Comments on FiveThirtyEight.com: Electoral Projections Done Right: Expectations and Realities

How about using those leaked Obama campaign number...

2008-04-22T14:41:00.000-05:00

How about using those leaked Obama campaign numbers? If I remember right they were something like "Clinton wins by 6-9".

However, that is my point.So because either Obama ...

2008-04-22T13:37:00.000-05:00

However, that is my point.

So because either Obama closed the gap from 12-16 points to 6 points (or Clinton lost that lead) he is now having to do better than 6 points to "exceed expectations".

The time to call a spread is before the teams officially take the field. By calling it now is like calling it when we are in the final minutes of the game with special teams on the field.

I am not saying that it should have been called a 20 point race 6 weeks ago. However, logic could have been used to call the spread at that time and up to 2 weeks ago.

To call it on the final day (or the day before) based upon the recent polling ignores the fact of movement up until that day (either way - up or down). It places it all on GOTV (get out the vote) instead of also on message and the full scope of campaigning.

I think the term 'expectations' can be understood ...

2008-04-22T12:11:00.000-05:00

I think the term 'expectations' can be understood in many ways. For sure,the media expectations many weeks ago were for a Clinton landslide (however that might be defined -- let's say more than 15 points). But I doubt that even then either the Clinton or the Obama campaign expected it to end up that way today.

The article in question here can also reasonably be understood as describing the 'expectations' by polling and political experts, and then translated into media frames for interpreting what's high or low and the implications for the future of the campaign.

Another way to think of expectations is strictly in a statistical sense. Poblano's regression model, which is based on the experience of previous primaries this year coupled with demographic data, provides a "statistical expectation," i.e., best estimate based on that data along with a range of uncertainty.

These are all reasonable, and not b.s.

I call Bullsh!t to McIntyre's article. Why? Becaus...

2008-04-22T10:23:00.000-05:00

I call Bullsh!t to McIntyre's article. Why? Because how can 6-9% be expectations on the day of the election? The term "expectation" implies a length of time involved. Calling an expectation using data on the day of an election uses no significant amount of time.

6 weeks ago, RCP had the race at 12 points. 2 weeks ago at 16 points.

When you look at the 3 major "northern" states bordering PA - Clinton won OH by 10.1, NH by 9.8 and NY by 17.5 - and yet PA should only be expected at 6-9???

Football Analogy. 2 minute warning - the team is on their own 20 yard line. The logical expectation is - they probably won't make a touchdown - just a field goal. Now it is 10 seconds on the clock and they are on their opponents 5 yard line and it's 3rd down. What is the expectation? The expectation is STILL a field goal. It ISN'T a touchdown because time has run down - it's not 2 minutes left on the clock. If they score a touchdown, they exceeded expecations - if they are held to a field goal, that is in line with logic. If the opposing team forces a fumble - that's a victory for the opposing team.

PA was supposed to be owned by Clinton - more so than OH. Clinton had PA's most successful statewide Democrat in her corner - and if there is a D-machine - it is Rendell. Rendell stated a couple of weeks ago that Clinton was going to win PA easily. Now, if Obama doesn't meet 6-9 points (less than OH), he somehow doesn't live up to expectations?

Bullsh!t.

Sounds fine, except I think caucus states may disa...

2008-04-22T07:51:00.000-05:00

Sounds fine, except I think caucus states may disagree that a popular vote win is as "legitimate" as a delegate win.