bei Google ScholarTitelangaben
Schüler, Jens ; Küsshauer, Alexander ; Anderson, Brian S. ; Baum, Matthias:
Test-retest reliability in metric conjoint experiments: Important requirement or overrated nuisance?
In: Academy of Management Proceedings.
(2022)
Heft 1
.
ISSN 2151-6561
DOI: https://doi.org/10.5465/AMBPP.2022.260
Abstract
Metric conjoint studies are a popular research design in the entrepreneurship domain to tap into the decision-making processes of individuals. For these studies, test-retest reliabilities of r = 0.70 or higher, are considered as the central and most important quality criteria that can make or break any study. Although the importance of this criterion is routinely highlighted in best-practice recommendations, we lack actual empirical evidence on its usefulness in the conjoint context i.e. the sensitivity of conjoint results upon varying test-retest reliabilities remains unclear. Therefore, drawing on a comprehensive review of metric conjoint studies published in the entrepreneurship literature, we conduct a Monte Carlo simulation to investigate how various test-retest reliabilities and modeling considerations affect regression outcomes in conjoint studies. Our simulations indicate a more cautious perspective towards test-retest reliabilities for metric conjoint designs and best practice recommendations are discussed.