Statistical correctness

This paper calls for improved survey research methods both in formal research projects and in everyday use of survey questionnaires in libraries. The author proposes that probability sampling be recognized as the best practice in library and information science. The problem of selection bias arising from the use of nonprobability sampling, and convenience sampling in particular, is explored. The paper presents examples of unfounded research conclusions appearing in studies using this sampling approach. It also espouses library research and assessment practices that aspire to higher standards of soundness, accuracy, validity, and impartiality.