Does the Gender of an Authority Figure Change Obedience? What a 2025 Milgram Replication Found
If you're studying Social Influence, you already know Milgram (1963). It's probably the single most cited study on the entire specification, and it's also one of the most heavily evaluated, mostly with the same handful of AO3 points: it lacks temporal validity, it raises ethical concerns, the sample was culturally narrow. All fair points. All slightly tired by the time you're the fortieth student your examiner has read this year.
But if you want an evaluation point that makes an examiner sit up rather than skim past like the other thirty-nine, here's a genuinely different one. A 2025 study went back to Milgram's paradigm and tested something the original experiment never actually varied: what happens if the person giving the orders is a woman instead of a man?
The question Milgram never actually answered
It's worth being precise about what Milgram did and didn't test, because this is where a lot of students get muddled.
Milgram's original 1963 study used only male participants: the people delivering the shocks were all men. He later ran versions with female participants too (1974), and found no meaningful difference in how obedient men and women were. That's the participant's gender.
What Milgram never systematically varied across his classic studies was the gender of the experimenter, the authority figure giving the instructions. Every "teacher" in his most famous studies took orders from a man. Whether a woman in that same position of authority would have been obeyed just as readily was never actually tested. As the authors of this new study put it, it's "a critical gap in the literature."
That's the gap this 2025 study, by Grzyb, Dolinski and Cantarero, set out to close.
What the new study did
The researchers ran two linked studies.
Study 1 was a lab experiment with 80 participants, using a modified, ethically adjusted version of Milgram's procedure (based on Burger's 2009 "obedience lite" paradigm). Participants believed they were taking part in a memory study and were instructed, by either a male or female experimenter posing as a psychology professor, to deliver increasingly intense electric shocks to a "learner" (in reality, a confederate) who made scripted mistakes. Shocks were capped at 150V rather than Milgram's full 450V, and the session ended automatically after ten shocks or an early refusal.
Study 2 moved online, with 793 Polish participants imagining the same scenario rather than experiencing it directly, and also completing a questionnaire measuring their own sexist attitudes.
What they found
In the lab study, 88% of participants fully complied under a female experimenter, compared with 90% under a male one. Statistically, that's not a meaningful difference (χ²(1, N=80) = 0.13, p = .723). A Bayesian analysis found the data were more than three times more likely under "no difference" than under "there is a difference."
The online study reached the same underlying conclusion (no significant effect of the authority figure's gender) through a larger, more statistically powerful sample, though the way obedience was measured differed enough between the two studies that it's more accurate to say each study independently found no gender effect, rather than treating them as reporting the exact same number.
What this means for Milgram's theory
This is the part worth actually understanding, not just quoting.
Milgram never argued that people obey because of who specifically is giving the order. His explanation was about legitimacy of authority: we obey people we perceive as occupying a recognised position of authority within a trusted system (a lab, a hospital, an institution), regardless of their individual identity. If that's the real mechanism, then swapping the authority figure's gender shouldn't move obedience much, because gender isn't what makes someone a legitimate authority in the first place.
That's exactly what happened. So rather than undermining Milgram, this replication supports one of his central theoretical claims, and does it with contemporary, gender-specific evidence that the original studies simply didn't have. That's a far stronger AO3 point than the generic "it's old, so who knows" line: you can now say specifically what was tested, what was found, and why it fits Milgram's own explanation of obedience rather than contradicting it.
The more interesting finding: it wasn't gender, it was attitudes
Here's the bit I actually find more interesting than the headline result.
In the online study, the authority figure's gender didn't predict obedience. What did was the participant's own score on a measure of sexist attitudes: the more sexist the participant, the more willing they were to keep going, regardless of whether the instructions came from a man or a woman (β = .30, p < .001).
What makes this genuinely good science rather than just a convenient finding: it's not what the researchers expected. They'd originally hypothesised that sexist participants would show lower obedience specifically to a female authority figure: an interaction effect. What they actually found was a simple, direct link between sexism and obedience that had nothing to do with the authority figure's gender at all. Being wrong about your own hypothesis in an informative way is a good sign a finding is real, not just convenient.
