Directional is the new null? A comment on Bushman and Anderson (2021).

Bushman and Anderson (2021) have recently argued that the evidence of harm after playing violent video games is so strong that this effect should be our starting point for future research. They base this claim on an argument that: (a) many professional bodies agree with this opinion; (b) strong theories, such as their General Aggression Model (GAM), predict such outcomes; (c) experimental and meta-analytic studies back such a claim; and (d) people who do not obtain this effect are in the minority and their studies have methodological shortcomings or they misanalyze their data. It is argued here that this is not consistent with the precepts of falsificationism and that: (a) their argument from authority is problematic; (b) they appear unconcerned with, or unaware of, increasing null studies, particularly missing out international research or preregistered studies; (c) the majority of research groups outside of Bushman/Anderson and their coauthors do not concur with their results; and (d) there are theories (e.g., the Immersive Media Prediction model) which better account for the data than their GAM. It is also argued that when theories and data collide it is the theories that need revision not the data, demand effects of researchers are strong in gaming research, and Bushman and Anderson’s results may themselves have been influenced by their methodology rather than accurately describing a naturally occurring weakness in the human condition. It is also argued that we need a more collegial approach to gaming research and make suggestions to facilitate this shift.