't Al Gore win this state in a landslide?

They predicted all of the above.

This seems to be a GOP talking point this year. The problem with claiming the Star Tribune / Minnesota Poll have always been wrong and therefore will always be wrong is that it ignores the other data. If you don't like the MN Poll, which one do you like? They all show the same trends: the GOP is losing support and the DFL is gaining support. Personally, I'd rather see the Independence Party gaining ground, but we don't always get what we wish for.

Maybe you seem to be ignoring the point here. I am talking specifically to the Star Trib. Maybe you missed the point but they are off by a substantial margin in every election. If you look at polling done by other companies the lead over Mark Kennedy is usually between 5-10%. The Trib has never had polling less than 19%. They claim to have about a 2% margin of error and so do the other polls which mean that in the instance of the highest poll would be 12% and the trib at 17% they are still a WIDE margin off.

I don't particularly care who is winning in the polls but if you are going to present a number at least try to make it look like it is close to credible. If a scientist came up to you and said "I did a study and the earth has warmed 100 degrees in the last 100 years" you would dismiss him as an idiot because while he may be using sound sciene to come to his conclusion, he certainly isn't using it correctly. The same with the star trib

This topic has been discussed extensively - and rather heatedly, I might add - at the Star Tribune's "Big Question" blog. Eric Black invited the MN Poll pollster over to explain their methodology. His explanation of their technique and numbers was very informative.

If I may ask, where do your numbers come from? These ones are particularly interesting to me: "Since 1998, the final Minnesota Polls have underestimated Republican results by an average of 7.26 percent, while understating Democratic results by only .054 percent."

Alas it does not matter if they extensively discussed it or explained their methodology...The fact stands that they are consistantly WRONG. If you worked as a cashier and your till came up $7.26 short every night and you explained to your boss that your methodology of counting, no matter how good it is, you will still be fired for being short that much every night.

The fact of the matter stands that the Star Tribune isn't representing a true vote count. If they did, their gap wouldn't be so huge. Take example in the last presidential election, they predicted Kerry to win by 12%, a landslide if you will. But instead the actual tally came to only 3%. You can explain methodology to me all day, but if you can't produce a credible result, I will just dismiss you as an idiot.

The numbers would be from averaging the the difference between the actual votes cast and their own predictions from the latest poll taken.

According to the pollster, again I don't have the link handy, but its somewhere in the Big Question blog, he explains that they've noticed a Republican 'surge' in the last several days. My understanding is that the poll they publish on Sunday does not reflect this 'surge.' This may explain why the numbers you cite seem so far off.

It appears that some people expect polls to predict the election outcome, even several days in advance. Given the GOPs recent successes with microtargeting and their get out the vote efforts, I can see how the pollsters' jobs are tougher than ever. In fact, it seems like the Sunday 'misleading' polls likely help the GOP GOTV efforts - simply put together a call script something like "Have you seen the Sunday Star Tribune poll that claims we're losing? We need your help to keep candidate XYZ from making this country less safe, blah blah blah. Go vote for candidate ABC on Tuesday to save freedom and democracy!"

Err...isn't this supposed to be a blog written by fence-sitters? Ken's posts lead me to believe he is in fact a very partisan Republican.

bsimon, if polling is not done to show who is leading an election then why is it done? What is the point of asking people who they are going to vote for if not to figure out in advance is winning?

David, does being a "fence-sitter" mean one holds liberal views? If this is the case then no, I am no fence sitter. While my views fall more in line with republican views, a party such as theres holds no weight in my heart. Of course if you had even bothered to read all my posts you would notice that I am not a partisan Republican at all.

Another thing came to mind when I was reading some of the comments, to be quite honest they under-represented Jesse Ventura by a good 10%+ depending on the poll you listened to but he took a majority easily.

Of course a poll is intended to show levels of support for candidates. Your mistake is in reading the Sunday polls as gospel. As the MN poll pollster has explained, they've seen a GOP surge in the final days of the campaign, resulting in GOP victories despite late polls (i.e. the polls published the Sunday before election day) showing DFL leads.

It supports the argument that the GOP has a far better get out the vote machine than the DFL. Its not a secret. In fact, it supports the conventional wisdom about the GOP GOTV efforts, their microtargeting program and the '72 hour strategy.' Frankly, its a brilliant strategy; your opponent doesn't make the final effort they should, thinking they have a comfortable lead; meanwhile you are able to use the appearance of being behind to motivate voters to get to the polls.

It appears that some people expect polls to predict the election outcome, even several days in advance.

Of course a poll is intended to show levels of support for candidates.

Please, explain how those two statements remotely agree with each other.

You fail to reach a logical conclusion from all the blathering you are doing. If a poll was an indication of support, then regardless of the get out the vote effort it would still be near the actual turn out data. Polls are done in category's of "Likely Voters" "Registered Voters" and then general population. If the Star Tribune had an honest polling system they would at least be close in their poll. Just because the turn out is higher doesn't change who they would vote for if they were asked today. If I asked you if you would vote for Mark Kennedy or Amy Klobuchar, would that change if a GOP pollster called you to go vote?

If it doesn't please explain to me how a get out the vote effort changes poll numbers.

The whole point of GOTV efforts is to get people to 1) decide on your candidate and 2) get them to the polls. They are targeting the people that respond to pollsters as 'undecided' or that pollsters are putting in the 'unlikely to vote' category. There are stories on microtargeting at both MPR and the Star Tribune, that get into the details of how the campaigns identify these voters & what some of the tactics are to get them to vote.

Regarding your interpretation of two comments as being in conflict, I'll try to explain it differently. Polls are showing the horse race. Sometimes its safe to call the race on the back straightaway, sometimes you have to wait until the home stretch. And sometimes its a photo finish. I can take a look at a race & predict who'll win once the horses are on the home stretch, but sometimes I'll be wrong because a horse finds an energy reserve & surges ahead in the last few lengths.

You dodge the question by stating something not even asked. How does a call from a political party change your poll? A poll is a broad spectrum of of people of which you reach a number of people for a candidate. If you polled people and they said they would vote for x or y, how does the call from a party change that? It simply does not.

Alas, I know of no horse race where a horse that was 29% of the track ahead of all the other horses lost as that is a substantial lead.

I'm a shop steward of my union and I do read the Star Trib from time to time, but I would never buy it. The reason is clear and is proven in the polling results that they print as fact.

If we believe the reported poll results have a bias, it supports our belief that the stories it reports have a bias. If I read that Mark Kennedy is losing by a margin of 19 percent, I have no reason to read further because I know that it's not accurate, and I will question the accuracy of the next story I read.

Can anyone tell me why a Democrat talks to a group of labor leaders and a Republican talks at a party put on by the Beltway Young Republicans, and poll results come from people who write blogs on a Web site at MPR.org?