The subheading is "and women overestimate men's preference for femininity". And furthermore, the study focuses specifically on <i>facial appearance</i>. Splitting these up like this and burying the context reminds me of the story of an editor magnifying gender conflict in the fantastic essay "How Americans edit sex out of my writing":<p><a href="https://web.archive.org/web/20221006180631/https://europeanreviewofbooks.com/how-americans-edit-sex-out-of-my-writing/en" rel="nofollow">https://web.archive.org/web/20221006180631/https://europeanr...</a><p>Anyway, the methodology for <i>quantifying</i> how masculine or feminine a face is in the study was kind of interesting, though it wasn't brought up in the press release.<p>><i>Transformation was achieved by applying or subtracting the linear difference between the average male face shape and female face shape to target face (i.e. average faces for illustration and each individual composite or base face for stimuli. The distance from the average female face shape to the average male face shape was equivalent to a +100% increase in masculinity. Thus, each of the 118 faces contributed to the definition of the dimorphism vector along which base faces were manipulated</i><p>So some notion of deformation — the medical imaging community uses the concept of a "deformation vector field" — is being applied to project face shapes onto an axis of feminine to masculine.<p>One limitation of face-shape studies like this is that the faces are always shown in orthographic projection on a flat screen, which creates a predictable distortion of facial features, tending to widen and flatten faces compared to what you see in real life. This is the "camera adds ten pounds" effect:<p><a href="https://petapixel.com/2016/07/28/camera-adds-10-pounds/" rel="nofollow">https://petapixel.com/2016/07/28/camera-adds-10-pounds/</a><p>Since the preferences for female faces were both predicted by women and chosen by men to be much more feminine (>100%) than the displayed faces, it seems possible that the method of display may have been slightly distorted as though projected from infinity, like the telephoto examples in the article, where the subject's face is widened.<p>But perhaps the greater lesson is that it isn't so simple to display a 3D object on a flat screen. In fact, the method of projection isn't even mentioned in the study, much less the article!