Describe the process of conducting survey research.

Question

What will be an ideal response?

Answer 1

Survey research consists of three major phases: planning, administration, and analysis. The planning phase involves selecting a topic, reviewing the literature, selecting the unit of analysis, assembling the survey instrument, selecting the survey mode, selecting the interviewers, training the interviewers, and conducting the interview.

To select a topic, researchers decide on a topic of interest, select a problem within the identified topic, and formulate general hypotheses or research questions based on the selected problem. Reviewing the literature involves locating and reviewing relevant background literature, including journal articles, books, and other published and unpublished materials, to determine what is known about the topic of study and what information remains unknown. During this review, the investigator clearly states the research objectives, refines the problem, and specifies the research hypothesis or question in operational terms. Investigators also identify existing measures, scales, or instruments, and evaluate them so that they may be incorporated in or adapted for the prospective survey.

The survey unit of analysis may be any unit that can be counted, including individuals or groups of individuals such as households, organizations, and cities. The choice of unit of analysis is determined by the research objectives, specified hypothesis, or research question. The survey instrument or questionnaire may be constructed by the researchers themselves or adapted from available instruments used by others. The variables selected depend on which characteristics should be studied to meet the research objectives.

When selecting the survey mode, researchers may use the self-administered questionnaire survey or the interview survey. The choice of survey mode depends on a number of factors, including the research topic, respondent size, sample size, and resources available to researchers. The sampling design for a survey depends on the characteristics and distribution of the population of interest, the availability of the sampling frame, and the survey mode chosen, as well as the overriding factor of cost. Prior to data collection, investigators must submit their study proposal for IRB review and approval. Informed consent is required for survey studies, and the statement of consent must include detailed information on the benefits and harms to study participants, the rights of anonymity and confidentiality, and how the participants’ protected health information is kept private. Health services researchers must understand and comply with HIPAA rules and related regulations for research.

Regardless of how interview surveys are conducted, their critical components are selecting the interviewers, training the interviewers, and conducting the interview. The ideal interviewer has a high sense of responsibility, a pleasant personality that gains respondents’ trust and cooperation, an interest in the research topic and in talking with people, an absence of prejudices toward or against the respondents, an ability to listen carefully and to speak articulately, and the skill to legibly and accurately record responses. In training interviewers, researchers first provide information about the study’s purpose, specific objectives, significance, sponsor, sampling method, planned uses, guidelines, and procedures. The interviewer should be trained to identify any perceived difficulties that the participants have in responding to the questionnaire items. Interviewers are then acquainted with the interview schedule or questionnaire, item by item. The interviewer must be familiar with the questionnaire instructions and specifications, must be able to follow the instructions, and must be able to determine when questions will need to be tailored to fit a given respondent’s situation. Interviewers will need to learn how to record responses, and should be taught basic interview techniques and rules for how to establish rapport with respondents and gain their cooperation, ask questions and probe in an unbiased manner, record observations, and handle interruptions and digressions. Demonstrations show interviewers exactly how to conduct the interview.

After obtaining the consent of respondents, interviewers conduct the interview immediately or schedule a later interview at the respondent’s convenience. Interviewers should dress properly, remain courteous, tactful, and nonjudgmental, and try to communicate a genuine interest in getting to know the respondents. Information gathered in the course of the interview should not be revealed to anyone except the interviewer’s supervisor. Throughout the survey, researchers should remain available to interviewers to answer questions and provide assistance. Interviewers should return completed questionnaires to the supervisor who would check completed questionnaire instruments to ensure that the survey questionnaire is properly filled out.

The administration phase involves pre-testing the survey instrument, obtaining access to respondents, administering the survey, and following up on nonrespondents. A pre-test consists of selecting a small but representative group of people and administering the survey instrument to them. Probability sampling is not required in selecting the pre-test group. Instead, the key is to choose those who share the major characteristics of the target population. Respondents are typically followed up on and asked for their opinions regarding the instrument, including its clarity, length, word choice, and any missing items. This helps identify areas requiring revision.

When respondents are from organizations, researchers must get permission from organizational administrators to obtain access to them. Endorsement from those who have an appeal to the respondents is also useful. A cover letter introducing the purpose and significance of the study may also facilitate access. In interviews, the cover letter is usually read to the respondent. In mail questionnaires, the cover letter is sent with the questionnaire as a separate sheet or attached to the questionnaire. Internet-based questionnaires should also include an electronic version of a traditional cover letter. Surveys may be administered by mail, in person, via telephone or Internet, or some combination of these methods.

The final step of field administration involves following up on nonrespondents to encourage their participation and increase the response rate. Methods of enhancing the response rate include shortening the questionnaire, obtaining sponsorship by a recognized authority, using a novel and appealing format, and incentivizing respondents with cash or other rewards. Regardless of the alternatives used, a follow-up mailing is a required sequence for increasing return rates, particularly in large-scale mail surveys.

Follow-up methods vary by survey modes. Mail questionnaires are usually coded so that researchers can tell who has not responded. Each follow-up mailing includes a new questionnaire and a cover letter that explains why each respondent’s cooperation is important to the validity of the findings. In practice, a total of three mailings (an original mailing and two follow-ups) seems to be the most efficient approach. In telephone interviews, follow-up procedures include calling back at different times of the day and/or different days of the week. In face-to-face interviews, neighbors or others who may know the whereabouts of respondents may be asked about when the respondents are usually at home or how they might be contacted. In Internet surveys, emails may be sent until responses are returned.

The analysis phase involves processing and analyzing the data. Data processing entails correcting data-inputting errors, dealing with missing values, assigning sample weights, and recoding variables into conceptual categories to facilitate data analysis. When surveys have large missing values, imputation is often used to minimize their effect on data analysis. Commonly used methods include deductive imputation, cold-deck imputation, hot-deck imputation, statistical imputation, and multiple imputation. Deductive imputation looks at information from other sections of the questionnaire to fill in the missing values. Cold-deck imputation applies group estimates, such as means, to assign values to the individuals with missing values. Hot-deck imputation assigns the actual responses of other individuals with similar characteristics to the individuals with missing values. Statistical imputation uses statistical procedures for the assignment of missing values. Multiple imputation generates multiple simulated values for each incomplete piece of data, then iteratively analyzes data sets with each simulated value substituted in turn.

Before analysis is conducted, the sample data must also be weighted if subgroups of the sample were selected at different rates. Weighting involves assigning varying statistical weight to groups so that the sample distribution mirrors that of the population. To adjust for disproportional sampling, the weight for each observation is computed based on the inverse of the sampling fraction. Data analysis applies statistical methods to survey data to test hypotheses or answer questions in research.