The Sound of Silence

When choosing between alternatives of uncertain quality, one can learn from other decision-makers either through communicating with them about the reasons underlying their choices, or through simply observing their choices. The former way of learning has been extensively studied in the word-of-mouth (WOM) literature, while the latter way, labelled silent word-of-mouth (SWOM), has received less attention. This paper is the ̄rst individual-level empirical study to structurally model how SWOM in°uences choices. I consider a setting where a group of individuals sequentially decide whether to adopt a product of unknown quality. Each individual is equipped with a private quality signal. Under WOM, one incorporates her predecessors' signals, as well as her own signal, in her quality evaluation. Under SWOM, she observes her predecessors' decisions, infers what their signals have been, and updates her quality evaluation. I show that WOM and SWOM can lead to distinct choice patterns in the marketplace in that decisions are more likely to be imitated under SWOM. The model is applied to a high-stake decision environment where patients on a transplant waiting list must choose whether to accept a kidney for transplantation. This data is desirable to study the SWOM e®ect because (1) decisions are sequential, (2) all previous decisions are observed, and (3) privacy concerns make WOM impractical, which helps to isolate the SWOM e®ect. An interesting puzzle in the data is that, other things equal, patients towards the back of the queue are less likely to accept a kidney. SWOM explains this phenomenon: A patient lowers her quality evaluation about a kidney once it is previously refused and becomes more inclined to reject as well. Her refusal in turn lowers the quality evaluation for subsequent patients, triggering more refusals down the queue. In this way, refusals are self-reproductive. Indeed, policy experiments show that kidney acceptance rate would have been substantially higher if there were WOM communication, or if there were no social learning at all. The competing explanation of unobservable (to the researcher) public quality information is ruled out. This paper contributes to the understanding of how social imitation originates and propagates with observations of others' choices. It highlights for marketing managers the importance of (1) ̄rst-impression management by building a strong initial customer base and by enhancing the visibility of early adopters, and (2) tipping-point management by targeting the marginal customer segment.