Really quite predictable redux

"It's what you learn after you know it all that counts."
-Earl Weaver, former MLB manager

Adjusting his model after missing the turnout in Pennsylvania by over 38% and by over 37% in Indiana, Poblano of fivethirtyeight.com confidently predicted Hillary Clinton would win West Virginia by "39 points, 105,000 votes."

Hillary Clinton won by 41 points and she out polled Barack Obama by over 147,000 votes, making Poblano's prediction of a 105,000 vote win off by over 40%.

Poblano's new model not only underestimated the actual turnout by over 84,000 votes, it also underestimated votes for Clinton by over 31%, votes for Obama by over 14%, and votes for Edwards by over 42%.

As for Poblano's snark, our West Virginia poll was closer to predicting the final results for Clinton, Obama, and Edwards, even with 6% undecided.

--Dick Bennett

May 16, 2008 - Kentucky Democratic Primary Preference

Hillary Clinton leads Barack Obama 56% to 38% among men (45% of likely Democratic primary voters). Among women, Clinton leads 72% to 21%.

Clinton leads 73% to 21% among white voters (87% of likely Democratic primary voters). Obama leads 91% to 7% among African American voters (11% of likely Democratic primary voters).

Clinton leads 59% to 35% among voters age 18 to 49 (46% of likely Democratic primary voters) and Clinton leads 71% to 23% among voters age 50 and older.

12% of likely Democratic primary voters say they would never vote for Hillary Clinton in the primary and 50% of likely Democratic primary voters say they would never vote for Barack Obama in the primary.

For details, click on the R or D for each state in the column on the left under 2008 Presidential Polls.

May 16, 2008 - Oregon Democratic Primary Preference

Barack Obama leads Hillary Clinton 57% to 39% among men (48% of likely Democratic primary voters). Among women, Clinton leads 51% to 44%.

Obama leads 51% to 44% among white voters (88% of likely Democratic primary voters). Clinton leads 50% to 46% among Hispanic voters (6% of likely Democratic primary voters).

Obama leads 55% to 38% among voters age 18 to 49 (49% of likely Democratic primary voters) and Clinton leads 52% to 45% among voters age 50 and older.

Clinton and Obama are tied at 49% each among voters saying they have returned their ballots (58% of likely Democratic primary voters) and Obama leads 52% to 40% among voters saying they will definitely return their ballots by May 20th.

20% of likely Democratic primary voters say they would never vote for Hillary Clinton in the primary and 22% of likely Democratic primary voters say they would never vote for Barack Obama in the primary.

For details, click on the R or D for each state in the column on the left under 2008 Presidential Polls.

A Surprise About Obama, Clinton, and Turnout

Conventional wisdom has it that Barack Obama's primary victories are based on his ability to increase turnout.

A look at what happens when voter turnout increases in the primaries proves that this notion is wrong. In fact, Obama has had his greatest primary (and caucus) victories when turnouts have been low.

Obama received 66% of the primary vote in Georgia when 22.7% of all registered voters in the state voted in the Democratic primary. In Virginia, Obama received 64% of the primary vote when 21.1% of all registered voters in the state voted in the Democratic primary. In Mississippi, Obama received 61% of the primary vote when 24.4% of all registered voters in the state voted in the Democratic primary.

Hillary Clinton received 67% of the primary vote in West Virginia when 30.1% of all registered voters in the state voted in the Democratic primary. In Pennsylvania, Clinton received 55% of the vote when 27.7% of all registered voters in the state voted in the Democratic primary. In Ohio, Clinton received 54% of the primary vote when 30.5% of all registered voters in the state voted in the primary.

There are exceptions (such as Arkansas, Tennessee, and Rhode Island), but Clinton, and not Obama, tends to win in the states where turnouts are higher. The relationship is strong enough that Democratic primary turnout can be predicted using Clinton's share of the vote.

--Dick Bennett

May 9, 2008 - West Virginia Democratic Primary Preference

Hillary Clinton leads Barack Obama 57% to 27% among men (43% of likely Democratic primary voters). Among women, Clinton leads 72% to 20%.

Clinton leads 70% to 19% among white voters (93% of likely Democratic primary voters). Obama leads 91% to 3% among African American voters (5% of likely Democratic primary voters).

Clinton leads 61% to 27% among voters age 18 to 49 (47% of likely Democratic primary voters) and Clinton leads 72% to 19% among voters age 50 and older.

13% of likely Democratic primary voters say they would never vote for Hillary Clinton in the primary and 45% of likely Democratic primary voters say they would never vote for Barack Obama in the primary.

For details, click on the R or D for each state in the column on the left under 2008 Presidential Polls.

The model got it right?

Mark Blumenthal of Pollster.com declares in his weekly column for NationalJournal.com that the regression model developed by Poblano of fivethirtyeight.com correctly predicted the 2-point margin in the Democratic presidential primary results from Indiana.

But, surprisingly, Mark fails to look at the projections on which the margin was based. Those projections clearly show, as was the case for Pennsylvania, that the model failed to come close to accurately predicting the actual turnout.

The model underestimated the actual turnout by over 37% (it was off by over 38% in Pennsylvania) and the difference over the actual by candidate has Obama up 4,349 votes over Clinton.

We will never know what the projected margin would have been if the model accurately predicted the turnout, but looking at the difference by candidate suggests it would have underestimated the vote for Obama.

The model was wrong, but it happened to get the 2-point margin right. Poblano's models continue to fail at something she or he confidently claims to be "really quite predictable" - this time by 346,337 votes.

--Dick Bennett

April 21, 2008

George W. Bush's overall job approval rating is holding steady at 22% while 68% of Americans say the national economy is in a recession and 54% say their personal financial situations are getting worse. Details from the nationwide survey conducted April 16-19 are available at The National Economy.

ARG's Electoral Vote Calculator has been updated to reflect the 2004 results. Assign states to the Republican or Democrat for 2008 and watch it calculate the electoral votes for president at Electoral Vote Calculator.

Why doesn't most advertising work? Take a look at 10 Rules for More Effective Advertising.

See your out-of-conscious processing system at work and learn how to improve your marketing and advertising efforts by understanding the power behind the Scintillating Grid.

The margins of error reported for most polls underestimate the actual margins of error for ballot estimates from the same samples. Use the Ballot Lead Calculator to determine if the lead for any survey is statistically significant.

The New Hampshire Poll:

Results from the March 2008 Quarterly New Hampshire Poll are available at The New Hampshire Poll. NH residents rate NH business conditions, their personal finances, John Lynch, and George W. Bush.