+ All Categories
Home > Documents > A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Date post: 16-Apr-2017
Category:
Upload: fincher-consulting
View: 698 times
Download: 0 times
Share this document with a friend
79
A Journalist’s Guide to Survey Research and Election Polls Cliff Zukin * Rutgers University AAPOR The American Association for Public Opinion Research Washington Press Club Briefing 9/24/12 *This presentation reflects the author’s views. It has not been vetted or endorsed by AAPOR.
Transcript
Page 1: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

A Journalist’s Guide to Survey Research and Election Polls

Cliff Zukin * Rutgers UniversityAAPOR The American Association for Public Opinion Research

Washington Press Club Briefing 9/24/12

*This presentation reflects the author’s views. It has not been vetted or endorsed by AAPOR.

Page 2: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

First, can we trust polls: 2004?

http://www.realclearpolitics.com/epolls/2004/president/us/general_election_bush_vs_kerry-939.html2

Page 3: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Can we trust polls: 2008?

http://www.realclearpolitics.com/epolls/2008/president/national.html3

Page 4: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

What AAPOR Thinks You Should Know

1. Who did the poll? Who paid for it?

2. When was it done?

3. Who was sampled?

4. How were respondents contacted?

5. What is sampling error?

6. Why are data weighted?

7. How is the question worded?

8. How are the questions ordered?4

Page 5: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

What You Want to Know

1. How are margins of error determined, and what does the error rate mean?

2. To what extent are traditional polls being replaced by targeted cell phone or Internet polls?

3. Discuss Internet polls and their validity.

4. Any new polling techniques for 2012?

5. Explain weighting, and how subjective it is, and how to evaluate it.

6. What are the most common errors made by political journalists in writing about polls?5

Page 6: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Our Roadmap (Through the Methods Box. See the handout. )

• Sampling—why polls work: Good and Bad Samples

• Developments and Challenges facing the SR profession: cell phones, IVR (robo-polls), etc.

• Question wording & ordering

• Things you want to be wary of…SR for journalists

• Election Polling—How it’s different & why polls differ

6

Page 7: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

SAMPLING: The Science of Polling

http://faculty.elgin.edu/dkernler/statistics/ch01/3-1.html7

Page 8: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

All Scientific Polling is Based on the Notion of Sampling

• Bloody Whiners I’VE NEVER BEEN CALLED

• Polls are estimates

• A sample is drawn to represent an underlying population

• The sample must be representative

• If you don’t do this well, the rest doesn’t matter

8

Page 9: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Telling Good Polls and Bad Polls Apart Starts with the Sample

• Probability Samples are GOOD– Scientific—known chance of inclusion– Random, or thereabouts, selection– Generalizable from sample to population

• Non-Probability Samples are BAD– Not representative: 1,200 people are not a sample of anything– Person in the street– ALL Internet opt-in surveys– Any self-selection

9

Page 10: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Journalistic Standards

• In order to represent the population statistically, a survey should be based on a probability sample.

-NY Times

• Methodologically, in all or nearly all cases we require a probability sample, with high levels of coverage of a credible sampling frame. Self-selected or so-called “convenience” samples, including internet, e-mail, “blast fax,” call-in, street intercept, and non-probability mail-in samples do not meet our standards for validity and reliability, and we recommend against reporting them

-ABC/Washington Post

10

Page 11: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Probabilities, Better Known as The Odds: Why Polls are Accurate…We’re the House

Dice Total How you can make this total How many ways

to do it For every 100 times, you should get this

outcome this many X

2 1,1 1 33 1,2; 2,1 2 54 1,3; 2,2; 3,1 3 85 1,4; 2,3; 3,2; 4,1 4 116 1,5; 2,4; 3,3; 4,2; 5,1 5 147 1,6; 2,5; 3,4; 4,3; 5,2; 6,1 6 178 2,6; 3,5; 4,4; 5,3; 6,2 5 149 3,6; 4,5; 5,4; 6,3 4 11

10 4,6; 5,5; 6,4 3 811 5,6; 6,5 2 512 6,6 1 3

36 Total Outcomes 11

Page 12: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Dice Outcome(Or, the Central Limits Theorem)

12

Page 13: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Probabilities: Why Polls are Accurateor “We’re the House”

• Chances of getting a 7 = 17% average or mean

(of 100 rolls (100%)/ 6 ways = 17%)

