Video games often contain violent and misogynistic imagery and game-play elements. This much is beyond dispute. What is very much in dispute — and it’s not going to be resolved any time soon — is whether and to what extent this sort of content affects players in damaging real-world ways.
In light of the welcome current focus on sexism, harassment of women, and crimes like rape, it’s understandable that a new study in PLOS One claiming to have found, as the Ohio State headline puts it, that “Sexist video games decrease empathy for female violence victims” is getting some attention, with various outlets picking it up. “Here’s What Sexist Video Games Do to Boys’ Brains,” went the Time headline. “Violent, sexist games decrease empathy for real-life female victims, says study,” went the CNET headline.
But the study doesn’t really show that. In fact, on the question of whether video games decrease empathy for female violence victims, it found no straightforward evidence to suggest they do. This is a highly indirect finding that really shouldn’t be freaking anybody out.
For the study, a team led Alessandro Gabbiadini of the University of Milano Bicocca in Italy, and that included OSU’s Brad Bushman (a psychologist and one of the leading academic voices arguing that there is a meaningful connection between playing some video games and acting in an aggressive or violent manner), divided 154 Italian high-school students into three groups, with each group playing one of two games in the “violent-sexist” category (Grand Theft Auto San Andreas or Grand Theft Auto Vice City), the “violent-only” category (Half-Life or Half-Life 2), or the “non-violent” category (Dream Pinball 3D or Q.U.B.E. 2).
The participants watched a one-minute introductory video of the game in question, played “a preselected scene for 5 minutes” to familiarize themselves with it, and then “played the game alone for 25 minutes.” Afterward, participants were asked a number of questions, including how sexualized they thought the game’s female figures were, how much they identified with the video-game character, and the extent to which they endorsed a number of rather creepy “masculine beliefs,” among them “Boys should be encouraged to find a means of demonstrating physical prowess,” and “It is OK for a guy to use any and all means to ‘convince’ a girl to have sex.”
Most important, the participants were shown “one of two photos (randomly determined) of an adolescent girl who had been physically beaten by an adolescent boy” and asked to rate how “sympathetic, moved, compassionate, tender, warm, softhearted, disregarded and indifferent” they felt for her. (It should be pointed out that the GTA games are, in many countries, rating-restricted to young people anyway — it’s obviously true that teenagers play them, but this was a young sample relative to the officially intended audience.)
The obvious question is whether the participants who played a violent-sexist game appeared to have less sympathy for the female victims than those who played the other types of games. And if you look at the table running down these numbers (it’s too wide to embed), the answer is … nope. There was no statistically significant difference between the three groups in how much empathy the players said they had toward the hypothetical victims.
Which, if you look at this in a straightforward way, answers the question, no? Playing a violent, misogynistic game for 25 minutes didn’t appear to reduce the players’ level of empathy for domestic-violence victims compared to those who played other sorts of games, violent or otherwise. (I confirmed via email with Bushman that even when one only looks at the results for male players, there are no statistically significant differences here.)
Not quite, argue the authors. You see, there was an inverse correlation between masculine beliefs and the level of empathy respondents reported for the female victims — that is, the more they endorsed those beliefs, the less empathy they had. And participants who played the violent-sexist games did endorse the masculine beliefs at a slightly higher rate than those in the other groups — though we’re talking only about .3 points on a 7-point scale. Also, for those who played the violent-sexist game, there was also a correlation between identifying with the character and endorsing masculine beliefs. (As a side note, it would have been nice to have included a control group of high-schoolers who didn’t play any games and see how much empathy they exhibited toward the victims.)
So perhaps playing a violent-sexist game could lead players to endorse more masculine beliefs, which could in turn lead them to have less empathy for abuse victims. Or something. Except there’s another catch: “conditional indirect effects of type of video game on empathy for female violence victims were significant only for males in the violent sexist-game condition at values of identification with the game character greater than 4.2583[.]” Translated: Even this indirect connection only applied to players who answered 5 or above on the question of how much they identified with the character they were playing as — a character who, in the case of the GTA games, was a mass murderer. And at this point, we’re also slicing things pretty thin statistically given that only 22 males played GTA games at all, and their average score on the identification item was 4.12. (The authors do point out that the kids played these games for only 25 minutes — a fraction of the period many teenagers play video games in a week. So maybe that’s why they didn’t find much. But still: They didn’t find much.)
Given all of this, some of the press release’s language is strange. Particularly lines like this: “Bushman said it was significant that males who played one of the Half Life games — which were violent but didn’t have a sexist component — didn’t show the same lack of empathy as those who played the GTA games that combined sexism and violence.” That’s not what Bushman found at all; there was no statistically significant difference between the two groups on the empathy number. But we all know that there is a misleading-press-release problem afoot, so maybe this shouldn’t be surprising.
Anyway, if the question the researchers were trying to answer is “Does playing a certain type of video game have a big enough effect on enough players’ behavior for us to be concerned, based on this study?,” there shouldn’t be a need for such meticulous hunting for indirect effects. And if the answer to that question is, in fact, “Well, maybe playing those games has a slight effect, assuming XYZ is also true, and the player responds to a particular stimulus in a particular way, and also … ” then a more useful way to rephrase the answer would be “Nope, not really.”