2013IS实证研究方法的讨论 DISCOVERING UNOBSERVED HETEROGENEITY
Assuming that data in empirical studies are homogeneous andrepresent a single population is often unrealistic in the socialand behavioral sciences, such as information systems, management,and marketing (Rust and Verhoef 2005; Wedel andKamakura 2000). There may be significant heterogeneity inthe data across unobserved groups, and it can bias parameterestimates, lead to Type I and Type II errors, and result ininvalid conclusions (Jedidi et al. 1997). Consider the followingtechnology acceptance model (TAM) example: Aresearcher is interested in individuals’ intention to use an ITsystem or service (Davis et al. 1989; Venkatesh 2000;Venkatesh and Davis 2000; Venkatesh et al. 2003). Informedby existing theory, the researcher proposes a model in whichperceived usefulness (PU) and perceived ease of use (PEOU)of the IT system explain intention to use the system (IU)(Figure 1). The empirical results reveal that PU and PEOUare equally important in explaining IU. However, the theoryand model overlook the two underlying groups: experiencedIT users (Figure 1a, segment 1) and inexperienced IT users(Figure 1a, segment 2). Experienced users show a strongpositive relationship between PU and IU and a weak, or nonsignificant,relationship between PEOU and IU. In contrast,inexperienced users show a strong positive relationshipbetween PEOU and IU and a weak, or nonsignificant, relationshipbetween PU and IU (Figure 1a). In this scenario,drawing inferences based on results from the overall samplewould lead to Type I errors as we would be overgeneralizingthe significant findings from the overall sample to theunderlying user groups, one with a nonsignificant estimate forPEOUIU and the other with a nonsignificant estimate forPUIU. If the model is not refined to accommodate thisunobserved heterogeneity, a system that is unsuitable foreither user group (i.e., one with average usefulness andaverage ease of use) may be provided to all users.