The Moral Licensing Effect
Eight morally mixed scenarios across four life domains. Four ask cold. Four ask after a prior moral act has been logged on your day. Hidden in pairs. You rate likelihood on a 0-100 slider. WIZ measures how much the prior good deed quietly bought you.
“After a virtuous act we relax; after a transgression we compensate. The moral self oscillates around a set-point.”Sachdeva Iliev & Medin (2009) self-regulation model
Monin & Miller (2001) JPSP vol 81 founded the paradigm. Subjects asked to choose a candidate for a job in a male-dominated industry (police chief, construction manager) were more likely to choose the male candidate over an equally-qualified female candidate IF they had first been given an opportunity to disagree with a blatantly sexist statement. The licensed group chose male 71% of the time; the unlicensed control chose male 49% of the time. Same hiring scenario, same candidates, same instructions. The only thing that changed: whether the subject had a prior moral act on the books.
Mazar & Zhong (2010) Psychological Science vol 21 ran the cleanest demonstration in consumer behavior. Subjects who shopped in a virtual green store cheated more on a subsequent dice-rolling task and stole more money from the experimenter than subjects who shopped in a virtual conventional store. The mere act of choosing environmentally responsible products produced a measurable drop in honesty. Khan & Dhar (2006) JMR vol 43 found that subjects who imagined volunteering for three hours next week were more likely to choose a luxury item over a utilitarian item in a subsequent unrelated choice — the licensing extends to mere intention. Effron Cameron & Monin (2009) JESP vol 45: endorsing Obama licensed subsequent stereotypical preferences. Sachdeva Iliev & Medin (2009) Psychological Science vol 20 generalized the mechanism into the moral-self-regulation model: people maintain a moral set-point and oscillate around it.
Meta-analytic status. Blanken van de Ven & Zeelenberg (2015) PSPB vol 41 pooled 91 studies and reported d=0.31 (small to moderate). Simbrunner & Schlegelmilch (2017) Management Review Quarterly vol 67 pooled 89 studies and reported d=0.32, with the largest effects in Western individualist samples. The effect replicates; the magnitude has moved downward from the early-2000s founding studies, in line with the broader social-psychology replication picture.
The exercise. Eight scenarios in four hidden pairs across hiring, environment, indulgence, and honesty. Each pair: same behavior, same stakes, same person. The only difference is whether a prior moral act has been recorded on the day. For each scenario you move a 0-100 slider for how likely you would be to do the morally mixed thing. At the end I compute your Moral Licensing Gap: the average likelihood you assigned to the four licensed versions minus the average for the four unlicensed versions. The Blanken 2015 modal gap is roughly 10-20 points on a 100-scale equivalent. A subject with no licensing sits at zero. A subject for whom one good deed clearly buys the next mixed one sits above 30.
The exercise is not a moral test. The bias is real and operates on almost everyone. Knowing it is the first step in the Conway & Peetz (2012) intervention: when you notice yourself reaching for a prior good act as cover for a current mixed one, ask whether the same behavior would look the same if the prior act had not happened. If yes, the act and the prior credit are independent. If no, the prior credit is doing work it cannot legitimately do.