'A Girl Is the Champion' Cut Adults' Ability Ratings in Two Studies. For Children It Did Not. The Games Were Invented.

The headline reads "A girl is the Gazzer champion." Gazzer does not exist, which is the point. Elsewhere in the session, another invented game announces its winner by name, in the style of "Joan is the champion." Each volunteer sees both styles. The researchers rotate which game gets which. Then come the questions: who plays, who is good at it, and would you want to try it?

That setup comes from a paper in the Journal of Experimental Psychology: General. Its authors are April Bailey of the University of Edinburgh, Maya Kreitman of The Chicago School, and Andrei Cimpian, the New York University psychology professor who is senior author. It pools eight experiments with 326 children aged 4 to 10 and 788 adults, all in the United States. Bailey's summary in the NYU release is that well-meant messaging "does not always promote more positive perceptions of women or girls."

Participation took the most consistent hit

The firmest finding is about who people think takes part. After a gender headline, volunteers inferred that fewer people of that gender play the game. That showed up in five of the seven experiments that used gender (1a, 1b, 1c, 2b and 3b). Pooled across experiments, the shift was -0.24 standard deviations on the indirect measure and -0.32 on the direct one, which the authors call small to medium.

Where the press release and the paper part ways

The ability result is where the two documents differ. The NYU release says gender-marked messages diminished perceptions of girls' and women's abilities "among children and adults of both genders." The paper's abstract is narrower. It reports lower ability "among adults."

The adult evidence is real. In Experiment 2b, 105 university students rated the referenced gender as less capable after the gender headline than after the name headline. The gap was B = -0.40 (p < .001), roughly half a standard deviation. Experiment 3b, with 179 online adults, found it again, and 3b was preregistered while 2b was not. Pooled across experiments, including the children, the ability estimate was -0.25. Its 95 percent interval ran from -0.48 to -0.01, so one edge nearly touches zero.

The children did not follow. In Experiment 2a, 90 children split by gender. Girls rated the winner's gender directionally lower and boys directionally higher. The authors caution that neither shift was significant (p = .190 for girls, p = .111 for boys). The pattern did not return in Experiment 3a, which had 174 children.

Interest is thinner too. Volunteers wanted to try the gender-marked activity less in three experiments (1b, 2b and 3a). In the other four the gap was not significant by the supplement's own tests, including Experiment 1c, where 293 online adults produced p = .100. The pooled interest effect was 0.10, the smallest of the four.

Invented games, and a group that flipped the result

The authors chose invented activities so nobody walked in with prior beliefs. That cleans the test and limits the claim. For real fields such as coding or chess, where people already know one gender is underrepresented, the paper says the messaging might reinforce those beliefs. It lists that as an open question. Nobody signed up for a course or stuck with a task. Every sample was American.

Experiment 4 gives the result a boundary. In it, 121 adults read about invented beings called Daxes. The headline said either "A square Dax is the Gazzer champion" or "Charlie is the Gazzer champion." Marking the group did not backfire. Volunteers rated the marked group as better at the activity (B = 0.18, p = .043). The authors read this as evidence that the "why mention it?" inference needs knowledge of how people talk about gender, which nobody has about Daxes. The win is narrow. Participation ratings leaned positive without reaching significance (p = .098 and .119).

The supplement also reruns most experiments using only the first game a person saw, which halves the data. Every effect kept its direction and several lost significance. Experiment 2b's ability result went from B = -0.40 (p < .001) to B = -0.18 (p = .213).

“What's counterproductive is highlighting women's gender as the reason for celebration.”

What the authors suggest instead

The labels in question are everywhere: Space4Women, Women's Chess Champion, Coding for Girls. The paper opens with the 2019 mission of Christina Koch and Jessica Meir, the first all-female spacewalk team, and the headlines that called it an "all-women spacewalk." It later turns to Space4Women, a United Nations program. The authors read the name as inviting the thought that space is typically "4Men." A related result fits that reading. A game called "Gazzer for Girls" led children and adults to link plain Gazzer with boys.

The suggested fix is to show women succeeding without framing the achievement as being for them. Gender can still come through in names and photos. In Experiment 3, same-gender role models led children and adults to infer that their own gender plays more and does better, even with no label attached. The authors report they never found gendered messaging to be equity-promoting.

The coverage that loads, including Phys.org's, repeats the NYU release almost word for word. For now the evidence is a warning about wording, drawn from games nobody plays.