SUSA POLLS for Michigan, Maine, Iowa, Oklahoma, Tennessee

[Floridude]

Tennessee

Bush 55   Kerry 41

Michigan  Kerry 52 Bush 42

Oklahoma Bush 64 Kerry 33

Iowa Bush 50 Kerry 46

Maine Bush 47 Kerry 46.

Nothing especially remarkable in these numbers, except that Michigan looks out of play

[The Vorlon]

Tennessee

Bush 55   Kerry 41

Michigan  Kerry 52 Bush 42

Oklahoma Bush 64 Kerry 33

Iowa Bush 50 Kerry 46

Maine Bush 47 Kerry 46.

Nothing especially remarkable in these numbers, except that Michigan looks out of play

Maine jumps out at me a bit.

Lots have polls have shown it Kerry up 0-5%, I think this is actually the first one to show a (statistically insignifigant) lead for Bush.

I don't think Bush is down 10 in Michigan, I also don't think he is all that close either however, I'd say 6 or 7, but no matter

[Sam Spade]

Bush has been advertising heavily in Maine from what I know and has campaigned up there of late.  I wouldn't be surprised if it's getting closer.

[MHS2002]

Tennessee:
http://www.surveyusa.com/2004_Elections/TN040923president.pdf

Michigan:
http://www.surveyusa.com/2004_Elections/MI040923president.pdf

Oklahoma:
http://www.surveyusa.com/2004_Elections/OK040923pressen3qs.pdf

Iowa:
http://www.surveyusa.com/2004_Elections/IA040923president.pdf

Maine:
http://www.surveyusa.com/2004_Elections/ME040923president.pdf

[Pollwatch99]

Maine is very good news.  That should be Kerry country

[CARLHAYDEN]

Lets take a good look at the SUSA Michigan poll.

First, they have 88% of registered voters as likely voters.

The highest percentage of registered voters actually voting in a Presidential election since the voting age went to 18 was 62.% in Michigan, with 58.8% (2000) being close to the mean and median.  

In short, the 88% turnout rate is totally implausible and should give one pause.

Now, lets look at some of the subsamples.

First, they have Kerry winning a higher percentage of the Democrats likely to vote than Bush does of the Republicans llikely to vote (very implausible).

Second, they have Kerry winning the military vote by 6 (all those who believe this, please post).

Third, they have Kerry winning the male vote by 8!

Fourth, they have Bush winning the White vote by 1 point.

Fifth, they have Bush winning the Regular Religious service attender vote by 1.

The SUSA poll has Wayne county (the Democrat stronghold) as casting 22% of the vote.  The history (including my projection for 2004) are included in the following table:

Year                    % of Michigan vote

1984                           22.84

1988                           20.39

1992                           19.70

1996                           19.01

2000                           18.16

2004                           17.47

(BTW, Wayne county has continued to decline in population both net loss of population and as a percentage of the state).

The poll is also skewed by age to include an absurdly high percentage of voters age 18-34.

I would really like to see a Tarrance poll of Michigan as the SUSA poll is absurd.

[dougrhess]

Lets take a good look at the SUSA Michigan poll.

First, they have 88% of registered voters as likely voters.

The highest percentage of registered voters actually voting in a Presidential election since the voting age went to 18 was 62.% in Michigan, with 58.8% (2000) being close to the mean and median.  

Where did you get that 62% of registered voters from? Registration rolls are very padded. Census has self reporting nationally at 86% in 2000. Granted that's self reporting, but the overall number of voters in 2000 from self reporting is not too far off the national. If I recall correctly; which I may very well not. ;-)

[Nym90]

Lets take a good look at the SUSA Michigan poll.

First, they have 88% of registered voters as likely voters.

The highest percentage of registered voters actually voting in a Presidential election since the voting age went to 18 was 62.% in Michigan, with 58.8% (2000) being close to the mean and median.  

In short, the 88% turnout rate is totally implausible and should give one pause.

Now, lets look at some of the subsamples.

First, they have Kerry winning a higher percentage of the Democrats likely to vote than Bush does of the Republicans llikely to vote (very implausible).

Second, they have Kerry winning the military vote by 6 (all those who believe this, please post).

Third, they have Kerry winning the male vote by 8!

Fourth, they have Bush winning the White vote by 1 point.

Fifth, they have Bush winning the Regular Religious service attender vote by 1.

The SUSA poll has Wayne county (the Democrat stronghold) as casting 22% of the vote.  The history (including my projection for 2004) are included in the following table:

Year                    % of Michigan vote

1984                           22.84

1988                           20.39

1992                           19.70

1996                           19.01

2000                           18.16

2004                           17.47

(BTW, Wayne county has continued to decline in population both net loss of population and as a percentage of the state).

The poll is also skewed by age to include an absurdly high percentage of voters age 18-34.

I would really like to see a Tarrance poll of Michigan as the SUSA poll is absurd.

I agree that Kerry isn't up 10 in Michigan, but he's up at least 5.