• Chances of getting between 6 and 8 = 45% (14% + 17% +14% = 45%)

• Chances of getting a 5 through 9 = 67%(11% + 14% + 17% +14% + 11% = 67%)

• So, the chances of being within 4 points on either side of the expected (mean) is 95%. Or the margin of error is + 4 % points

13

Page 14: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Probability(good) Samples Depend on Mathematical Axioms: The Central Limits Theorem & the Law of Large

Numbers

1. Life (or a sample of life) is distributed normally and generally fits a bell-shaped curve Central Limits

2. The more observations you have, the more your distribution will approximate a bell-shaped curve Large numbers

3. After a certain point, sample size matters, but population size doesn’t. Law of M&Ms

14

Page 15: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

At a Secret Location in Hacketstown NJ…

15

Page 16: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

…I Counted the M&Ms in the Last Slide

Color Number

Tan 4

Brown 8

Red 9

Orange 10

Green 7

Yellow 2

Total 40

16

Page 17: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Relationship between Sample Size and Sampling Error

17

Page 18: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

• Sampling Error is a theoretical minimum• It is only one kind of error, but it is quantifiable• It applies not to the GAP between candidates,

but to each point estimate

Obama 47 Romney 44

Obama 47 ± 3 Romney 44 ± 3

Obama 44 + 50 Romney 41 – 47

MOE ± 3

Sampling Error

18

Page 19: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

SO WHAT? Why is this Important? • INTERNET POLLS ARE NON-PROBABILITY SAMPLES

(WITH ONE EXCEPTION)

• It is not possible to calculate sampling error on a Non-Probability Sample

• And even with lipstick, a pig is still a pig

Ipsos Poll for Reuters DAILY ELECTION TRACKING 9/11/12:These are findings from an Ipsos poll conducted for Thomson Reuters from Sept 7-

11, 2012. For the survey, a sample of 1,269 American registered voters (age 18 and over) was interviewed online. The precision of the Reuters/Ipsos online polls

is measured using a credibility interval. In this case, the poll has a credibility interval of plus or minus 3.1 percentage points for all respondents.

19

Page 20: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

It’s Not Just Sampling Error:Total Survey Error Comprises 4

Sources• Sampling Error: Sample, not population

surveyed

• Coverage Error: Sample may not map to pop

• Measurement Error: Question wording/ordering

• Non-response Error: Many decline to be surveyed

20

Page 21: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Trends in Survey Research

Page 22: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Trends in the Survey Research Industry

• Cell phones

• IVRs (Robo-Polls)

• Fewer high-quality polls

• Challenge of interviewing a representative sample

• Increased cost of finding respondents and dialing

22

Page 23: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

*Data Source: CDC/NCHS National Health Interview Survey

Percentage of Wireless Only Adults 2001-2005

23

Page 24: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Percentage of Wireless Only Adults 2001-2011

*Data Source: CDC/NCHS National Health Interview Survey 24

Page 25: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Percent of Adults Ages 18-29

Source: Current Population Survey and Pew Research Center surveys25

Page 26: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

RDD Samples without Cell Phones – Age Bias

26

Page 27: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Phone Status of Adults, Dec 2011

Among those households with telephones…

•There are four times more “cell phone only” (32%) in the population as “landline only” (8%)

•Taken together, fully half of the public is “cell phone only” (32%) or “cell phone mostly” (18%)

•Did you know: Most IVR Polls do not include cell phones in their samples? This is definitely a question you should ask before reporting

27

Page 28: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Journalistic Standards: NY Times

• Interactive Voice Response Polls — Interactive voice response (IVR) polls (also known as "robo-polls") employ an automated, recorded voice to call respondents who are asked to answer questions by punching telephone keys.

The Times does not publish IVR polls.

• Internet Polls — Non-probability samples are commonly used in Internet polls, call-in polls, blast e-mail polls and a variety of others. The entire population does not have an equal chance of being contacted. Most Internet polls are based on panels of self-selected respondents.

The Times does not publish most Internet polls.

28

Page 29: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Manipulating Public OpinionOr can you make a poll find anything you want?

NO!

