Essay on Discuss the various ways of determining the validity of a particular measure. Of what importance is triangulation in resolving the issue of validity?
Speaking about the validity of measurement, we are interested in how precisely the values obtained by this measurement correspond to the true values of the measured variable. High validity means that the individual characteristics of the respondents in this psychological and socially homogeneous sample identified by the methodology are really essential for the diagnosis and prognosis of their behavior and performance in the assumed situations and conditions (Kirk and Miller 23-25). Next, we will consider the main ways of identifying the validity of measurements.
In establishing the validity the crucial role belongs to the grounding and the subsequent verification of the hypothesis of relevance, i.e. semantic accordance of estimations obtained by the methodology on the one hand, and formalized assessments of behavior and activities of the respondents on the other. It is assumed that both assessments are conditioned by the same individual characteristics. Theoretically grounded relevance is subject to practical test. The higher the level of matches between both estimates is, the higher the validity is. Complete coincidence in practice is highly improbable, because the level of matches is influenced by a number of factors which are impossible to calculate and formalize: the dynamics of motivation, random fluctuations of conditions under which the estimates are obtained, etc. (McCrae, Kurtz, Yamagata and Terracciano 28-50). There are three types of establishing validity.
- Validity by criterion.
The assessments of the respondents obtained by the applied method are compared with formalized estimation of their behavior and performance that serve as a research subject. The latter estimates are called criterial or criterion. For determining the level of accordance most frequently the calculation of the coefficient of linear correlation is made, there are also chi-square method and other statistical techniques that reveal the strength of connection (Kirk and Miller 48-55). It should be noted that the relationship between the estimations obtained by the method and the criterial assessments should not necessarily be rectilinear (for example, between the level of professional qualification and performance level).
- Validity by content.
It is determined in cases when it is necessary to prove that the questions and tasks included in the methodology represent the problems faced by the individual in the areas of behavior and activity examined by the study. This validity is verified first by a substantive analysis of all questions and tasks the methodology consists of to give a semantic characterization of their representativeness. Then the method is tested on socially and psychologically homogeneous sample. All individuals included in the sample are interviewed by the commission of experts, and it is agreed beforehand that the experts’ questions by volume or content will not fully correspond to the methodology, but they must relate to the represented problems (Kirk and Miller 57-63). Further, the evaluations obtained by the method are compared with the formalized assessment of the commission of experts: the higher the level of matches is, the higher the validity is. In this case the same statistical methods are applied as for the establishment of the first type of validity.
- Validity by construct, correlation ratio.
The results of the method being verified are compared with the results of other methods designed to deal with research problems of the same class which passed a preliminary check of validity. For this purpose statistical comparisons are applied: multiple correlations and regressions, factor analysis (Kirk and Miller 65-68).
Validity of all three types is characterized by the statistical value obtained through comparisons, the coefficient of correlation, regression, factor weight. All estimations, including those obtained by the method, criterial ones, expert assessments on other methods must have sufficient statistical reliability otherwise validity cannot be established.
Postmodern authors discuss the radically new concepts and principles of justification of knowledge: the ironic (J. Baudrillard), neopragmatic (J.-F. Lyotard), risomatic (J. Derrida) validation (Shadish, Cook, and Campbell 358-372). N. Denzin substantiated the triangulation method as a radical alternative to traditional approaches to validation. Triangulation is the combining and interactive use of different techniques, methodologies, data, theories and/or researches. Multiplicity of approaches and methods is designed to reduce the epistemological limits, cross the established boundaries, and reveal new aspects of the phenomenon (Shadish, Cook, and Campbell 411-418). It should be noted that the idea of triangulation [of the truth] was generated in postpositivism (D. Campbell), where it was understood as a procedure, a set of useful tricks, e.g.: in addition to a survey of “ordinary” people to interview experts; to randomly split the sample of subjects in two parts, and analyze the data separately; to exclude one variable from the analysis and see how the model changes; to validate the construct using the approach “many traits, many methods”, etc.
In general, social research possesses value only when it gives equal opportunity to speak to representatives of different groups, creates equiprobable versions and suppresses the dominant, “correct” interpretations.
