Presidental survey page

From WikEd

Jump to: navigation, search

Update for 2008 [Done for EPSY490OL, by Michael Geraghty]:

The first big polling/Survey issue of the 2008 election is the polling that took place prior to the New Hampshire Primary. On the Democratic side especially, the polling data did not match up with the actual election results. Barck Obama's loss to Hillary Clinton in New Hampshire suprised many, given polling both internal to the campaign and and external by news outlets, had Obama winning by at least several if not double digit points. Why did this happen? There are many opinions. The New York Times offers this explanation: "In the end, it seems, the preferences of a considerable number of New Hampshire voters were very much in flux in the final days of the campaign. What pollsters call “considered opinion” — the kind of opinion born of reflection rather than one elicited in an instant by a poll taker — registers only when people step into the ballot box." [1]


Pick two surveys of the presidental race that differ substantially and discuss reasons why they differ. (Or you can email me a Word file and then link it from here. Or you can email your survey to everyone in our seminar.)
jal

I like this quote from today's (Monday, Oct 25th) Electoral Vote Predictor site: "Some of these results are very surprising. Is Kerry really leading by 4% in Colorado? Is Bush really leading by 5% in New Mexico? I don't believe either of those. They are in conflict with too many other polls. Another example: the current Ohio University poll gives Kerry a 6% lead in that state, whereas Zogby puts Bush ahead by 5%. The MoE on these polls is 4%, so an 11% change in a couple of days in a state with so few undecideds is impossible. I think there are serious problems with all the polls."
jal

Comments from Heather

On Monday, October 25, the USA Today poll has Bush ahead by 2 points, whereas the Washington Post / ABC News poll has Kerry ahead by 1. Both polls have a margin of error of +- 3%, so technically, both polls are possible, nevertheless, it is striking that at this stage of the campaign, there is still so much difference between various polls (see Jim's comments above). In the case of these two polls, both are based on telephone interviews, both asked similar questions (though it is interesting that the USA Today poll, which has Bush ahead, placed Kerry's name first in the question and vice versa), sample sizes for both polls are similar, both use rolling samples over three days (the same three days -- 10/21 - 10/24), and both provide present results of "likely voters". The main difference seems to be in how the polling organizations identify and weight "likely voter" responses. Both use a series of questions (though different questions) to determine current likelihood of voting and past voting behavior. Both then use weighting formulas of past demographic trends in presidential elections to determine how to weight inidividual responses in their poll results. These formulas differ and may fail to reflect actual voter turnout in this election.

Additional details:


USA Today Poll http://www.usatoday.com/news/politicselections/nation/polls/usatodaypolls.htm

Date reported: Monday, October 25, 2004

Current standing]: o Bush- 49% o Kerry 47% o Nader 1%

Population surveyed: Likely voters

Question: “Suppose that the presidential election were being held today, and it included John Kerry and John Edwards as the Democratic candidates and George W. Bush and Dick Cheney as the Republican candidates. Who would you vote for?�?

Methodology: Results are based on telephone interviews with 1,538 National Adults, aged 18+, conducted October 22-24, 2004. For results based on the total sample of National Adults, one can say with 95% confidence that the margin of sampling error is ±3 percentage points.

Results based on likely voters are based on the subsample of 1,195 survey respondents deemed most likely to vote in the November 2004 General Election, according to a series of questions measuring current voting intentions and past voting behavior. For results based on the total sample of likely voters, one can say with 95% confidence that the margin of sampling error is ±3 percentage points. The likely voter model assumes a turnout of 60% of national adults. The likely voter sample is weighted down to match this assumption.

For results based on the sample of 1,461 registered voters, the maximum margin of sampling error is ±3 percentage points. This includes respondents who say they plan to register in states that allow same day voter registration.

In addition to sampling error, question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of public opinion polls.


Washington Post / ABC News Poll http://www.washingtonpost.com/wpsrv/politics/elections/2004/charting.html

Date reported: Monday, October 25, 2004

Current standings: o Bush – 48% o Kerry – 49% o Nader – 1%

Population surveyed: Net leaning... Likely voters

Question: If the 2004 presidential election were being held today, would you vote for George W. Bush and Dick Cheney, the Republicans, John Kerry and John Edwards, the Democrats, or Ralph Nader and Peter Camejo, the independents? Which candidates are you leaning toward?