BUT question wording and question order are extremely

important

29

Page 30: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Good Questions:

1. Are simple, direct, clear to all, and avoid jargon

2. Don’t presume information

3. Are balanced

4. Ask about only one thing

5. Don’t tax respondent’s memory or cognitive ability

30

Page 31: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Bad Questions:

1. Are complex or presume information

2. Are leading or unbalanced

3. Are double-barreled or double negative

4. Are loaded through emotional or red flag words

5. Give biasing or unequal information in the Q stem

31

Page 32: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Bad Questions:

1. Are complex or presume information:

Do you favor or oppose a 1 percent tax by the United Nations on international air travel which by itself would produce half of all the revenue the organization needs and a savings for all members, which for the United States would be about half a billion dollars a year. The tax itself would add about a billion dollars to America’s international travel and air freight bills. Do you favor or oppose that?

32% Strongly favor 30% Somewhat favor 11% Somewhat oppose 25% Strongly oppose 2% Don't know/No answer

32

Page 33: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Bad Questions:

1. Are complex or presume information:

President Obama has announced changes to federal student loan programs that would: allow some college graduates to limit federal student loan repayments to 10% of their discretionary income starting in January, two years before the cap was due to take effect under federal law; forgive remaining debt on the federal loans after 20 years, five years earlier than under current law; and allow those with more than one student loan to consolidate their debt. Do you approve or disapprove of these changes?

33

Page 34: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Getting Real:Citizen Knowledge and Attentiveness

• 17% follow “news about political figures and events in Washington very closely”

• 28% named John Roberts as Chief Justice out of four choices (7/10)

• 43% know Republicans control the House (11/11)

• 53% Know it is the Republicans who want to reduce the size of government

Source: Pew Research Center 34

Page 35: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Bad Questions:

2. Are leading (or unbalanced):

Are you in favor of the mandatory drug testing of professional athletes? BIASED

Rather,

Do you favor or oppose the mandatory…UNBIASED

or… professional athletes, or not? UNBIASED

35

Page 36: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Bad Questions:

Totals may not add up to 100% due to rounding.

2. Are leading (or unbalanced):

Should the federal government increase its role in providing loans with a goal of making sure everyone who puts in the effort to graduate from college can afford to do so?

• It’s okay for you to ask leading questions of experts, but not for us to ask them of random folks

36

Page 37: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Bad Questions:

3. Are double-barreled or double negative:

Did you vote in the 2008 and 2010 elections – yes or no?

37

Page 38: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Bad Questions:

3. Are double-barreled or double negative:

Do you want to see less money spent on defense and more on social problems?

38

Page 39: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Bad Questions:

3. Are double-barreled or double negative:

(Agree or disagree) Should the federal government no longer be involved in college loans, and instead leave that entirely to the private sector?

39

Page 40: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

4. Are loaded through emotional or red flag words

Fifty state legislatures have passed resolutions calling on Congress to pass the flag amendment and send it to the states for ratification. Do you think members of Congress who may personally oppose the amendment should vote in favor of it anyway so that “we the people” can decide the issue?

Bad Questions:

40

Page 41: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

4. Are loaded through emotional or red flag words

Do you believe in killing unborn babies?

Bad Questions:

41

Page 42: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

5. Through giving one-sided information

If it would result in increased opportunities for educating New Jersey citizens, would you favor or oppose building a new TV transmitter at liberty science center?

Bad Questions:

42

Page 43: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

5. Through giving one-sided information

Many gambling experts believe that internet gambling will continue no matter what the government does to try to stop it. Do you agree or disagree that the federal government should allocate government resources and spend taxpayer money trying to stop adult Americans from gambling online?

Balance is necessary: Some people say…while others say…

Bad Questions:

43

Page 44: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

ALASKA, or (I’ll Ask her)

• Zogby: for environmental interests

Do you think oil companies should be allowed to drill for oil in America’s Arctic National Wildlife Refuge?

Allow 38% Not allow 55%

17 percentage points against drilling

44

Page 45: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

ALASKA, or (I’ll Ask her)

45

Page 46: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

ALASKA, or (I’ll Ask her)

46

Page 47: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Maxims of Question Wording

1. Question wording matters most when respondents know the least

2. Projections and hypotheticals are generally unreliable

3. You can never really underestimate how much people know about politics and government

4. Giving information before asking the question is a double-edged sword

5. It must be balanced--as easy to disagree as to agree

6. No ONE question tells the story

47

Page 48: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

CONTEXT EFFECTSEach Question Affects the Following Ones

• How important is Social Security to you?

• Do you think Social Security will be there when you need it?

• Are you worried about private investment options for Social Security?

• What is the most important problem facing the country?

48

Page 49: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

WHAT’S THE TOPIC HERE….

