Chapter 8
Survey Research

I. Introduction

A. Communication research concerns a wide range of situations, from intimate exchanges to mass appeals, from one-to-one interaction between members of a family to messages sent from one person to a large audience.

1. Survey research is particularly relevant to the latter end of this continuum.

2. Researchers use the survey method to ask questions about the beliefs, attitudes, and behaviors of respondents for the purpose of describing both the characteristics of those respondents and the population they were chosen to represent.

II. The Prevalence of Surveys

A. Surveys have been used almost as long as people have gathered into social groups.

1. The Romans used them as a census to tax people and the English, in 1086, conducted a survey to list all landowners.

2. Surveys came of age in the twentieth century.

a. Survey research is particularly useful for gathering information about populations too large for every member to be studied.

b. The survey method also is a relatively straightforward research strategy: You ask people questions and analyze their answers.

III. Applied Uses of Survey Research

A. Survey research has become a big business; survey research organizations grossed over4.4 billion in worldwide revenues in 1994 (J. K. Stewart, 1994a).

1. Surveys are most used as a tool in public opinion research, with three major applications.

a. Political polls: Results of political polls are presented to us routinely by the media, especially during election years.

i. Decades ago, political polls were held in disrepute after highly publicized election predictions were shown to be inaccurate.

ii. One reason today’s pollsters do make more accurate predictions on election days is because of the use of exit polls, asking people how they voted just after they cast their ballot.

b. Market research: Seeks to determine current, and predict future, levels of consumption of products and services, evaluate consumer satisfaction with the performance of existing products, predict consumer preferences for new products, and identify persuasive strategies for product packaging, pricing, and advertising.

i. Advertising readership surveys: Conducted regularly to identify differences in demographic characteristics between readers and nonreaders of a publication and to identify those who read advertisements for specific products.

ii. Editorial content readership surveys: Conducted to determine which articles newspaper and magazine subscribers like and don’t like and what topics they would like to see covered in the publication.

iii. Broadcast audience surveys: Identify the size and composition of the audience that television and radio stations and specific programs reach.

(a) A variety of measurement services are used by rating services, such as A. C. Nielsen.

iv. Market research is a dynamic field that is constantly developing new research techniques.

(a) Usability testing: Involves watching potential users interact with a new product under carefully controlled conditions, such as in a laboratory (see Chapter 7).

(b) Contextual inquiry: Involves studying customers’ use of a product at their place of work.

© Evaluation research: Surveys are used to evaluate the effectiveness of specific programs or products by inquiring about audience members’ and customers’ experiences and feelings.

(i) Formative evaluation: Conducted while a program or product is in the process of being developed to identify ways to refine it.

(ii) Summative evaluation: Conducted after a program or product is completed to learn its overall effectiveness, usually to determine whether to continue or discontinue it.

(iii) Need analysis: Research to identify specific problems experienced by a target group, usually by identifying gaps between what exists and what is preferred, as well as to test reactions to potential solutions to those problems.

(iv) Organizational feedback surveys and audits: Organizational members and representatives of groups those organizations serve are questioned about current or potential opportunities and constraints facing the organization.

v. Network analysis: Used to identify the patterns of interaction within a social system by gathering information about who system members communicate with and how often, either by surveying system members via questionnaires or interviews or by observing them.

IV. Use of Surveys in Communication Research

A. Many contributions to our understanding of people’s communication behavior have been made by scholars who use the survey method.

1. Figure 8.1 presents some recent examples of how researchers use the survey method to study communication.

2. When studying relationships between variables, survey researchers rarely exercise the same amount of control as experimental researchers.

a. Correlational design: A questionnaire or interview is used to assess all the variables of interest at one point in time, and then analyze the relationships among them.

b. Survey research typically establishes noncausal relationships among variables.

3. One survey can sometimes provide data for multiple studies.

a. Reanalysis by one scholar of data collected by someone else is called secondary data analysis, to distinguish it from primary research collected by the original researcher.

V. Survey Research Design

A. Designing an effective survey study involves a series of challenging decisions centering on the sample’s selection; survey strategy such as phone or face-to-face; wording of the survey questions; and the amount of time for data collection.

