The slow rollout of Vox Games content continued today with Brian Crecente’s article on a recent paper by Douglas Gentile and his group, which I didn’t really care for. While Crecente did some things right, and notably improved on typical science reporting by actually providing a link to the paper (an unacceptably rare practice), I thought the article fell short.
The standard of success in science is whether research is convincing to peers. For this reason it is essential to get an opinion on the research from a scientist in the field not associated with the project. Crecente didn’t do this, even though there are some obvious options: Chris Ferguson, Wai Yen Tang, and Lawrence Kutner would all have been reasonable choices, though Ferguson is probably tired of getting the “contrasting opinion” call every time Gentile publishes. Crecente also could have looked through the citations in the article to find some experts to discuss it with. Fred Zimmerman is one possibility from this list, as he has frequently written on similar topics, but not co-authored any papers with Gentile. The point here is not to dig up any old crackpot who disagrees with Gentile’s conclusions, but to find somebody who can look at the research with a fair and critical eye.
The external perspective is particularly desirable here because Gentile is (perhaps unjustly) perceived as an axe-grinder. That Gentile has written numerous articles and books trying to link videogame violence with real-life violence and aggression, and that he was Research Director at the National Institute on Media and the Family, is something that Crecente might have noted.
In addition, external perspective might have prevented Crecente from both overstating and understating the implications of the research. The lede paragraph mentions a vicious cycle of reinforcement between videogames and attention problems, and also mentions a comparison with television. The former is good, and reflects the key finding. The latter isn’t really addressed by Gentile et al.’s paper. Crecente reports several ideas about how games might cause attention problems, but these come from the paper’s introductory material. Nothing in this research directly addresses any mechanistic hypothesis, but these speculations constitute a significant proportion of Crecente’s article.
The idea that reducing videogame exposure (VGE) could address attention problems seems an optimistic take on the results. A very simplistic analysis of these results indicates that VGE at any given time point explains 1-4% of the variation in current or subsequent attention problems (Table 1). That’s something, but it’s not much. The VGE-attention correlation grows stronger in the third wave of surveys, which may mean that newer games or systems are more likely to cause attention problems, or that the effect on attention grows stronger as a child ages. Or, this could simply reflect intrinsic noise in the data. Here an outside expert’s opinion would be valuable.
The path diagrams in the later part of the article are meant to illustrate possible causal relationships identified through mathematical modeling. They show that attention problems are a better predictor of future VGE than VGE is of future attention problems. However, there is a modest predictive relationship both ways, and Crecente understates the findings by failing to report that a causal effect, rather than just a correlation, was found (notice how some commenters missed this).
While the finding may imply, as Gentile believes, a vicious cycle of reinforcement, it may also signify that he has simply got the causal relationship entirely backwards, and that the power of VGE at wave 2 to predict attention problems at wave 3 is simply bleed-through of the (unmeasured) attention problems at wave 1 causing VGE at wave 2. The results on impulsiveness seem to argue against this, but there the predictive capacity is even lower. Or, there may be a third factor, perhaps related to parenting or genetics, that promotes both. In my opinion Gentile et al. are unjustly dismissive of this possibility, and Crecente does not even mention it.
With or without an external opinion, Crecente’s article doesn’t do a good job of putting the research in perspective. How surprising or novel is this finding? How does it fit with other things we know about children’s use of and reaction to media? Might cultural factors (the research was performed in Singapore) affect the causal relationships found here? What recommendations for parents does the research imply? On these subjects, Crecente’s article gives us less than we would have gotten had he merely reproduced Iowa State’s press release.
My point here is not that Crecente is a bad journalist, or even that this is a terrible article. As far as I can tell, it contains only one minor factual error (journal name), and the errors he did make are ones that happen all the time in science reporting. On the positive side, Crecente reports the research methods and central conclusion clearly and correctly, and he provides a link to the paper. That may sound like damning by faint praise, but in the context of the media’s chronic inability to accurately report science it is actual straight-up praise. However, the absence of an external opinion is a serious problem, and certain omissions in the article can mislead the audience. As science reporting in gaming media goes, it’s not bad (e.g. this), but Vox Games aspires to a higher standard.