Thursday, January 13, 2005
By Brad Bannon
Because of declines in urban tax bases and the increasing demand to provide city services to needy populations, mayors are always on the hot seat and face a unique set of pressures. For this reason, mayoral campaigns are increasingly competitive. Polling in mayoral contests also present a unique set of challenges which are worthy of treatment that you might not get attention in a general discussion of polling techniques.
MORE AFTER THE JUMP…One thing to keep in mind is that there is no free lunch in good polling. You can get a political science professor at the state university to write a questionnaire and to process the data and you can have your friends and volunteers do the interviewing but chances are that you will get back bad data that will lead to bad tactical and strategic decisions that might cost you a victory and a lot of money.
Do not attempt to poll within the privacy of your own home. Hire a trained professional who will cost your campaign some money at the front end but save you from spending a lot of money foolishly at the back end. And keep this advice handy when you discuss the process with your pollster.
SAMPLING AND INTERVIEWING
Since mayoral contests often take place in odd number years when municipal campaigns are the only game in town, turnout is usually low which presents unique sampling problems.
The first step is to make sure you have a good list. If you sample off a list that has good phone numbers for only half of the voters, you can do anything else right and you will get bad numbers. Garbage in is garbage out. Your pollster can probably obtain an up to date voter list if you do not have one.
The next step in the sampling process is determining the type of voter screen to use to select the voters that you interview. Screen tightly but not too tightly. It would be a mistake for example just to interview voters who have voted in past municipal elections.
Your campaign should be bringing new supporters into the electorate and if you rely on past voters, you miss people who haven’t voted in the past but might vote because of the turnout efforts of you and your opponent.
The best thing to do is to start with a base of all registered voters in the city and to screen for people who will probably vote or are certain to vote in the election. You will reel in just about any “good” voter with this net and you can examine this past voter group as a sub sample of the possible electorate. You can use the entire sample of likely and probable voters as a high turnout scenario and the smaller sample of past voters as a low turnout situation.
The next question is the size of your sample. A survey of 400 likely voters is a good place to start in a medium sized city and gives you a margin of error of about plus or minus 5% which means that you can be 95% confident that if you had interviewed every voter in the city, the actual total from that census would not differ more than 5% from the results of the survey. But you may have to interview more voters to do precise geographic or demographic targeting. If there are 6 wards in the city and you want to compare the responses of voters in each of the 6 wards to each other, than you should have at least 75 interviews out of each ward for a total of 600 interviews.
There are two rules for conducting the interviews. First, have professional interviewers conduct the surveys. There is a skill to getting voters to open up and talk and your campaign volunteers will not know enough about the dynamics of a successful interview to do the job correctly. The other danger in using volunteers is that they will consciously or not ask questions in a way that will bias the results in your direction. The second rule is that you should always have the capacity to interview voters in their native language for the many people in any city who don’t know much English. You should as much as possible match interviewers and respondents by race.
QUESTIONNAIRE DESIGN
There are some questions that everyone knows you just have to have in the survey. You need questions that measure the personal favorability of the incumbent mayor and the opposition candidates. You also need to measure the job performance of the incumbent mayor and get a clean head to head match up between the all of the candidates. To measure the mood of the voters, you need to find out weather they think the city is headed in the right or wrong direction. These basic questions should be at the beginning of the survey, so you can warm up respondents before you get to complex questions that require more thought.
Mayoral campaigns are equal parts issues and images. Voters make decisions by listening to the candidates talk about city services and issues in order to make judgments about the character of the candidates running for mayor. So you need to write a survey that measures both factors.