Methodology: The Washington Post tracking poll will be conducted daily until election day. Results are based on telephone interviews with a randomly selected national sample of adults, including self-identified registered voters and likely voters.

The Post and ABC News collect data jointly but are responsible for developing their own methods to identify likely voters. This may produce slightly different estimates of candidate support. The margin of sampling error is plus or minus 3 percentage points for registered and likely voters, and slightly larger for subsamples. Sampling error is only one of many potential sources of error in this or any other public opinion poll. Interviewing was conducted by TNS of Horsham, Pa.

This tracking poll is based on a rolling three-day sample. About 350 likely voters are polled each day. To update the numbers, a new day's sample of respondents is added to the total sample and the oldest day's sample of respondents is dropped out. The Post uses seven variables to define likely voters, including whether the respondent states they are registered to vote, their intention to vote, past voting history, interest in the presidential campaign, age, whether the respondent is voting for the first time in 2004 and whether the voter knows the location of his or her polling place. These variables produce a sample of likely voters that is largely composed of individuals who regularly vote in presidential elections but does include newly registered as well as other first time voters. In a typical sample, about one in 10 likely voters are self-described first-time voters and one in six are between the ages of 18-29.

The Post adjusts, or "weights," each day's randomly selected samples of adults to match the voting-age population percentages by age, sex, race, and education, as reported by the Census Bureau’s Current Population Survey. The Post also adjusts the percentages of self-identified Democrats and Republicans by partially weighting to bring the percentages of those groups to within three percentage points of their proportion of the electorate, as measured by national exit polls of voters in the last three presidential elections.

Polls (Chris)

Tracking the polls for this year’s presidential race is both interesting and frustrating. It is interesting how the numbers flow and change. It becomes frustrating trying to find out how the poll was conducted. The websites we have and the ones I have found discuss the different methods used and how those methods can change the statistics.

I have always heard statistics referred to as the Art of lying with numbers.

As this poll indicated, President Bush has anywhere from plus 5 point over John Kerry or is tied. But with the margin of error for these polls being +/- 4 to 5 points, most of these polls come down to a statistical tie anyway.

What I found to be most surprising was the small sample sizes for these polls. None of the ones listed on the major poll tracking services had more that 2000 people, and they varied from registered voters to likely voters.

So what does this mean? Well, we will see next week.

Pam’s West-Wing-laced polling potpourri

I went to Real Clear Politics (is that an oxymoron or what?) and looked at the array of polls. When I headed off to look at the specifics behind some of them, like Chris, I was frustrated that it wasn’t always possible (or easy) to find out how the poll was conducted. I downloaded the report for both the Fox and Marist polls (for 10-17, +7 and +1 for Bush, respectively - Marist is conducted by a bunch of college students) and in addition to the differences between how questions are worded, I was noticing the different kinds of questions asked.

Marist did more to find out what was behind people’s thinking in making their choice - like how important particular issues were, or what traits they most liked about their candidate. Fox went more for outcome oriented data - e.g. who will fight the war on terror more aggressively. The contrast struck me as nuanced and investigatory vs. validating pre-existing ideas.

The bulk of the data in these polls is really more for campaign organizers. Most of what the public gets is the who-would-you-vote-for numbers. It’s interesting to think about what campaigners do with it all, and what they ask for from the pollsters. Which leads me to a recent rerun of...

West Wing! :)

So... Martin Sheen was faced with whether to make an issue out of flag burning. Some guy (the one that played Q on Trek) had done a poll that showed a surprising number of people thought it should be banned (esp. people who are likely to think Sheen was a wimpy prez) so he was jazzed about how it would be a good way for this president to win points with those voters. BUT, Marlee Matlin happened to have done a poll that asked how much do you CARE about flag burning. Answer: not that important, not a vote swayer. Shulman woulda been proud. :) Marlee “carefully framed that question most important...�? She really impressed (and relieved) everybody with that little tidbit (and it looks like Josh is falling for her but wait! she’s sleeping with Q! Ewwww!).

