Field Coding

Field coding involves the use by an in-person or telephone interviewer of a standardized listing of response options to categorize open-ended responses given by respondents to questions that provide no specific response options to the respondent. This approach differs from the administration of a closed-ended question, where the response options are read to the respondent, and differs from the administration of open-ended questions, where the response is typically recorded verbatim.

With field coding, an interviewer typically asks the respondent an open-ended question and waits for a response. As the respondent replies, the interviewer records the information into one or more of the predetermined response options. Should the respondent give an answer that is not on the interviewer's list of response options, the interviewer either must interpret the answer as close as possible to one of the [Page 277]predetermined response options or ask follow-up probes to clarify the response. Creating as inclusive a set of response options as possible is important, which means that the researcher must anticipate (or know from previous research studies) how the sample population might respond to the particular survey questions administered in this manner. As a last resort, when an interviewer cannot map an open-ended response to a pre-specified response option, the researcher should provide the interviewer with an "other" response choice and ask that some verbatim specification of what the respondent said be written by the interviewer.

By allowing respondents to reply in their own words, field coding techniques help to establish a dialogue between the respondent and the interviewer that more closely resembles a conversation than is typically the case with the administration of closed-ended survey questions. A positive rapport can help facilitate more sincere and detailed answers from the respondent. Additionally, the use of a predetermined set of response categories allows for greater standardization of the process than might be the case with recoding of verbatim responses, which can often be incomplete or unrelated to the actual question asked. A researcher must anticipate possible responses by the respondent, which requires development of an inclusive but mutually exclusive set of response options. Pretesting of the initial response options helps create the set of possible responses to use in the coding list.

Use of a standardized set of response options may, however, limit the capture of more complex responses, those which do not fit into the predetermined categories. In this respect, field coding may produce data that are less comprehensive than the recording of full verbatim responses or the taking of field notes by the respondent. Because the interviewer takes an active role in "creating" the respondent's response, field coding is susceptible to reactivity (i.e. changes in the respondent's answers caused by interaction with the interviewer or the setting) and to coder variance (variation in how an identical response is coded across a number of different interviewers). Further, the respondent may take a longer time to reach the same answer than he or she would have had the response options been presented as part of the question. This can increase the cost of conducting a survey.

...