Ballot Measures

[CARLHAYDEN]

In a discussion I had several months ago with the Myster Pollster, we agreed that surveying ballot measure is harder than candidates.

The Survey USA folks proved it this year.

In Colorado they had the anti-gay marriage measure failing by a point (it won by 11 points), a difference of 12 points!

Also in Colorado, they had Referendum K (which instructed the Colorado Attorney General to sue the U.S. Government for failing to secure the border) tied, whereas it passed with a twelve point margin, a difference again of twelve points!

In Washington, SUSA had Initiative 920 (repealing the state Estate tax) winning by five points, whereas it actually won by 20 points, a 15 point diffence!

[Grad Students are the Worst]

I have no idea where you are getting these numbers, but you are entirely off.  The estate tax repeal failed by 24 points, not passed by 20.  Also, the last SurveyUSA poll had it failing 51%-40%, which - while still quite off - is more than five points.

[CARLHAYDEN]

Thank you for the correction.

[Grad Students are the Worst]

I meant "they are entirely off," by the way.  The post wasn't meant to come off as passive-aggressive as I now read it as coming off as. 

[Beet]

Survey USA is so random. IIRC, they had most of the California ballot measures passing in 2005, while at the same time PPIC was showing wildly different results (that ended up much closer to the mark).

[CARLHAYDEN]

Survey USA is so random. IIRC, they had most of the California ballot measures passing in 2005, while at the same time PPIC was showing wildly different results (that ended up much closer to the mark).

The point I was making was threefold:

first, too many people place too much faith in the accuracy of polls,

second, ballot measures tend to be harder to accurately survey than candidate races, and

third, where the liberal media is pouring vitroil on one side of a ballot measure, the polls tend to significantly "off" the actual results by a considerable margin.(I first saw this on Prop. 15 in California in 1982).

[Grad Students are the Worst]

third, where the liberal media is pouring vitroil on one side of a ballot measure, the polls tend to significantly "off" the actual results by a considerable margin.(I first saw this on Prop. 15 in California in 1982).

What do you mean by this?

[CARLHAYDEN]

third, where the liberal media is pouring vitroil on one side of a ballot measure, the polls tend to significantly "off" the actual results by a considerable margin.(I first saw this on Prop. 15 in California in 1982).

What do you mean by this?

Well, in California in 1982 only ONE major media outlet (the Orange County Register) opposed Prop 15 while the majority of major media outlets (newspapers and television stations in that state) that took a position on the issue supported the proposition.

The Field poll at one point showed the measure winning by approximately two to one, but it lost by about that margin.

The liberal media was particularly nasty on that proposition.

[Grad Students are the Worst]

Well, in California in 1982 only ONE major media outlet (the Orange County Register) opposed Prop 15 while the majority of major media outlets (newspapers and television stations in that state) that took a position on the issue supported the proposition.

The Field poll at one point showed the measure winning by approximately two to one, but it lost by about that margin.

The liberal media was particularly nasty on that proposition.

But what effect do you think the media has on polling results differing from final results?  To follow your specific example, I do not see anything in the Field Poll's methodology (depends on how they asked the question?) that would matter.

It's also worth considering that initiative polling is even harder because most people won't know what just "Initiative 400" is, and might say "yes" on the phone if it's the "Tax Lowering Initiative," while in the voting booth they might think of it as the "Education Cuts Initiative."

[CARLHAYDEN]

Well, in California in 1982 only ONE major media outlet (the Orange County Register) opposed Prop 15 while the majority of major media outlets (newspapers and television stations in that state) that took a position on the issue supported the proposition.

The Field poll at one point showed the measure winning by approximately two to one, but it lost by about that margin.

The liberal media was particularly nasty on that proposition.

But what effect do you think the media has on polling results differing from final results?  To follow your specific example, I do not see anything in the Field Poll's methodology (depends on how they asked the question?) that would matter.

It's also worth considering that initiative polling is even harder because most people won't know what just "Initiative 400" is, and might say "yes" on the phone if it's the "Tax Lowering Initiative," while in the voting booth they might think of it as the "Education Cuts Initiative."

First, Field is one of the oldest (and most reputable) state limited pollsters.  There was NO implication in my prior postings of any criticism of their methodology.

Second, there was MASSIVE coverage of Prop 15 (indeed, it was credited for getting Duke elected Governor that year).  So, NO it was not a matter on which their was little knowledge/media coverage.

Third, the liberal media in California called those who opposed Prop. 15 every dirty name they would think up.  So, when contacted by Field, many respondents gave the answer the liberal media wanted, but in the privacy of the polling place, they voted intelligently (i.e. against Propl 15).

[Grad Students are the Worst]

You really think that HALF of people responded to a survey differently than in the polling place because they were afraid the liberal media controlled the polling company (or whatever the case it was)?  Not meaning to present a strawman here, but that seems pretty extreme.  Field may be the oldest pollster, but they also could have just f'ed up.

