Nonsampling Error

There are many forms of potential error in social research that a prudent researcher will (a) try to minimize and/or (b) measure (cf. Groves, 1989). Error is due to the various causes of BIAS (i.e., constant error) and VARIANCE (i.e., variable error) that can occur whenever data are gathered, regardless of whether the research is QUANTITATIVE RESEARCH or QUALITATIVE RESEARCH. SAMPLING ERROR is the variance associated with any research sample that is not a census of the population of interest. Nonsampling error is the term that traditionally has been used to refer to all other sources of bias and variance, and it includes coverage error, nonresponse error, and measurement error.

Coverage error refers to the bias that may result when the frame (the list) used to represent the population of interest fails to adequately cover (include) the entire population. Nonresponse error refers to the bias and variance that may result when data are not gathered from all of the elements sampled from the frame. Measurement error refers to potential bias and variance that can be caused by any of the following: (a) the person who gathers the data (e.g., an interviewer, observer, or coder); (b) the instrument that is used to gather the data (e.g., a questionnaire, observational or coding form); (c) the subject/respondent being measured; (d) the mode of data collection (e.g., in-person vs. self-administered). The literature abounds with methodological and statistical studies that address various techniques that can be used to try to reduce the possibility that nonsampling errors will occur and/or ways to measure and adjust for such errors when they do result. Of note, many think that these forms of errors apply only to sample surveys, when, in fact, each type of error has its counterpart in any form of social research, including OBSERVATIONAL RESEARCH, CONTENT ANALYSIS, and all forms of qualitative research.