[Sam Spade]

You can usually go with the rule on these battleground states that if no public poll has shown one candidate up for about a month or so, then the other candidate is probably leading by 5+ points, somewhere outside the MOE. (which could lead to another candidate showing a lead at any time)

This is the reason why I consider Michigan and Washington to be 5+ point leads for Kerry, whereas Ohio, Missouri and Wisconsin have become 5+ point leads for Bush.  In none of those states have we seen a poll showing a lead for the other candidate in over a month.

[CARLHAYDEN]

Lets take a good look at the SUSA Michigan poll.

First, they have 88% of registered voters as likely voters.

The highest percentage of registered voters actually voting in a Presidential election since the voting age went to 18 was 62.% in Michigan, with 58.8% (2000) being close to the mean and median.  

Where did you get that 62% of registered voters from? Registration rolls are very padded. Census has self reporting nationally at 86% in 2000. Granted that's self reporting, but the overall number of voters in 2000 from self reporting is not too far off the national. If I recall correctly; which I may very well not. ;-)

Go to the links page, Michigan, SOS, Elections, historical turnout, 1992 (it was 62.5%)

[dougrhess]

Lets take a good look at the SUSA Michigan poll.

First, they have 88% of registered voters as likely voters.

The highest percentage of registered voters actually voting in a Presidential election since the voting age went to 18 was 62.% in Michigan, with 58.8% (2000) being close to the mean and median.  

Where did you get that 62% of registered voters from? Registration rolls are very padded. Census has self reporting nationally at 86% in 2000. Granted that's self reporting, but the overall number of voters in 2000 from self reporting is not too far off the national. If I recall correctly; which I may very well not. ;-)

Go to the links page, Michigan, SOS, Elections, historical turnout, 1992 (it was 62.5%)

I see. Well, that's the problem then. Like I said, states have very padded reg rolls. Cann't trust those numbers.

[CARLHAYDEN]

Well, according to the Bureau of the Census, 60.1% of the Voting Age Population of Michigan claimed they had voted in the 2000 Presidential election in Michigan (that approximately 4,345,000 votes).

There were actually 4,232,711 votes cast for President in Michigan in 2000.

[electcollfan]

I wonder where the voters they polled were in Maine? CD2 has always been more conservative than the CD1.
I don't think Bush will pull Maine, but he might have a shot at CD2.

[dougrhess]

Well, according to the Bureau of the Census, 60.1% of the Voting Age Population of Michigan claimed they had voted in the 2000 Presidential election in Michigan (that approximately 4,345,000 votes).

There were actually 4,232,711 votes cast for President in Michigan in 2000.

VAP does not equal adult citizens and that doesn't equal registered. About 8 percent of the US population is non-citizen.  It appears that in MI that number is only about 4percent (in 2000), so about 60-64% of all of the adult citizens voted.  However, over 85% of those Registered did vote, which is what it all comes down to. It will be interesting to see how high the registered voter turnout will be this year with all the claims to high interest and large GOTV projects in the works.  See what I mean?

Here's the link to the data for these numbers.

http://www.census.gov/population/socdemo/voting/p20-542/tab04c.pdf

[CARLHAYDEN]

Here's the url for the real numbers:

http://www.michigan.gov/sos/0,1607,7-127-1633_8722-29616--,00.html

BTW, the data you cited is "reported" i.e. self-identification NOT actual.

There are several problems with this.

First, more people tend to "report" they voted than actually voted.

Second, many people do not realize they are registered to vote.

Third, some people say they are registered when they are not.

Fourth, eligibility to vote in most states (Wisconsin and I believe North Dakota) is based on ACTUAL, not REPORTED registration.

Fifth, you are correct that many of the registered voters seldom (if ever) actually vote.

[CARLHAYDEN]

Well, according to the Bureau of the Census, 60.1% of the Voting Age Population of Michigan claimed they had voted in the 2000 Presidential election in Michigan (that approximately 4,345,000 votes).

There were actually 4,232,711 votes cast for President in Michigan in 2000.

VAP does not equal adult citizens and that doesn't equal registered. About 8 percent of the US population is non-citizen.  It appears that in MI that number is only about 4percent (in 2000), so about 60-64% of all of the adult citizens voted.  However, over 85% of those Registered did vote, which is what it all comes down to. It will be interesting to see how high the registered voter turnout will be this year with all the claims to high interest and large GOTV projects in the works.  See what I mean?

Here's the link to the data for these numbers.

http://www.census.gov/population/socdemo/voting/p20-542/tab04c.pdf

Well, it may seem hard for you to believe, but I do know that:

VAP does not equal adult citizens and that doesn't equal registered.

I also know that non-citizens aren't supposed to vote (unfortunately they do in some instances).

If you're really interested in the subject, and not just being snotty, I suggest you also consider the number of persons ineligible to vote based on conviction of a felony.

