PWOM vs NWOM: More on the Debate
The debate regarding PWOM vs NWOM (see comments from Gary Stein, Pete Blackshaw, and Walter Carl ) is one that has emplored me for years, and others for decades. What follows here is some of what others have contributed to the debate, a few of my thoughts, and some of what I too wonder about.
Before we enter the debate, as a WOM researcher or manager, it is important to first determine precisely what it is that we are comparing -- the "unit of analysis," if you will. For example, is it impact of PWOM vs NWOM or is it prevalence of PWOM vs NWOM that you care about? Both are important. Regarding impact, it makes sense that consumers may weigh and recall negative information more heavily than positive, although this is not always the case. Consider yourself buying a new car... you would probably pay more attention to warnings from friends regarding which cars to avoid, as opposed to recommendations regarding which cars to buy. On the other hand, it is conceivable that a strong positive recommendation from a good friend to buy a house in a certain neighborhood may have more bearing than a warning from just about anyone else... Regarding prevalence, the proportion of positive (satisfying) versus negative (dissatisfying) consumption experiences is certainly relevant in determining PWOM vs NWOM propensities, as highlighted by the Verde Group. Of course, this is also likely to vary by product category. For example, consumers may have fewer dissatisfying consumption experiences with their shampoos than they do with the DVDs that they rent.
A few comments and studies regarding the PREVALENCE debate:
In my conversations with others regarding WOM, one "fact" invariably comes up: "Research has shown" that when consumers are miffed, they blab to 10 people, whereas when they are delighted, they might tell one other person. Where did this phantom statistic come from? I did some digging and found that independent research conducted by TARP for Coca-Cola in 1981 concluded that "on average, twice as many people are told about a bad experience than are told about a good experience." Since then, popular press and trade articles have generalized and exaggerated this claim about the propensity for dissatisfied consumers to talk, and vaguely credit the statistics to "research has shown."
Here is how academics weigh in ... Research on gossip and rumors has noted a negativity bias, where negative rumors (e.g. fear and aggression-based) far outweighed positive rumors ("pipedreams") (Rosnow 1988). Note that this could conceivably be because people believe negative rumors are more important (impactful) and therefore they spread them more (prevalence). In a study that considers only dissatisfied consumers, it was concluded that as the severity of the problem increases, so too does NWOM (Richins 1983 JM). This seems sensible. In another study that I find particularly compelling, the researcher demonstrated "an asymmetrical u-shaped relationship between satisfaction and WOM" (Anderson 1998), where extremely satisfied or extremely dissatisfied consumers engage in more WOM than moderately satisfied consumers, and this relationship is exaggerated for dissatisfied consumers. This one makes sense to me: why would consumers bother talking about something that is neutral? Give consumers some credit - they are only going to talk about your product if it is newsworthy.
However, several other studies contradict the above findings and demonstrate the prevalence of PWOM across many other product categories, including coffee (Holmes & Lett 1977), car dealerships (Swan & Oliver 1988), television shows (Godes & Mayzlin 2004) and books (Chevalier & Mayzlin 2005). The Keller Faye Group concludes that PWOM outweighs NWOM by a ratio of 6:1!
So where does this leave us? I wonder why and how it is that such inconsistencies exist. A few ideas:
1. Inter-category differences may explain some of the inconsistencies. For example, it has been suggested that higher involvement product categories may have higher ratios of NWOM to PWOM (Richins 1983). That said, the TARP Coca-Cola research yielded similar results for both high and low involvement products and other research has demonstrated that in the high involvement automobile category, WOM increases with satisfaction (Swan and Oliver 1989).
2. Differences between consumers may also account for the inconsistent results. For example, age and expertise may effect people's propensities to generate PWOM vs NWOM. Even the medium across which they talk (online, face-to-face, over the phone, etc) may impact people's WOM behaviors.
3. Different levels of specificity and definitions regarding WOM itself may account for these contradictions. For example, the constructs "PWOM" and "NWOM" are quite different from "WOM following a satisfying experience" and "WOM following a dissatisfying experience." The former refers to any positive or negative consumer-to-consumer conversations regarding a product or service. The latter is more specific, since it is limited to conversations regarding one's past consumption experiences.
4. The proportion of satisfying versus dissatisfying consumption experiences (as stated above) will have a significant impact on the proportion of PWOM vs NWOM. The proportion of satisfying vs dissatisfying experiences may also vary by category, by type of consumer... what else?
It seems I have asked more questions than I have answered here. I suppose that is the point? Let's keep this conversation going...