Anyway, it’s obviously frightening to think about the ways in which polling data affect political discourse and turn into sound bytes. The fact that it is so established, as an operating influence, makes the fuzziness of the numbers all the scarier. And Chris’s point about the margins for error is spot-on. They really do make all this data almost irrelevant, given the closeness of it all. Personally, I’m hoping (likely in vain) for some sort of shocking outcome that defies the polls and upends the reliance on them. They’ve taken on too big a voice in the election imho.

A footnote from Real Clear Politics. I noticed that Nader consistently gets higher numbers from the Time poll. What does that tell us? I wonder if he gets left out of certain polls. And I wonder what HE thinks of polling data.

Time magazine had an interesting comment on polling... It can be found at http://www.time.com/time/election2004/article/0,18471,733814,00.html

Heather

I assume each of your textbooks' chapters on surveys describe the infamous Literary Digest poll in 1936. If not, take a look at this web site. The same issue has come up this year because of the increasing number of people, especially young people, who have just cell phones and no "land line" phones, and therefore who are unavailable for pollsters to contact. jal

Another issue that has relevance for any survey that "selects" who to report:
"Gallup ran a poll in Iowa Oct. 22-25, only a day before Zogby's, and found that among registered voters, Kerry leads 48% to 47%. However, among what Gallup believes are likely voters, Bush is ahead 50% to 46%, a surprisingly large difference. This immediately brings up the issue of how Gallup decides to reject some respondents as unlikely voters. The likely voter screen involves questions about past voting behavior, interest in the election, and knowledge of where the polling place is. Given the huge number of new voters this year, there are legitimate questions about how accurately pollsters can predict likely voters." (from http://www.electoral-vote.com/)

First I have to say how much I enjoyed the posts here! I am fascinated by the way in which these polls create a mind-set that is manipulated by (and manipulates) the media and agree with Pam that this is way out of control. As someone who regularly makes good use of caller ID to avoid pesky callers, I would probably be one of those who is unreachable, yet will vote. However, my household includes individuals who might be considered less likely to vote (although these same folks usually have filled out their sample ballots and head to the polls even earlier than I do). So depending upon who answered the phone... The way in which all this gets weighted as Heather says is quite dubious to me.

In that spirit I decided to take an internet poll, lured by the promise of an iPod Mini I saw in an ad (of course this turned out to be a CHANCE to win an iPod Mini, like I really need another toy like this anyway). That survey at PollingPoint (see link below) was much like both polls described by Pam: a mixture of outcome-type questions (Who will keep us safer from terror?) and questions about beliefs on gay marriage, gun control, and other issues. These were of course combined with all the demographic info on respondent education, party affiliation, etc. Several of these questions were also internally weighted (do you consider yourself a "strong democrat"? "Just a democrat"? and so on).

PollingPoint.Com [2] has done several additional surveys related to the elections, and has their own way of "weighting" for what they admit is the decidedly "unscientific" nature of internet polling (given the selection bias). However, as Jim's post points out, changing uses of technology are already skewing selection and so new methods will have to be developed if polls are to continue to play a role. I think it's also interesting that while polls do create some kind of collective media mind set, on an individual level I'll bet that not many people-on-the-street actually "trust" them (sounds like a survey).

Cheryl

According to AlterNet, there are 40 million Latinos in the US making them the largest minority in the country. At the same time, there are many misconceptions about this population, among them the belief that this group consists of social conservatives, and that their main electoral issue is immigration. In fact a Washington Post/Univisión/Tomás Rivera Policy Institute poll suggests that the most important issues among registered Latinos are:

  • 33% economy
  • 18% education
  • 15% terrorism
  • 13% war in Iraq

Survey 1: Washington Post/Univisión/Tomás Rivera Policy Institute

  • Methodology: Random phone interviews of 1,605 registered Hispanic voters done between the 6th and 16th of July by a group of USC independent academics. Eleven states participated including California, Texas, Florida, New York, Arizona, Illinois, Nuevo México, New Jersey, Colorado, Virginia and Massachusetts. Together these states represent 88% of the total of Hispanic voters. The margin of error was of approximately 3%. Field research was conducted by Interviewing Service of America (ISA) from Van Nuys, California.
  • These are some of the poll findings:
    • 54% disapprove of Bush’s job as president
    • If elections were held today (July 2004) 58% of the participants would vote for Kerry, 28% for Bush, and 2% for Nader
    • Who are you most inclined to vote for? Bush 15%, Kerry 20%, Nader 1%, Don’t know 54%
    • Democrat 56%, Republican 18%, Independent 22%, Don’t know 4%