B. Selecting Survey Respondents

1. To select a sample, researchers first identify the population they want to describe.

a. Sampling frame: A list of the population from which they will sample.

b. It may not be possible to obtain a list of all the population.

d. In the case of large cities, for example, survey researchers often use telephone directories as the sampling frame.

i. Random digit dialing: Helps solve the problem of unlisted phone numbers by randomly generating via computer all the possible combination of telephone numbers in a given exchange.

2. Sampling method is selected after the sampling frame is identified or created.

a. A random sample provides the best guarantee of generating an externally valid sample from a population.

b. Many large-scale surveys use cluster sampling to select respondents, and this procedure entails: Selecting states, counties within the state, cities within the counties, streets within the cities, houses on the streets, and finally, household members, all in a random manner.

c. Stratified samples help ensure that individuals with particular characteristics are selected, often in numbers proportional to their occurrence in the actual population.

3. Obtaining random samples is fraught with problems.

a. In a recent study, for example, it was reported that in 1990, the contact rate—the percentage of phone calls that result in contact with an English-speaking interviewee—was only 56.4 percent.

b. The average cooperation rate—the percentage of phone calls in which the interviewees agree to participate—was only 53.2 percent and 26 percent of the respondents had been interviewed too frequently by other researchers; referred to as the surveyed rate.

4. Survey researchers usually obtain responses from individuals, but individuals are not always the unit of analysis in the survey.

a. Individuals are often asked questions about a larger unit of analysis, such as their family or work unit, and cross-level inferences are drawn about this larger unit.

b. There is some danger involved in such generalizing and if an ecological fallacy exists—where the data collected from one unit do not describe accurately the larger unit of analysis—the study may have little value.

c. Researchers have to also be concerned with the response rate, that is, the number of usable responses divided by the total number of people sampled.

i. The response rate is crucial for evaluating survey research, for if there are important differences between those who did and those who did not respond, the results of the study may not be valid.

ii. There is no generally accepted minimum response rate, and researchers have to be concerned about nonresponse bias.

iii. Researchers try very hard to maximize response rates and some strategies include: people are more likely to respond if the topic is important to them, or they believe they are contributing to an important cause; monetary rewards can also be used; and confidentiality or anonymity is assured when reporting results.

iv. Despite such strategies, studies show that about 35 percent of adults refuse to cooperate with survey researchers (Looney, 1991).

v. Two practices that discourage people from participating in surveys are sugging and push- polling: The former involves selling under the guise of a survey and the latter involves asking questions in a way to discredit a person or product.

VI. Cross-sectional versus Longitudinal Surveys

A. Surveys that study respondents at one point in time are called cross-sectional surveys, and surveys that study respondents at several points in time are called longitudinal surveys.

1. Cross-sectional surveys are easy to do, used most often, and very effective for studying the status quo.

2. When evaluating the results from a cross-sectional survey , it is important to take into account the particular point in time when the survey was conducted.

a. The results of cross-sectional surveys can be misleading if done at unrepresentative point in time.

3. Longitudinal surveys help to overcome the limitations of cross-sectional surveys.

a. By gathering data from respondents at several points in time, longitudinal surveys are much more effective at capturing the processual nature of communication.

4. Three primary techniques are used to conduct longitudinal surveys.

a. A trend study measures people’s beliefs, attitudes, and/or behaviors at two or more points in time to identify changes or trends.

b. A cohort study gathers responses from specific subgroups of a population, usually divided on the basis of age, and compares the responses over time.

c. A panel study (also called a prospective study) obtains responses from the same people over time to learn how beliefs, attitudes, and/or behaviors change.

i. Panel attrition: This occurs when panel members cannot be replaced and fall out of the study, also known as mortality.

5. Researchers can sometimes combine cross-sectional and longitudinal survey designs into an accelerated longitudinal design to study individuals over time.

VII. Survey Measurement Techniques

A. Questionnaires and interviews are the primary measurement techniques employed by survey researchers; therefore it is essential to focus on question design and issues about using such self-report methods.

