Information variability of online word-of-mouth as a context for customer decision making

This dissertation explores how the variability of online word-of-mouth (O-WOM) valence affects customer decision making. Valence is the positiveness/negativeness of a posted opinion and variability is the distribution of customer opinions. When customers share opinions on platforms without possibility for feedback from other customers, like in rating-type websites, the valence distributions tend to be U shaped (i.e. only extreme opinions are present). When customers share opinions on platforms with possibility for feedback from other customers, like in forum-type websites, the valence distributions tend to have bell-shaped form (i.e. customers hold and share moderate point of view).;To address the issue for the effect of O-WOM variability it was hypothesized that customers do not hold a single point estimate evaluation for a brand but expect the whole range of possible outcomes with different probabilities resulting in an expectation distribution. Such a distribution can be described by three parameters: mean, variance, and entropy. Hence, the two major research questions were; first, can O-WOM variability be mapped onto the customers' expectation distributions; and second, how the parameters of the expectation distribution affect the customers' decision making process?;The first question was addressed through experimental design. Four groups with total of 395 students were provided with treatments of assembled opinions for an imaginary computer notebook brand. The four treatments were organized in a two by two matrix with high/low levels of variance and entropy. This simulation of distribution was meant to correspond to different platforms for O-WOM collection. The results indicated that neither of the parameters related to variability (variance and entropy) were mapped on the expectation distributions.;The second research question resulted in the test of a customer decision making model where the parameters of the expectation distribution (i.e. mean variance, and entropy) affect value equity, brand equity, and confidence, which in turn affect purchase intentions. Product expertise was used as a control variable. In general, the model was supported. The results indicated that variability affected mostly brand equity with a positive sign. Value equity was almost not affected by variability but on the other hand the mean exerted greater influence on it. Expertise affected both mean and entropy negatively.;Additional post-hoc analysis for mean comparison revealed that different groups of customers tend to form different shapes of expectation distributions. Expertise, gender, and confidence all affected the parameters of the expectation distributions, while involvement did not. The meaning and the significance of the results are elaborated on in the discussion section and guidance for future research is provided.