1. “The First Amendment of the United States applies to everyone regardless of gender, race, religion, age profession, or point of view.” (96%)

2. “The First Amendment protects the right of individuals to create a private organization consisting of a specific group of people based on age, gender, race, ethnicity, or interest.” (77%)

3. “The First Amendment protects the right of organizations like the Boy Scouts, the Girl Scouts, and the National Association for the Advancement of Colored People to exist.” (91%)

49

Page 50: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

IT’S BEEN IN THE NEWS RECENTLY….

4. “Individuals have a right to join a private group, club, or organization that consists of people who share the same interests and personal backgrounds as they do if they so desire.” (91%)

5. “Private organizations that are not funded by the government should be allowed to decide who becomes a member and who does not become a member on their own, without being forced to take input from other outside people or organizations.” (76%)

50

Page 51: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

The Masters Polling Co…

6. “Private clubs that have members only of the same gender are simply a harmless way for similar individuals to get together and associate with each other.” (71%)

23. “Although currently there are no women members of the Augusta National Golf Club, the Golf Club does allow women to play on their golf course, and visit the course for the Masters Tournament. In other words, women are welcome to visit the Club and they often play as guests.”

“Knowing this, would say you support or oppose the Augusta National Golf Clubs decision to keep their membership policy as it is?” (60%)

51

Page 52: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Some Common Problems in Reporting Polls

Page 53: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

What Do I Write if it’s 50 – 46?

AP Style Book:

•If the difference between the candidates is more than twice the sampling error margin, then the poll says one candidate is leading.

•If the difference is less than the sampling error margin, the poll says that the race is close, that the candidates are "about even."

– (Do not use the term "statistical dead heat," which is inaccurate if there is any difference between the candidates; if the poll finds the candidates are tied, say they're tied.)

•If the difference is at least equal to the sampling error but no more than twice the sampling error, then one candidate can be said to be "apparently leading" or "slightly ahead" in the race.

53

Page 54: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Be Wary of the “Slight” or “Modest” Trends

From the January 14, 2005, Hartford Courant:

• …Rell's stellar approval rating crept even higher in a poll released Thursday, rising from 80 percent to 83 percent in the first survey since her cancer surgery and State of the State address….

• Telephone poll of 1,287 voters from 1/7-10/05. Margin of error of +/- 2.7 percentage points.

54

Page 55: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Was Jesus married?

Beware Small Groups and Subgroups: Size Does Matter

55

Page 56: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Inadequate Sample Size: Independent Soccer Mom Voters in New Jersey

56

Page 57: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

What’s the Difference?

7 79

7 79 9

1112

1012

15

11

14 14

17

02468

1012141618

1987 1988 1990 1994 1997 1999 2002 2003

Political and Policy Attitudes Social and Personal Attitudes

Republicans and Democrats are Further Apart than Ever: Average Difference in Republican and

Democratic Attitudes

57

Page 58: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

An Alternative Perspective:

0102030405060708090

100

1987 1988 1990 1994 1997 1999 2002 2003

Political and Policy Attitudes Social and Personal Attitudes

Republicans and Democrats are Further Apart than Ever?

58

Page 59: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Comparing Two Different Polls

“After a day of action at the Republican convention in Tampa, GOP presidential candidate Mitt Romney is looking better to voters. Romney’s image has received a five-point bump after the convention’s first day, according to data presented at a Wednesday breakfast sponsored by The Hill at the law offices of Holland & Knight.”

“Romney had a 43% favorable and 44% unfavorable rating in nine battleground states heading into the convention, according to the average compiled by Real Clear Politics. A survey conducted by Wilson Perkins Allen Opinion Research in nine battleground states Tuesday evening found Romney’s favorable rating among likely voters to have jumped to 48%. His unfavorable rating dipped to 39%.”

59

Page 60: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Beware the Costless Question:

Would you favor or oppose preserving more open land in New Jersey?

• Favor• Oppose• Depends how much (VOL) • Don´t Know / Refused (VOL)

60

Page 61: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Don’t Report Answers to Silly Questions

“Sixty-seven percent of American Catholics believe John Paul II was the best pope the church has ever had, according to a new

CNN poll of 254 Catholics.”

