Word of Mouth (hereafter, ‘WOM’) communication is one of the most pervasive and influential mediums consumers use to learn about products and services in markets (Alexandrov & Sherrell, 2006; Balter, 2008; Cotte, Coulter, & Moore,
2005; Grewal, Gotlieb, & Marmorstein, 1994; Whyte, 1954). Despite this acknowledgement little is known about the nature of the information that is typically communicated by WOM (Frenzen & Nakamoto, 1993). This research aims to provide this insight by examining a sender’s decision to communicate by WOM.
Drawing on Random Utility Theory, a general model of WOM communication is proposed (Brown & Reingen, 1987; Louviere, Hensher, & Swait, 2000; Roloff,
1981; Thurstone, 1927). This model emphasises the sender of information as the controller of the flow of information in WOM networks. It accommodates the influence of context and the resulting motivations on the sender’s choice of communication behaviour. At the heart of this model is the idea that a sender will choose to pursue rewards from a WOM exchange, thereby motivating them to participate. They will then choose behaviour that maximises the probability of obtaining these rewards. This allows the linking of the literature regarding motivation and behaviour in WOM communication.
This model was proposed within a choice based experimental framework. The advantage of using such a framework is that it allows precise measurement of individual level behaviours. The outputs of the models from this are also the probabilities of specific communication behaviours. These outputs are able to be used as inputs in systems and network based models of aggregate level WOM phenomenon. This offers one of the first methodological approaches to link the individual and aggregate level aspects of the WOM literature.
To test the robustness of this model two particular classes of WOM are considered: assisting and covering. Assisting refers to the typical form of WOM in which a sender will offer genuine assistance to the receiver to help them make the ‘best’ decision given their preferences. Covering refers to a less typical form of WOM in which a sender wishes to appear to be offering help but in fact is attempting to be as obstructive as possible. These two classes of WOM offer a strong test of this new model.
The results of this research indicate that there is strong support for the underlying model used to link WOM context, sender motivation and sender behaviour. The estimation approach using Individual level scale adjusted Feasible Generalised Least Squares (FGLS) regression offered choice predictions that correlated 0.79 v/ith the observed choice probabilities. The specific results regarding assisting and covering type WOM also provide new insight into these WOM forms.
It is found that a sender generally provides helpful information when assisting. This is achieved by providing information about relatively important product features, by confirming a receiver’s existing knowledge to make them more certain in their beliefs, by expressing the facts in the information with certainty, and by using language that is unambiguous, i.e. terminology consistent with previous communications. Each of these types of information would allow the receiver to improve the quality of their decisions. One exception can be noted; the sender also chooses to communicate information of lower economic value even when assisting. More recent research into interpersonal communication suggests that this may be a manifestation of risk averse behaviour, whereby a sender does not wish to be responsible if a receiver suffers losses as a result of a decision (Young, Donald, Freeman, & Benn, 2008).
With regard to covering type WOM it is found that senders exhibit largely the same behaviour as when assisting with two critical exceptions. These are choosing to communicate information in an uncertain manner, and using marginally more ambiguous language, that is, language using terminology less consistent with prior communications. The effect of both of these would be a dramatic reduction in the usefulness of the information for the receiver; however, the remaining similarity to assisting type WOM prevents the receiver from detecting this covert action.
Results from this research offer new insights into how consumers use WOM communication each day. Furthermore, a number of additional gender specific insights are obtained offering even greater detail regarding how individuals choose to communicate by WOM.