Dual-faceted multidimensional IRT models with hierarchical structure

In marketing, there are many important multidimensional constructs, such as service quality, market orientation, and consumer-based brand equity (CBBE). Survey methods and multidimensional scales are used to measure these constructs. But the survey responses collected for these constructs are also dual-faceted. Except for consumers, the other facet (such as services, firms or brands) is involved too. Assessment of multidimensional marketing scales (such as SERVQUAL and MARKOR) has relied on confirmatory factor analysis (CFA) and structural equation modeling (SEM), which may not give accurate and unbiased evidence about the estimates and dimensionality across both facets. The authors introduce a hierarchical extension of many facet item response theory (MFIRT) to empirically investigate the causal relationships between a construct and its dimensions for scales that generate dual-faceted data. CBBE data for soft drink brands are used in an empirical study, which demonstrates the method and finds CBBE best modeled as formative from its dimensions along the consumer facet.

• Many multidimensional constructs in marketing are dual-faceted. • Only single facet assessment of multidimensional marketing scales is available. • The causal relationships between a construct and its dimensions are examined across facets.