Reliability and Validity: Misnomers for Qualitative Research

Abstract

Validity and reliability, as indices of measurement, have been well developed in and for quantitative methods. Validity has been classified and defined in the following ways: face, content, predictive, concurrent, construct, internal and external. Reliability has been classified and defined as quixotic, diachronic, synchronie, external and internal (Duffy, 1985; Kirk & Miller, 1986; Knapp, 1985; Le Compte & Goetz, 1982). Essentially the terms validity and reliability refer to accuracy, consistency and equivalence in research that is designed for quantification in the natural sciences. Many tests have been and continue to be developed to prove validity and reliability. Not all quantitative researchers agree however, that the terms are used accurately. For example, Knapp (1985) in his article "Validity, reliability and neither" notes errors that are commonly made: describing internal consistency as a distinct and different property, when in fact it is an aspect of reliability; or, the citing of a correlation between parallel forms as evidence for validity. Knapp illustrates how some researchers incorrectly use these measures and he raises a second issue - that knowledge about reliability and validity is continuing to evolve. The thesis of this paper is that the terms validity and reliability should not be applied to qualitative research methods. When Le Compte and Goetz (1982) wrote their, seminal article on how qualitative research has adhered to the canons of reliability and validity, what was their intent? They acknowledged that reliability is difficult to measure in qualitative research because of the nature of the narrative data and the involvement of the researcher in a change process; yet, they proceeded to force qualitative methods to fit criteria for external and internal reliability. Other qualitative researchers such as Goodwin and Goodwin (1984), Duffy (1985), Swanson-Laufman (1986), and Atwood and Hinds (1986) have followed suit, describing how their research addresses these measures.