B. The value of survey instruments rests on asking effective questions.

1. Good questions elicit useful information; poor questions evoke useless information.

C. Asking the “right” question is always important and evaluation criteria depend on the purpose(s) or objective(s) of the survey.

1. Many types of questions can be asked (see Figure 8.2) and these can be grouped into two category schemes (see Figure 8.2).

2. The type of question asked influences the accuracy of people’s answers, especially when they call for recall of past events.

a. Forward telescoping: People tend to think that past events either occurred more recently then they did.

b. Backward telescoping: People tend to think that past events occurred longer ago than they actually did.

c. People also tend to report inaccurately incidents that are unpleasant or ego-threatening

D. Phrasing Questions: The way a question is phrased influences the kind of responses received.

1. Closed questions: Provide respondents with a limited number of predetermined responses from which to choose.

2. Open questions: Ask respondents to use their own words in answering questions.

3. Whether closed or open, the wording of questions needs to be appropriate for the specific respondents being surveyed.

a. Researchers must take into account respondents’ educational levels and cultural backgrounds and pose questions that use language with which they are familiar and, thereby, maximize the chances that the questions will be understood and answered.

b. Good survey questions are straightforward, to the point, and stated clearly.

c. To keep wording simple, or at least appropriate for the intended audience, researchers conduct a readability analysis to measure how easy or difficult it is to read a particular passage.

d. Good survey questions should ask about one and only one issue.

e. Double barreled questions ask about several issues at once and should be avoided.

f. The responses to closed questions should also be limited to a single referent.

g. Good survey questions are not loaded questions, that is, they don’t lead people to respond in certain ways.

h. Good survey questions avoid using emotionally charged, inflammatory terms that can bias people’s responses.

i. Good survey questions avoid using double negatives.

E. Questionnaires and interviews are structured in a variety of ways, depending on the type and arrangement of questions used.

1. The tunnel format asks a series of similarly organized questions.

2. The funnel format begins with broad, open questions followed by narrower, closed questions.

3. The inverted funnel format begins with narrow, closed questions and builds to broader, open questions.

VIII. Questionnaire Survey Research

A. Questionnaires are commonly used in survey research to gather information from large samples.

B. There are two common types: self‑administered questionnaires and mail surveys.

1. Self‑administered questionnaires (two types): Those completed by respondents in the presence of the researcher; and those completed by respondents without the presence of the researcher.

2. Group administration of questionnaires is obviously more efficient than individual administration because a large number of respondents can be surveyed in a short amount of time.

C. There are three types of paper‑and‑pencil self‑administered questionnaires.

1. Traditional paper questionnaire is the most common form.

2. Scannable questionnaires* ask respondents to mark their answers on a special form that is scanned by a computer. Provides a question sheet and a separate answer sheet.

3. A customized scannable questionnaire provides the questions and answers on the same scannable sheet.

a. A scannable questionnaire must rely on only closed questionnaires and limits the number of categories from which respondents can choose.

4. Mail surveys provide an efficient and inexpensive way to reach representatives of large, geographically dispersed populations.

a. Must be self sufficient and user friendly.

b. Afford respondents more privacy and anonymity than questionnaires completed in the presence of a researcher.

c. Potentially jeopardizes the response rate and result in a much lower rate than other forms of questionnaires.

IX. Interview Survey Research

A. Structured interviews are the only other major tool of survey research.

1. Involves the interviewer (person conducting the interview) and the interviewee (person who is being interviewed).

2. The degree to which the interviewer establishes rapport with interviewees influences somewhat how much and what kind of information interviewees offer in response to questions.

B. Training interviewers is important as the quality of the data gathered in the interview is largely dependent on the skills of interviewers.

1. Many researchers train interviewers before the actual case study because unskilled interviewers can destroy the validity of a well‑designed interview, or can impugn the reputation of the research organization that sponsors the survey.

2. All interviewers should be educated about goals and methods of the survey so that they can answer respondents’ questions accurately.

3. Interviewers should become familiar and comfortable with the interview schedule and be given opportunities to practice before the actual interview.

4. Directive interviews use more closed questions and demand only that interviewers read questions to respondents clearly and accurately and record their answers.

5. Nondirective interviews, in contrast, use more open questions and demand added sensitivity, flexibility, and communication skills.

