Vormistamine ülesanne 3
Among those ar
the relatively well understood effects of sampling variability, but also more difficult to measure
effects. Deming incorporates effects of the interviewer, method of data collection, nonresponse,
questionnaire imperfections, processing errors and errors of interpretation. Other authors (e.g., Kish,
1965, see also Groves, 1989) basically classify threats to survey quality in two main categories, for
instance differentiating between errors of nonobservation (e.g., nonresponse) and observation (e.g.,
in data collection and processing). Biemer and Lyberg (2003) group errors in sampling error and
nonsampling error. Sampling error is due to selecting a sample instead of studying the whole
population. Nonsampling errors are due to mistakes and/or system deficiencies, and include all
errors that can be made during data collection and data processing, such as coverage, nonresponse,
measurement, and coding error (see also Lyberg & Biemer, Chapter 22).