61

Page 62: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

About WEIGHTING

• It’s necessary in almost all surveys

• Everyone does it

• It corrects for the problem of not interviewing people in the sample in correct proportion to their size in the population

62

Page 63: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Weighting Example

Education PopulationLess than High School

10

High School 25Some College 35College Graduate 30

Sample5103550

Weight2.02.51.00.6

63

Page 64: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Limitations of Weighting

• You can only weight to KNOWN population parameters

• Weighting to “attitudes” (like Party ID) is risky—they are not fixed attributes

• You can’t weight to something in the future (like turnout)

64

Page 65: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Election Polling

A Special Case

Page 66: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

ELECTION POLLING: A Special Case

• Time and field dates/procedures

• Type of sample used: random or listed

• Resp. selection/screening; Identifying likely voters

• Question wording and ordering

• Weighting

66

Page 67: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Time, Field Dates and Procedures

• Polls are a snapshot (yada yada)

• Number of days in the field: – Callbacks and refusal conversions

• Tracking polls/rolling averages

67

Page 68: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

• Random digit dialing (RDD)

• Addressed based sampling (ABS)

• Listed based sample (LBS)

Types of Samples: Advantages and Disadvantages

68

Page 69: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Source: Pew Research Center 69

Page 70: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Respondent Selection, Screening and Determining “Likely Voters”

• Hard screen to get into the survey, or remove low likelihood of voting respondents after the fact?

• The problem: There is a consistent over-report of intentions to vote by the public

What do you do if 80 percent of your eligible voters tell you they will definitely vote for President, but you’re looking at the table on the right?

Election Year Percent of Eligible Voters

1992 60.61996 52.62000 54.22004 60.12008 61.1 70

Page 71: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Pew Likely Voter Questions

• How much thought have you given to the coming presidential election?

• How closely have you been following news about the candidates?

• Do you plan to vote in the presidential election? How certain are you that you will vote?

• Rate your chance of voting in November on a scale of 10 to 1, with 10 being “definitely will vote” and 1 “definitely will not vote.”

• How often do you follow what’s going on in government and public affairs?

• Have you ever voted in your precinct or election district?

• How often would you say you vote?

• Do you happen to know where people in your neighborhood go to vote?

• In the last election, did things come up that kept you from voting or did you vote?

71

Page 72: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Likely Voter Computation

Score

Percent (%)

9 408 87 76 125 204 153 42 31 10 0

Target

72

Page 73: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Likely Voter Vote Division

Candidate Turnout(All)

Obama 50Romney 40

Don’t know 10Margin 10

Top 67% Top 55%

50 5343 48 7 47 5

73

Page 74: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Question wording and ordering

• Asking the vote question: The presidential candidates and ….?

• …VP candidates, • …party labels, • …randomized names, • (3rd pty if applic)?

• Where the vote question is put: What came before it that it will be reactive to?

74

Page 75: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Weighting

• To what parameters, even on demographics?

• Weighting on party is controversial; Best practices” says this is NOT a good idea

• But what’s a pollster to do if faced with:

75

Page 76: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

The Special Case of Party ID

Party ID Phone IVR Internet

Democrat 33% 36% 46%

Independent 35% 29% 15%

Republican 30% 34% 36%

Totals 98% 99% 97%

* Mark Blumenthal @ Huffington Post August 24, 2012 76

Page 77: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Why Election Polls May Vary by a Few Percentage Points

• Sampling error• Length of field period• Live interviewers vs.

IVR• Type of sample used• Mode of administration• Respondent selection• Likely voter indices• Question wording &

ordering• Weighting

Actually, it’s a wonder they are as close as they are!

77

Page 78: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

Know Your “Rights”AAPOR Minimum Disclosure Standards

• Name of the survey sponsor

• Name of organization that conducted the survey

• Exact wording of questions being released

• Definition of population under study

• Description of sampling frame used to represent population under study

• Explanation of respondent selection

• Total sample size• Method or mode of data

collection• Dates and location of

data collection• Estimates of sampling

error (if appropriate)• Description of data

weighting• Use of sub-groups

disclosed

78

Page 79: A Journalist’s Guide to Survey Research and Election Polls by Cliff Zuskin

We’re (not from the government) Here to HELP

• AAPOR– 3 Presidents

• Paul Lavrakas [email protected] President• Scott Keeter [email protected] Past President• Rob Santos [email protected] President-elect

– Rapid Response (Election Polling) Team• Quin Monson [email protected]• Mike Traugott [email protected]• Rob Daves [email protected]• Cliff Zukin [email protected]

79


Recommended