[CARLHAYDEN]

You really think that HALF of people responded to a survey differently than in the polling place because they were afraid the liberal media controlled the polling company (or whatever the case it was)?  Not meaning to present a strawman here, but that seems pretty extreme.  Field may be the oldest pollster, but they also could have just f'ed up.

Alcon,

First, the difference was one third of the electorate, not half.

Second, there is a widespread tendency for people to give 'politically correct' answers in instances where the atmosphere has been poisoned by vitroil from the liberal media.

Please not the variation in the two Colorado measures this year!

[Grad Students are the Worst]

You really think that HALF of people responded to a survey differently than in the polling place because they were afraid the liberal media controlled the polling company (or whatever the case it was)?  Not meaning to present a strawman here, but that seems pretty extreme.  Field may be the oldest pollster, but they also could have just f'ed up.

Alcon,

First, the difference was one third of the electorate, not half.

Second, there is a widespread tendency for people to give 'politically correct' answers in instances where the atmosphere has been poisoned by vitroil from the liberal media.

Please not the variation in the two Colorado measures this year!

I'm sorry; I haven't slept in a few days - you are right.  One third.  Nonetheless, one third.

If what you say is true, why were 2004 Presidential polls so accurate?  I am not necessarily challenging your conclusions; just engaging them.

It's also worth noting the SD measure (I think there was a poll that showed it passing handily, although I am not entirely sure).

[CARLHAYDEN]

You really think that HALF of people responded to a survey differently than in the polling place because they were afraid the liberal media controlled the polling company (or whatever the case it was)?  Not meaning to present a strawman here, but that seems pretty extreme.  Field may be the oldest pollster, but they also could have just f'ed up.

Alcon,

First, the difference was one third of the electorate, not half.

Second, there is a widespread tendency for people to give 'politically correct' answers in instances where the atmosphere has been poisoned by vitroil from the liberal media.

Please not the variation in the two Colorado measures this year!

I'm sorry; I haven't slept in a few days - you are right.  One third.  Nonetheless, one third.

If what you say is true, why were 2004 Presidential polls so accurate?  I am not necessarily challenging your conclusions; just engaging them.

It's also worth noting the SD measure (I think there was a poll that showed it passing handily, although I am not entirely sure).

A few simple rules rules for accuracy in polling.

First, polling of general elections with partisan candidates is easier than general elections with non-partisan candidates.

Second, polling general elections is easier than primary elections.

Third, polling elections with incumbents seeking reelection/renomination is easier than polling of an election without an incumbent seeking reelection/renomination.

Fourth, polling elections for candidates is easier than ballot measures.

Fifth, inaccuracies of polls on ballot measures are highest where the media is overwhelmingly on one side of an issue AND they take a virulent position (example, only Klansman would oppose affirmative action preferences).

[Grad Students are the Worst]

A few simple rules rules for accuracy in polling.

First, polling of general elections with partisan candidates is easier than general elections with non-partisan candidates.

Of course.

Second, polling general elections is easier than primary elections.

Yes.

Third, polling elections with incumbents seeking reelection/renomination is easier than polling of an election without an incumbent seeking reelection/renomination.

Yes, again.

Fourth, polling elections for candidates is easier than ballot measures.

Much.

Fifth, inaccuracies of polls on ballot measures are highest where the media is overwhelmingly on one side of an issue AND they take a virulent position (example, only Klansman would oppose affirmative action preferences).

Do you have proof of this beyond the few polls you cited?  E.g. a general report that says such?

[CARLHAYDEN]

A few simple rules rules for accuracy in polling.

First, polling of general elections with partisan candidates is easier than general elections with non-partisan candidates.

Of course.

Second, polling general elections is easier than primary elections.

Yes.

Third, polling elections with incumbents seeking reelection/renomination is easier than polling of an election without an incumbent seeking reelection/renomination.

Yes, again.

Fourth, polling elections for candidates is easier than ballot measures.

Much.

Fifth, inaccuracies of polls on ballot measures are highest where the media is overwhelmingly on one side of an issue AND they take a virulent position (example, only Klansman would oppose affirmative action preferences).

Do you have proof of this beyond the few polls you cited?  E.g. a general report that says such?

Only about two per cent of ballot measures meet the criteria set forth in my fifth point.  So, there isn't a large body of data on the subject.  However, if you check with the MysteryPollster (or Vorlon) I believe they will confirm my point.

Oh, and btw, an indirect way in which poll results are tilted in cases meeting the criteria cited in my number five is that the number of respondents declining to answer or complete the survey is higher in such instances.

[Grad Students are the Worst]

I don't dispute any of your conclusions, except the final.  I think the media does increase inaccurate responses in polling, but I think general error is much more significant.

I don't dispute that it is true - I am sure it is, in part - but I can't objectively believe it until I see numerical, unslanted proof.