Measuring Reliability: A Comparison of Alternative Techniques

Abstract

Few scholars would dispute the need to quantify and report the reliability of measurement in marketing research. There appears to be considerable confusion, however, about the appropriateness of various reliability measurement techniques. A comparison is made of the results from several alternative techniques applied to a lifestyle questionnaire. One technique rarely used to date in marketing research is suggested as superior to the others on the basis of the likelihood of its assumptions in an actual measurement setting.

Validation Guidelines for IS Positivist Research

“…Non-randomized ordering of questionnaire items could easily explain the high Cronbach's αs that are reported as well as other methods artifacts related to discriminant and convergent validity [Straub et al, 1995]. Straub [1989] did not discuss some traditional tests for reliability, including split-half, test-retest, and alternative forms [Cronbach, 1971, Parameswaran et al, 1979. Although Parameswaran et al [1979] criticize these traditional reliability tests for the assumptions in their theoretical underpinnings, IS researchers need to understand the basic approach in these techniques to be able to intelligently review manuscripts that choose to use these tests.…”

“…Straub [1989] did not discuss some traditional tests for reliability, including split-half, test-retest, and alternative forms [Cronbach, 1971, Parameswaran et al, 1979. Although Parameswaran et al [1979] criticize these traditional reliability tests for the assumptions in their theoretical underpinnings, IS researchers need to understand the basic approach in these techniques to be able to intelligently review manuscripts that choose to use these tests. In some specific cases these techniques continue to be useful in IS research, and IS researchers need to understand why these cases justify the use of these techniques.…”

Validade e fidelidade de instrumentos de medida em psicologia: Tudo o que sempre quis saber e nunca teve coragem para perguntar

“…Quanto ao método das "formas paralelas" (Parameswaran et al, 1979), utiliza-se, normalmente, um instrumento que possua formas "quase idênticas" (e.g., o teste de inteligência de Raven -SPM -possui diversas "formas paralelas" do "teste mãe"; avaliam exatamente o mesmo construto, mas possuem uma "aparência diferente" entre si) e em que cada uma delas é aplicada em momentos separados no tempo. Este método é útil para obviar possíveis ameaças que se colocam ao método teste-reteste, tais com a aprendizagem, familiarização com o instrumento de medida, entre outros.…”

Η Συμβολή Της Λογιστικής Πληροφόρησης Σε Πελατοκεντρικές Στρατηγικές Και Συστήματα CRM

“…Για την εκτίμηση της αξιοπιστίας μιας κλίμακας μέτρησης υπάρχουν διαφορετικές τεχνικές (Parameswaran et al 1979). Επίσης, υπάρχουν εναλλακτικοί τρόποι για τη βελτίωση της αξιοπιστίας μιας κλίμακας μέτρησης (Churchill & Peter 1984).…”