Mode of Data Collection

Within the context of survey operations, mode refers to an employed method or approach used for the collection of data. For example, surveys may be conducted face to face, by telephone, mail, or Internet (the four most commonly used modes), or through other types of approaches (such as interactive voice response [IVR], disk-by-mail, etc.) or combinations of modes. Modes can differ along a number of dimensions, including whether an interviewer is present, how the questions are presented and the responses recorded, the infrastructure required, field time, and costs.

One of the primary distinctions between modes of data collection is the presence or absence of an interviewer. When an interviewer is present, the survey questions are generally read to the respondent, and the mode is referred to as interviewer-administered. Telephone and face-to-face (in-person) surveys are examples of interviewer-administered data collection. When an interviewer is not present and the respondent must deal directly with a paper or electronic questionnaire, the mode is generally said to be self-administered. Examples of these include mail and Internet-based surveys.

The method of presenting the questions and receiving the responses also defines the mode of data collection. Questions presented visually are typically read by respondents, whereas those presented verbally are heard by respondents. The way in which the respondent receives the stimuli of the question has been shown to affect how a person responds to a particular survey question. Likewise, responses provided to survey questions can be written by hand, typed, or spoken. Each of these methods presents different memory and perception issues. Questions and response options that are read to respondents generally need to be shorter than those that are read by respondents, because of working memory limitations. When response categories are received visually, respondents tend to choose categories early in the list (a primacy effect). When they are received aurally, respondents tend to choose categories toward the end of the list (a recency effect). Thus, researchers must pay special attention to possible mode effects on data quality, especially in mixed-mode surveys in which some answers to a question come from respondents who were contacted via one mode (e.g. mail) and other answers to these same questions come from a different mode (e.g. telephone).

The infrastructure (and thus the financing) needed to conduct a survey also differs by mode. A self-administered, Web-based survey of several thousand individuals could potentially be carried out by an individual person, while a face-to-face survey of the same size would require a staff of interviewers and field managers. If a telephone survey is being considered, a centralized telephone interviewing facility often is required. Within any specific mode, the infrastructure [Page 483]requirements may depend on the sample size and on the needs of the researcher. A telephone survey of a few hundred individuals could be conducted by a team of students using paper-and-pencil questionnaires. However, a national face-to-face survey of several thousand will probably require a large survey research center with appropriate staffing and experience. Some recommend that the same professional and scientific standards be followed regardless of the particular structure of a research project. While a team of students can conduct the interviews, they should be trained on ethical issues of conducting research and on appropriate interviewer behavior (e.g. how to read questions) at the same level as staff in a professional research organization. This will include very specific instructions about whether questions are to be read as worded, whether and how the interviewers should probe respondents, and whether clarification or elaboration can be given by the interviewer.

...