Survey 2: Latino Coalition (per HispanicBusiness.com)

  • Methodology: This national Hispanic survey was conducted by Latino Opinions between September 27-October 3, 2004, among 1,000 Hispanic adults. All interviews were conducted by professional English- and Spanish-speaking interviewers via telephone. Respondents were given the option of conducting the survey in English or Spanish. Interview selection was at random within predetermined population units. These units were structured to correlate statistically with the nation's adult Hispanic population according to the 2000 U.S. Census. The accuracy of this national survey of 1,000 Hispanic adults is within ±3.1%, at a 95% confidence interval.
    • The Latino Coalition (TLC) is a non-profit, non-partisan organization based in Washington, D.C. TLC was established to address policy issues that directly affect the well-being of Hispanics in the United States. TLC’s agenda is to develop and promote policies that will foster economic equivalency and enhance overall business, economic, and social development of Hispanics.
    • The Latino Coalition describes itself as non-partisan, but one of Kerry’s spokespeople claims that they are tied to Republican interests. Indeed Deposada, president of the Latino Coalition, was director of Hispanic Affairs for the National Republican Committee (89-92) and headed G.H.W. Bush’s Hispanic publicity campaign in 1988.
    • Survey was conducted by Opiniones Latinas:
      • “Kerry’s lead among Hispanic Registered Voters is only nine points (47%-38%). At this stage in the game, the Democratic Presidential candidate should have been with a two-to-one lead and securing around 60% of the Hispanic vote. This is a serious problem for Senator John Kerry, if he is to carry key swing states with large Hispanic populations like New Mexico, Arizona, Colorado, Nevada and Florida.�?
      • Majority of Latinos last year thought economy and unemployment to be key electoral elements; now it is national security.
      • 50% see Bush as decisive leader, 34% have that opinion of Kerry.
      • Registered to vote or not, 59% of Hispanics identify more with the Democratic Party and 29.2% with the Republican Party.
      • 54% believe Kerry would do a better job than Bush in managing the economy and creating jobs.

Why do these surveys differ?

  • Methodology:
    • Survey 1 was conducted among Latino registered voters in 11 states. Survey 2 was conducted among the general Hispanic population (it is not clear what states, regions were involved).
    • Survey 1 was conducted by “academics�? versus (?) in Survey 2.
    • There is a gap of over two months between both surveys. The campaigns are getting more aggressive as November 2nd approaches; thus possibly influencing voter opinions.
    • Survey 1 does not specify the language used in the interviews. Survey 2 used English and Spanish.
  • Who is behind the survey?
    • Survey 1’s sponsors may be considered somewhat neutral. Although from my personal experience watching Univisión TV news and reading editorial pieces by its two top anchors, Salinas and Ramos, it is evident that they tend to favor Democratic values and interests.
    • Given Deposada’s political work, Survey 2’s sponsors seem to favor Republican values and interests.
  • Questions asked dealt with common issues: Healthcare, education, the war in Iraq, etc. Even though both groups seem to have different agendas, it is interesting to see that both surveys show that the majority of Latino voters identify with the Democratic Party. Where they seem to differ is on issues of national security, where those polled in Survey 2 found it to be more important than the economy. In addition, according to Survey 1, both candidates are considered to be strong leaders by over 50% of those polled, whereas according to Survey 2, Bush is seen as a more decisive leader. Thus, there seems to be a tendency by Survey 2 sponsors to correlate decisiveness as a leader with national security issues. Whereas in Survey 1, the majority of those polled (62%) did not find that Bush is doing a good job in this area. But then again, that was back in July…

Luz

Retrieved from "http://wik.ed.uiuc.edu/index.php/Presidental_survey_page"
Personal tools