3 ms·

Did you read the entire paper, or just the abstract? I ask because I only was able to get the abstract, so there might be something in the rest of the paper fo

Did you read the entire paper, or just the abstract? I ask because I only was able to get the abstract, so there might be something in the rest of the paper for your arguments. However, the abstract seemed very careful not to make overly broad conclusion. It says the playing violent videogames is a good predictor for being agressive, and being agressive is not a good predictor of playing violent video games. This is not sufficient to claim causation, which they do not (in the abstract at least), however this type of corralation that exists only where A preceeds B definantly suggests it. As to you concerns about sampling, they seem to be following best practices for this type of study, random. As long as both groups you are comparing are selected using the same random, your results are valid (for the group which they were selected from). Most statistical tests (which I assume were used in the main paper) are designed assuming the groups are random, and will provide valid statistics.

lutusp 14y ago
> However, the abstract seemed very careful not to make overly broad conclusion. In general, but when they say "... long-term relation between violent video games and aggression, as well as the specific game characteristics (e.g., violent content, competition, pace of action) that may be responsible for this association", they're moving from evidence to opinion, an opinion not remotely supported by the study. The only way they could sort out causes and effects in a study like this would be to compel some people to play video games, and compel other people not to. The subjects are pre-teenagers, and that is not going to happen. Such a study is called "prospective", and they are rare in psychology. Even with adults, prospective designs are terribly expensive -- and there have never been a study of that kind on this topic. > As to you concerns about sampling, they seem to be following best practices for this type of study, random. But the study has no control. Instead of choosing two groups as identical as possible and applying different stimuli to each, they rank subjects based on their existing behavior, meaning there is no control group as that term is understood. > Most statistical tests (which I assume were used in the main paper) are designed assuming the groups are random, and will provide valid statistics. But there aren't two groups selected in advance and randomly assigned a behavior. The study participants are pigenholed based on their observed behavior, the reverse of the strict statistical approach. On can see this without knowing the details, simply by imagining the opposite -- locking up a bunch of 12-year-olds and compelling them to play, or not play, video games. Obviously that didn't happen. > As long as both groups you are comparing are selected using the same random ... There weren't two groups selected in advance and subjected to different conditions, as in a truly scientific study. This study design is called "retrospective" (meaning subjects are chosen after the fact, not before), and is notoriously unreliable. http://en.wikipedia.org/wiki/Retrospective_cohort_study http://en.wikipedia.org/wiki/Retrospective_cohort_study A quote: "Caution needs to be exercised in particular with retrospective cohort studies because errors due to confounding and bias are more common in retrospective studies than in prospective studies."

lutusp 14y ago

> However, the abstract seemed very careful not to make overly broad conclusion. In general, but when they say "... long-term relation between violent video games and aggression, as well as the specific game characteristics (e.g., violent content, competition, pace of action) that may be responsible for this association", they're moving from evidence to opinion, an opinion not remotely supported by the study. The only way they could sort out causes and effects in a study like this would be to compel some people to play video games, and compel other people not to. The subjects are pre-teenagers, and that is not going to happen. Such a study is called "prospective", and they are rare in psychology. Even with adults, prospective designs are terribly expensive -- and there have never been a study of that kind on this topic. > As to you concerns about sampling, they seem to be following best practices for this type of study, random. But the study has no control. Instead of choosing two groups as identical as possible and applying different stimuli to each, they rank subjects based on their existing behavior, meaning there is no control group as that term is understood. > Most statistical tests (which I assume were used in the main paper) are designed assuming the groups are random, and will provide valid statistics. But there aren't two groups selected in advance and randomly assigned a behavior. The study participants are pigenholed based on their observed behavior, the reverse of the strict statistical approach. On can see this without knowing the details, simply by imagining the opposite -- locking up a bunch of 12-year-olds and compelling them to play, or not play, video games. Obviously that didn't happen. > As long as both groups you are comparing are selected using the same random ... There weren't two groups selected in advance and subjected to different conditions, as in a truly scientific study. This study design is called "retrospective" (meaning subjects are chosen after the fact, not before), and is notoriously unreliable. http://en.wikipedia.org/wiki/Retrospective_cohort_study http://en.wikipedia.org/wiki/Retrospective_cohort_study A quote: "Caution needs to be exercised in particular with retrospective cohort studies because errors due to confounding and bias are more common in retrospective studies than in prospective studies."