Author List: Bapna, Ravi; Goes, Paulo B.; Wei, Kwok-Kee; Zhang, Zhongju (John);
Information Systems Research, 2011, Volume 22, Issue 1, Page 118-133.
Despite much hype about electronic payments systems (EPSs), a 2004 survey establishes that close to 80% of between-business payments are still made using paper-based formats. We present a finite mixture logit model to predict likelihood of EPS adoption in business-to-business (B2B) settings. Our model simultaneously classifies firms into homogeneous segments based on firm-specific characteristics and estimates the model's coefficients relating predictor variables to EPS adoption decisions for each respective segment. While such models are increasingly making their presence felt in the marketing literature, we demonstrate their applicability to traditional information systems (IS) problems such as technology adoption. Using the finite mixture approach, we predict the likelihood of EPS adoption using a unique data set from a Fortune 100 company. We compare the finite mixture model with a variety of traditional approaches. We find that the finite mixture model fits the data better, controlling for the number of parameters estimated; that our explicit model-based segmentation leads to a better delineation of segments; and that it significantly improves the predictive accuracy in holdout samples. Practically, the proposed methodology can help business managers develop actionable segment-specific strategies for increasing EPS adoption by their business partners. We discuss how the methodology is potentially applicable to a wide variety of IS research.
Keywords: clustering analysis; electronic payments systems; finite mixture model; hierarchical logit regression; logistic regression; market segmentation
Algorithm:

List of Topics

#48 0.334 dimensions electronic multidimensional game transactions relative contrast channels theory sustained model predict dimension mixture evolutionary results unique traditional likely finite
#215 0.133 data classification statistical regression mining models neural methods using analysis techniques performance predictive networks accuracy method variables prediction problem measure
#191 0.104 model models process analysis paper management support used environment decision provides based develop use using help literature mathematical presented formulation
#49 0.088 adoption diffusion technology adopters innovation adopt process information potential innovations influence new characteristics early adopting set compatibility time initial current
#182 0.078 percent sales average economic growth increasing total using number million percentage evidence analyze approximately does business flow annual book daily
#44 0.077 approach analysis application approaches new used paper methodology simulation traditional techniques systems process based using proposed method present provides various
#67 0.058 production manufacturing marketing information performance systems level impact plant model monitor does strategies 500 unit present fortune integrated sales plants