Interactional Justice

What it is

Robert Bies and Joseph Moag named it in 1986 as a third kind of justice, beside the fairness of outcomes and the fairness of procedures. Their point was that people also judge the quality of the interpersonal treatment they receive when a procedure is carried out, whether they were treated with respect and given an honest account. Later organisational research often splits it in two, interpersonal treatment and the adequacy of explanations, but the service recovery literature usually keeps them together.

In complaint handling, Stephen Tax, Stephen Brown and Murali Chandrashekaran's 1998 study made it the third of the three tests customers apply: outcome, procedure and the nature of the interpersonal treatment during the process. Moshe Davidow's 2003 review split the response further, and three of his six dimensions sit here: apology, the acknowledgement of the complainant's distress; credibility, the firm's willingness to explain what went wrong; and attentiveness, the quality of the communication between the employee and the customer.

In effect

The weighting comes from Katja Gelbrich and Holger Roschk's 2011 meta-analysis in the Journal of Service Research. Distributive justice, the fairness of the outcome, was the main driver of satisfaction with the single transaction. Cumulative satisfaction, the main antecedent of loyalty, depended even slightly more on interactional justice than on distributive justice. Interactional justice mattered most when the failure was not about money and when it occurred in service industries.

What makes the treatment work has been taken apart. Holger Roschk and Susanne Kaiser's 2013 experiment in Marketing Letters broke the apology into three components, empathy, intensity and timing. Empathy had the strongest effect on satisfaction, intensity came next and timing last, and a late apology lowered satisfaction. The empathy effect was stronger for process failures than outcome failures, and for process failures an apology did about as much as compensation.

Davidow reported his own earlier finding that attentiveness was the single most important dimension, and that credibility, the willingness to explain, had a significant effect on satisfaction and repurchase intentions.

The treatment also travels further than the incident. Tom DeWitt, Doan Nguyen and Roger Marshall's 2008 hospitality scenario study of 459 people found that perceived justice after a recovery predicted trust and positive emotion, and that trust fully mediated the link between justice and loyalty.

What it does not say

It does not say an apology is enough. In Nira Munichor and Anat Rafaeli's 2007 telephone experiment, callers on hold who heard repeated apologies abandoned the call about as often as those who heard only music, while those told their position in the queue abandoned far less. Treatment without anything useful attached did nothing the study could detect.

It does not say treatment replaces compensation. The meta-analytic edge for treatment on loyalty is slight, and for satisfaction with the incident itself the outcome was the main driver.

And Davidow's review recorded apology as one of the least-studied dimensions at the time, with mixed results. The later experiments narrow that gap without closing it, and most of them use scenarios rather than customers' own failures.

How the article reads it

The article on what a customer apology must contain sorts a recovery into three parts and puts the apology, the explanation and the attention a customer gets in this one. Its reading is that treatment is the part a firm can have ready before any failure happens, since it depends on trained staff who are allowed to act, and that no per-incident refund buys that after the fact. The practical order the research suggests is to apologise at once with empathy for what the failure cost the customer, then to explain what happened plainly, and only then to repair and repay. Most of the studies behind this were read at the level of their published abstracts, and most use scenarios, so the order is a reading of the evidence and not a tested sequence.

---