Authority (the authority principle)

Evidence

The evidence usually cited for the authority principle comes from obedience research. Milgram (1963) led 40 men aged between 20 and 50 to believe they were administering electric shocks of increasing strength to another person. An experimenter told them to continue. 26 of the 40 went to the highest level, 450 volts, and none stopped before 300 volts. Fourteen final-year psychology students had predicted beforehand that at most three in a hundred would go that far. Bickman (1974) tested the everyday form on the street — and his finding is narrower than the title of his paper suggests. In Brooklyn, 153 passers-by were told by a young man to perform some trivial act: pick up a paper bag, give a stranger at a parking meter a dime, move to the other side of a bus stop. He wore either a sports jacket and tie, a milkman’s uniform or a guard’s uniform.

It was not the uniform that worked but this one uniform. Between milkman and civilian the difference was not significant; only the guard stood clearly above both (against civilian χ² = 7.73, p < .01; against milkman χ² = 4.04, p < .05). The χ² tests whether two frequencies differ more than chance would lead you to expect; the p-value beside it says how likely such a difference would be from chance alone. From p < .05 it counts as meaningful. Counting only the cases where people obeyed without a follow-up justification, the milkman at 14 percent falls even below the civilian at 19 — the guard reaches 38. Bickman draws the conclusion himself: the power lay in the type of uniform and not in the mere presence of a uniform.

The guard’s uniform was quasi-police. Bickman records that it superficially resembled a policeman’s; the badge and insignia were different and nobody carried a gun. What works here is a signal of official enforcement authority — not an occupational badge, and certainly not a mark of quality.

Where that power comes from is unresolved. In a second experiment 48 passers-by received the dime request either under surveillance or in a way that left the person asking long gone round the corner before it could be acted on. Surveillance made no difference whatever (χ² = 0). Reward and punishment are thereby ruled out; legitimacy Bickman arrives at by elimination, and he calls that explicitly a logical analysis, not a measurement.

His own two questionnaire studies contradict that reading. 141 and 189 students were asked to rate the requests, and to predict what they and others would do. In none of the three situations was the guard seen as more legitimate, and on predicted compliance the uniform had no effect at all (χ² = 0.05, n.s.). People do not correctly predict their own response to signals of authority — the same finding as in Milgram, where fourteen students were out by orders of magnitude.

Burger (2009) repeated the procedure 45 years later. Three things about the two papers belong together.

The result: 70 percent of participants continued past the critical 150-volt mark, against 82.5 percent in Milgram. The difference is not significant.

The stopping point: Burger ended every session at 150 volts. That participants would have gone on to 450 volts is an inference from Milgram’s data, not a measurement. Burger writes himself that only a full replication could settle it.

The sample: about a third of those interested were screened out by telephone, a clinical psychologist excluded 38 percent of the remainder, and anyone who had taken more than two psychology courses was excluded. That is ethically right and makes the sample not comparable with Milgram’s.

One side finding does not match the expectation: in a second condition Burger had a confederate refuse to obey. He had predicted this would lower the rate. It stayed practically the same (63.3 percent).

Doliński and colleagues (2017) ran the paradigm in Poland in 2015, the first team to do so in Central Europe. Eighty participants, half women and half men, aged 18 to 69. 72 of the 80 pressed the tenth button, that is 90 percent, against 85 percent in Milgram’s comparable condition. The authors speak of a ceiling effect themselves: with so few refusals no moderating factor can be demonstrated any more — not even their own question of whether it makes a difference when the victim is a woman. Three times as many withdrawals, but p = 0.56.

What matters, though, is what this supports. What was measured is compliance with an instruction given by a present person in an institutional role, usually against the inner resistance of the person complying. That is a different situation from a trust seal next to a pay button or an expert quote in an advert.

Limits

The gap between finding and application is wide. For the usual marketing transfer — seals, logos, titles, expert voices in advertising — there is no peer-reviewed paper showing an effect on purchasing behaviour. The one study that actually recorded the purchase finds none. That does not mean none exists anywhere; it means the literature usually cited measures something else. Justifying a trust seal with “Milgram” overstretches the finding.

There is research on trust seals, and it finds little. Kim, Ferrin and Rao (2008) surveyed 468 online shoppers in two rounds, before the purchase and after delivery. The attitude towards trust seals correlated with trust in the vendor at 0.05 and with purchase intention at 0.02. What explained trust there lay elsewhere: perceived privacy protection, information quality, reputation. On top of that, 73.7 percent of respondents did not know that websites are certified by third parties at all. The work is a survey, not an experiment — the guide on trust badges works this through with a second paper.

Bickman’s cells are small. 14 to 24 passers-by per combination of uniform and situation, twelve in the second experiment. The percentages above hang on a handful of individual cases.

Most people obeyed only after a justification. Only 25 percent complied with the bare demand; those who hesitated heard a follow-up line — “I have a bad back”, “I have no change either”. Bickman himself notes that this line may have turned the order into a request for help.

False authority is deception. Invented seals, borrowed titles or actors in doctors’ coats without disclosure are not a design decision. Assessing any specific case legally requires professional review.

Three papers, three different setups. Cited here is Milgram (1963): a single condition in which the victim does not protest at all until he pounds on the wall at 300 volts. Burger (2009) reconstructed Experiment 5 from Milgram’s 1974 book, in which the victim mentions heart trouble and demands to be released — the version familiar from the film footage. Doliński and colleagues (2017) chose Experiment 2 on ethical grounds, in which the victim screams but is not ill. The three versions are related and not the same thing.

Both replications end at 150 volts. Burger and Doliński alike stop after the tenth of thirty buttons. How far people actually go is not observed in either paper but extrapolated from Milgram’s data; Burger notes that no study using a Milgram-like procedure has been published for more than three decades. For the direction of the finding the picture is nonetheless good: two countries, two independent teams, consistently high figures.

Milgram himself explains the finding situationally, not by a principle. He lists thirteen features of the situation meant to account for the high obedience: the reputation of Yale, the evidently worthwhile purpose, the victim’s voluntary participation, the subject’s own commitment, the payment, the novelty of the situation, the complete absence of any chance to confer with others. Turning that into a general authority principle for advertisements leaves out precisely this list.

What is repeated is the direction, not the setup. Burger (2009) in the United States and Doliński and colleagues (2017) in Poland arrived independently at 70 and 90 percent respectively. Both, however, stop at 150 volts, and both reconstruct a different version of the original. What is well supported is therefore that people follow a present authority a long way — not that a trust badge or an expert voice influences a purchase.