The best way to get a handle on the issues in the campaign is to read a list of municipal problems to voters and find out how often they worry about each problem. The list of issue concerns might look something like this:
Now, for each of the following issues, please tell me whether you worry about it VERY OFTEN, FAIRLY OFTEN, ONLY OCCASIONALLY or NOT AT ALL? First, take… [ROTATE STATEMENTS]
VERY FAIRLY ONLY NOT AT DK/NA
OFTEN OFTEN OCCAS ALL EF
A. The failure of schools to
prepare children for the future -1 -2 -3 -4 -5
B. Inadequate bus service -1 -2 -3 -4 -5
C. Poor condition of the streets
And sidewalks -1 -2 -3 -4 -5
D. The amount of real estate
taxes that you pay -1 -2 -3 -4 -5
E. Corruption in city government -1 -2 -3 -4 -5
F. Things getting run down in your
neighborhood -1 -2 -3 -4 -5
H. Crime, drugs and gangs
in your neighborhood -1 -2 -3 -4 -5
In my opinion, the best way to determine how much impact an issue will have on a voting decision is to find out how often a voter worries about an issue. I f a voter worries about an issue very often, then that issue is going to be a key issue in the campaign. Measuring a behavior is much more effective than trying to gauge process by asking a voter what issue they think the mayor should address. Also, avoid open ended questions like “what is the most important issue facing the city.” Voters give open ended responses to open ended questions and the answers are not usually specific enough to give you much guidance.
One of the advantages of polling in a mayoral contest is that is that a lot of voters will not only know the mayor but also know the opposition candidates. Therefore you can identify the strengths and weakness of the candidates by asking them direct questions instead of hypothetical questions like “would you be more likely or less likely to vote for Jones if you knew that she was full of hot air” Once you know how mayor many voters believe the mayor is full of hot air, you can determine the importance of that factor by examining the statistical correlation between that perception and voting behavior.
I am going to read you a list of words which people use to describe political figures. For each word or phrase, tell me whether you think it best describes Joe Jones, Sue Kelly or Tony Cruz.
(ROTATE ORDER OF READING PHRASES)
JONES KELLY CRUZ ALL NONE DK/NA
A. Is inexperienced -1 -2 3 -4 -5 -6
B. Is a strong manager -1 -2 3 -4 -5 -6
C. Is dishonest -1 -2 3 -4 -5 -6
D. Cares about people -1 -2 3 -4 -5 -6
E. Is tough on crime -1 -2 3 -4 -5 -6
F. Will get things done -1 -2 3 -4 -5 -6
G. is full of hot air -1 -2 3 -4 -5 -6
ANALYIS AND INTERPRETATION
Analyzing and interpreting the results of a survey is a lot like assembling a jigsaw puzzle. In the puzzle you have hundreds of pieces of colored cardboard which don’t show anything until you put the pieces together correctly. In the survey, you have thousands of numbers that are meaningless until you put them together in a systematic way in order to see what the big picture is.
The analysis of a poll is a search for patterns and the explanation of variance. Let’s say that three out of every five voters like the incumbent mayor, think she is doing a good job and plan on voting for her. That constitutes a pattern. But only two out of every five voters think that the city is heading in the right direction which is a variation to the pattern. The mayor starts out with enough support to win reelection but many of her supporters, especially voters over 60 are uneasy about the condition of the city and might jump ship once they hear the challenger argue that better times will come after he wins the campaign. At this point the mayor has to either convince seniors that city is better off then they think it is or demonstrate that the challenger doesn’t have what it takes to turn things around.
The analysis process is nothing more than organizing a lot of data in a way that the pollster and client can see patterns and variances in the data. In the analytical stage, the pollster examines the relationships between the questions in top line or overall results and also looks at variances in the answers in each question from one voter subgroup to another in the cross tabulations.
A set of cross tabs for a poll in a mayoral race might look something like this.
All city voters
Good voters
Ward 1
Ward 2
Ward 3
Ward 4
Ward 5
Ward 6
Anglo
Latino
African American
Owners
Renters
Women under 40
Men under 40
Women 40-59
Men 40-59
Women 60 and over
Men 60 and over
Democrats
Republicans
Independents
But it is the interpretation of the data that separates the men from the boys and the women from the girls in this business. I can’t begin to tell you everything that I have learned after being a pollster for a quarter of a century but your pollster should be able to interpret the data to answer these three questions once the survey process is complete:
1. What are voters thinking or feeling?
2. Why are voters thinking or thinking whatever it is they are thinking and feeling?
3. How do you talk to voters to get them to change the way they are thinking and feeling and move them in your direction?
The answer to the last question is essentially is the campaign message. And once your have a message everything else should begin to fall into place.
Bannon is president of Bannon Communications Research which for 20 years has designed poll driven messages for Democrats, labor unions and issue groups. You can reach him at Brad@BannonCR.com with questions, comments or complaints.