C. There are relative advantages and disadvantages to face‑to‑face and telephone interviews.

1. The advantages of face‑to‑face interviews follow:

a. Interviewers have more cues available for identifying when interviewees do not understand questions, are unsure about their answers, or even when they appear to provide misleading answers.

b. With the advent of portable computers, it is also possible to set up question sequences based on respondents’ answers prior to the interview; see for example, computer‑assisted personal interviewing (CAPI).

2. The disadvantages of face‑to‑face interviews follow:

a. These take more time to gather data.

b. These can decrease the privacy and anonymity of interviewees, making it more difficult to gather valid data about personal, risky, or embarrassing topics.

3. The advantages of telephone interviews follow:

a. These can overcome some of the time, expense, and reactivity problems of face‑to‑face interviews.

b. These cost about 25 percent less than personal interviews.

c. These can reach interviewees over long distances and can increase respondents’ privacy and anonymity.

d. Computer‑assisted telephone interviews (CATI) selects and calls respondents automatically, cuing interviewers to the questions to be asked, and providing a simple mechanism for recording, coding, and processing responses.

e. Two new computer‑assisted telephone interviewing techniques don’t require a human interviewer: Touchtone data entry (TDE) and voice recognition entry (VRE) are both systems in which a “computer reads questions from a record and the respondent has to answer by use of the telephone” (Saris, 1991, p. 30).

4. The disadvantages of telephone interviews follow:

a. Amount of nonverbal information available to the interviewer is limited to paralinguistic cues, and thus, may make it more difficult to assess honesty of answers.

b. Limits interviewer’s ability to identify respondents’ demographic characteristics.

c. Answering some closed questions can be difficult.

D. In considering whether to use individual versus group interviews, the purpose(s) of the study help researchers make an informed decision.

1. Survey researchers usually interview respondents individually, unless their research purpose(s) call for respondents to be interviewed as a couple or group.

2. Focus group interviews: A facilitator leads a small group of people (5‑7) in a relatively open discussion about a specific product or program.

a. Focus group interviews encourages participation;

b. Make it easier for reluctant communicators to participate by encouraging people to “piggy‑back” on others’ ideas;

c. Are becoming increasingly popular for gathering data in communication survey research;

d. Are demonstrating effectiveness in developing survey questionnaires;

e. Can help generate information that can be used to design persuasive messages for communication campaigns.

E. Beginning the interview

1. How interviews are begun often determines whether, and how fully, respondents cooperate.

2. In‑depth interviews may require a great deal of rime and disclosure from respondents.

F. Planning the interview questions

1. Interviewers usually prepare the major questions they will ask in advance.

2. List of questions that guide the interview is referred to as the interview schedule or protocol.

3. Type of interview schedule depends on the purpose of the research.

a. Structured interviews require the interviewer to follow a preset list of questions the same way each time.

b. Semi‑structured interviews require the interviewer to ask the specific primary questions but then allow for probing secondary questions.

c. Unstructured interviews provide interviewers with a set of questions, but allow them maximum freedom to decide what questions to ask and how to phrase them.

G. Responding to answers

1. Interviews are influenced as much by how researchers respond to the answers they receive as by how initial questions are posed.

2. To obtain accurate and complete information, interviewers probe for more information when responses are too brief or general.

a. Probes often planned in advance, but can be improvised.

X. Using Multiple Methods in Survey Research

A. Questionnaires and interviews often are used together in the same research study.

B. Sometimes researchers conduct interviews after asking people to complete a questionnaire.

1. This is done when researchers want to learn more about questionnaire answers.

C. Survey research may help to verify findings that have been obtained using other methodologies.

D. Researchers, especially those who conduct longitudinal research, sometimes design a study to include both survey and other methods.

XI. Conclusion

A. What we learn from the survey method may be likened to what we see from the window of an airplane—an overview of a vast territory. We must be careful that we record and analyze accurately the terrain we are observing.

B. Many scholars are employing new technologies and more ingenious methodologies to overcome traditional survey limitations. They will be measuring more elusive phenomena and getting more accurate results as social science evolves in the twenty-first century.