[dougrhess]

Here's the url for the real numbers:
http://www.michigan.gov/sos/0,1607,7-127-1633_8722-29616--,00.html

>> Thanks for the link. The only number that is "real" on this is the number that voted. The number of registrations is "real" in that it's the number they have on the books, but not in the sense that it is accurate: many states have absurdly high numbers of people registered. Most voter registrations that come into states (like 40%) are people changing their address. States don't have a good system for removing people if they move to another state, or often even another county, etc.  For instance, you say people over report their registration. That seems reasonable, but when then is the Census self report at 4.9 million for MI and the state at 6.9 million?!? The state has 2 million more people registered than say they are! <<

BTW, the data you cited is "reported" i.e. self-identification NOT actual.
>> True, and I know. <<

There are several problems with this.

First, more people tend to "report" they voted than actually voted.

>> True. This seems for MI to be off by only a little. The states has 4.279 million voting in 2000. Census has 4.343 million voting in 2000. A differnce of about 64,000 people. So, only about 1 percent off. Not bad I say. In fact, the MI ballot count is within the Census estimate's confidence interval. Considering that a certain number of ballots are cast but spoiled, and I assume not counted as cast by the state, Census appears even more accurate. <<

Second, many people do not realize they are registered to vote.

>> More people think they are and are not than the other way around. I state that based on experience in voter registration work with some national nonprofits (Project Vote and others).<<

Third, some people say they are registered when they are not.

>>Yes, but again, the state seems far more likely to be off than self-reporting. Just look at the numbers between your link and mine on registered. <<

Fourth, eligibility to vote in most states (Wisconsin and I believe North Dakota) is based on ACTUAL, not REPORTED registration.

>> Doesn't have anything to do with this argument. It's true, but not related. The states list people registered that are not even in the state or are double registered. The question is what inflates the estimate of real people truly registered more: the state or census. I think for the reasons given the census self reporting does a better job. In fact, I hadn't look at this issue in such detail until now and it seems that census is doing a better job on this than I had thought. I'll have to see how they deal with self reporting issues, non reponse and weights, etc.  The data is based on the Current Population Survey wich is of about 50,000 households. <<

Fifth, you are correct that many of the registered voters seldom (if ever) actually vote.

>>I didn't state that. Did i? If I did, I meant that the problem is registering people, not getting them to vote. Of citizens registered, for Prez elections, the VAST majority (>85%) do VOTE! The idea that Americans don't vote is somewhat of a myth. We've often based our estimates of this on VAP which is note the same as the VEP (voting eligible population). US citizens may vote less faithfully than other nations, but it's not as bad as people make it out to be when they use the shoddy data from the states.

Thanks for discussing this issue with me. It's forced me to look at some data that I wouldn't have otherwise. However, I think relying on the state data, for all these reasons, can lead to errors.

[dougrhess]

Well, according to the Bureau of the Census, 60.1% of the Voting Age Population of Michigan claimed they had voted in the 2000 Presidential election in Michigan (that approximately 4,345,000 votes).

There were actually 4,232,711 votes cast for President in Michigan in 2000.

VAP does not equal adult citizens and that doesn't equal registered. About 8 percent of the US population is non-citizen.  It appears that in MI that number is only about 4percent (in 2000), so about 60-64% of all of the adult citizens voted.  However, over 85% of those Registered did vote, which is what it all comes down to. It will be interesting to see how high the registered voter turnout will be this year with all the claims to high interest and large GOTV projects in the works.  See what I mean?

Here's the link to the data for these numbers.

http://www.census.gov/population/socdemo/voting/p20-542/tab04c.pdf

Well, it may seem hard for you to believe, but I do know that:

VAP does not equal adult citizens and that doesn't equal registered.

I also know that non-citizens aren't supposed to vote (unfortunately they do in some instances).

If you're really interested in the subject, and not just being snotty, I suggest you also consider the number of persons ineligible to vote based on conviction of a felony.

How am I being snotty? Jeesh.

[CARLHAYDEN]

BTW, self-reporting of voting overstates actual voting by about five per cent.

[dougrhess]

BTW, self-reporting of voting overstates actual voting by about five per cent.

For MI? For the US?

[CARLHAYDEN]

I gave the national figure with the 5%.

In Michigan in 2000 the self-described behavior overstated actual voters by about 3%.

[dougrhess]

I gave the national figure with the 5%.

In Michigan in 2000 the self-described behavior overstated actual voters by about 3%.

Looked like much less than that based on the website you pointed me too and the webiste I gave. How did you get 3%?  

[CARLHAYDEN]

Here's the url.

http://www.census.gov/prod/2004pubs/03statab/election.pdf

7,231,000 x 60.1% = 4.345.031 / 4,232,711 = 1.0267.

This closely agrees with the url which you posted which alledged that 4.343,000 voted in Michigan in 2000.

[dougrhess]

Thanks for that link.

I don't get your formula, however, as you're using the percent that is derived from the number registered (the inflated state number) and number who voted to get back to the original number. Oh well, guess we've beaten this to death.  

[CARLHAYDEN]

Apparently you are misreading my post.

Your own source suggested 4,343,000 voted for President in Michigan whereas 4,232,711 actually voted.

If you divide 4,343,000 by 4.232,711 you get the result i provided.

What part of this don't you get?