People taking part in a study usually work out what it is supposed to show, and then behave accordingly. We told one group of participants that the keyboard they were testing came out of research and said nothing to the other: the first group typed faster and rated the user experience higher, on exactly the same keyboard. The paper gives guidelines for wording instructions and designing studies so that demand characteristics do not do the work for you. https://doi.org/10.1016/j.ijhcs.2024.103379

Studies of virtual avatars almost always rely on the same control: move the avatar out of sync with the person and see whether the sense of owning the body breaks. We tested that control as a usability question about the avatar. An online study with 202 participants found that different implementations of asynchrony produce different responses. A think-aloud usability study with 16 participants showed why: the conditions felt strange or unpleasant, and some people imitated the avatar, which kept the illusion intact rather than breaking it. The paper gives recommendations for choosing and reporting these conditions. https://doi.org/10.1145/3706598.3713506

We took every CHI paper that shared both its data and its analysis code and tried to run the analyses ourselves. Only 49% reproduced. We then surveyed and interviewed the authors about what got in the way, and turned their answers into practical guidelines for building an analysis repository that someone else can actually run. https://dl.acm.org/doi/full/10.1145/3772318.3791129

We asked 60 people, aged 21 to 41, which apps on their phone matter to them and why. Apps turned out to be special for reasons that rarely show up in UX metrics: they support a personal goal, they hold memories, they carry a relationship with someone. People described their identity as bound up with certain apps, including ones they find annoying or depend on more than they would like. They also actively manage that dependence, through how they arrange their phone and the rules they set for themselves. The anonymised transcripts are published alongside the paper. https://doi.org/10.1145/3613904.3642820