The researchers tentatively connect this to the idea of an authoritarian personality, already on your specification as a dispositional explanation for obedience: the idea that some people are simply more predisposed to defer to authority and enforce rigid hierarchies. They're careful to call this interpretation tentative, since they didn't directly measure authoritarianism itself, only sexism. But it's a genuinely interesting thread, and their own conclusion is worth quoting directly: "the capacity to induce obedience, even of a troubling nature, is not bound by gender." Pathological authority, as they put it, "knows no gender."
A genuine limitation worth knowing
Two honest caveats, both worth knowing if you're using this as evidence rather than just a headline.
First, in the lab study, obedience was so high (88% and 90%) that only nine of the eighty participants disobeyed at all. The researchers themselves flag this as a possible ceiling effect: when almost everyone complies, there's very little variation left to actually explain, which limits how much the study can tell us about what predicts refusal. It's a real methodological limitation, not just a footnote: a strong finding and a statistically powerful one aren't always the same thing.
Second, worth remembering that 88% and 90% aren't directly comparable to Milgram's famous 65% obedience rate. This study used a capped, lower-intensity version of the procedure (150V rather than 450V), specifically for ethical reasons, so a higher completion rate here doesn't mean people have become more obedient since 1963. It means a less severe procedure produces less resistance, which is itself a reasonable thing to point out in an evaluation paragraph.
There's also a fair generalisability question. Study 2's sample was Polish, and the authors themselves note that Poland has a narrower gender pay gap (around 10%) than the EU average (13%), and a higher proportion of women in managerial roles (43% versus a 34% EU average), not as a flaw, but as a reason the findings may not transfer neatly to cultural contexts with more pronounced gender inequality.
How to use this in your exam
If you're building an AO3 evaluation paragraph on Milgram, this gives you something specific to say instead of something generic:
Supports Milgram's legitimacy-of-authority explanation, since obedience held steady even when the authority figure's identity changed, precisely what his theory would predict.
A genuine point on temporal validity: obedience here (88% and 90%) was considerably higher than Milgram's original 65%, which you could use to argue that something about how people respond to authority may have shifted since 1963, supporting the temporal validity concern rather than just asserting the study is old.
But push that point further: this comparison isn't necessarily a fair one, since this study capped shocks at 150V rather than Milgram's full 450V. A less severe procedure would be expected to produce less resistance regardless of the era, so the higher obedience rate may reflect an easier procedure rather than a genuine shift in how obedient people have become. Raising that as a counter to your own point, rather than stopping at the first observation, is exactly the kind of layered evaluation that separates a mid-band answer from a top-band one.
Most students can describe a study. Fewer can build a point, then question their own point, in the same paragraph, and that's the actual skill examiners are marking.
If turning research like this into marks consistently, rather than just understanding it, is the part you find hardest, that's exactly what Accelerator is built around: identifying where knowledge isn't converting into marks, and closing that gap directly.
Source: Grzyb, T., Dolinski, D., & Cantarero, K. (2025). Authority Knows No Gender – Gender Effects in Exerting Obedience in Milgram's Experiment. Social Psychology, 56(2), 85–97. https://doi.org/10.1027/1864-9335/a000575 (open access)
👋 About the Tutor
I'm a full-time independent tutor with over five years of experience helping students master AQA A-Level Psychology. With a Master's in Political Psychology (Merit) and a BSc in Criminology (First), I move students beyond the "overwhelmed" phase by focusing on high-level subject mastery and flawless exam technique. I am Enhanced DBS-checked and safeguarding trained.
How to Work With Me
👥 Study Club: A supportive weekly space to revise key A-Level Psychology topics together. We break down tricky content, practise exam questions, and build confidence.
🚀 Masterclasses: Intensive courses designed for the crucial build-up to exams. These sessions focus on the topics students find most challenging, helping you consolidate knowledge, strengthen evaluation skills, and feel more prepared walking into the exam.
📚 1:1 Tutoring: Personalised support tailored to your exact needs. These sessions focus on strengthening understanding, building exam technique, and boosting confidence so that learning starts to feel manageable again.
Comments