If violent video games were actually making people violent, what would you do?

[Jordi]

For instance, if you take a random sample of people, and randomly let half of them play a game, but not the other half and it turns out that the game-players are more aggressive afterwards, you have totally just shown that that game in that population caused more short term aggression than whatever the hell the control group was doing.

Thats also correlation, not causation.... unless you get a practically 100% result AND can explain the failures.

You seem to think that partial causation is not causation. That is wrong, because you can transform this partial causation (A causes B in X% of the cases) to "regular" causation (A causes a X% chance of B) without changing the meaning.

Then you would have a qualitative experiment you've repeated instead of a quantitative one. Also if you just "let" them play a game they're making a choice. If they're making a choice to play the game your entire experiment is shot for causation as you can't disprove that B leads to A, IE the violent ones picked up the game. You could force people to play... but that adds another external factor. Theres a reason cause and effect are seldom mention in psychological examples.

In general:
Quantitative experiments show correlation
Qualitative experiments show causation

I'm really wondering what you mean by "qualitative" experiments here and how you think they can show causation. The qualitative research that I'm familiar with basically involves asking people questions. Aside from the fact that they can either lie, or just fail at introspection, or the researchers can fail at observation, it seems to be that the best possible result you could get are fruitful avenues to pursue with quantitative research. How will you ever prove/infer/show anything without quantitative methods?

If your wanting to convincingly prove causation you would have to find a procedural and repeatable experiment. You would need to break the event chain into small testable parts and have an explanation for each causal step. The really simple litmus test, can you explain it? If you can't, then your experiment shows correlation. Statistics based approaches may sometimes imply causation, but this is a huge challenge even for physiological experiments, let alone psychological ones. And it is an implication, at some point a strong enough correlation will be the result of a causal factor. Although the problem is that it goes both ways, even showing a 100% correlation doesn't give you the direction of the causation, you must find that via qualitative means.

You don't need to explain every little thing. If there are a hundred intermediate steps between A and Z (i.e. A causes B, which causes C, etc.), A is still causing Z. The intermediate steps may be interesting, but they are not necessary to know for the causation to be shown.

Also, if you have found a correlation between two variables, there are really only three options:
1) A causes B
2) B causes A
3) An outside factor is causing both A and B

If you can eliminate (2) and (3), you have shown causation. If you take randomized groups, you can calculate the odds of not having eliminated them. The idea is that you assume that the groups are equal before the experiment, so if at the end the test group has more "B" than the control group, you can assume that it wasn't a pre-existing condition. If there is an outside factor it would work just as much on both groups. And if B was causing A, but not the other way around, that wouldn't explain the increased amount of "B" in the test group. Of course, the groups aren't entirely equivalent, but you can basically calculate the odds that that affected your score. It is true that it is impossible to definitively prove causation. The most you will ever get is a kind of probability that A causes B.

BTW it seems to me that experiments like these are also much more repeatable than qualitative research, because it depends far less on the particular people in the study and the researchers carrying it out. I'm really curious how you would prove something with that. Can you give an example?

 

[Galaxy Roll]

I would work hard to move the violence from them.

With my fists.

 

[badgersprite]

Alcohol is something that definitely causes or is associated with lots of violence that wouldn't occur otherwise, yet we don't ban alcohol or stop drinking it. Anything can result in bad side effects or negative consequences if it's taken to extremes. But we're free individuals; people need to be able to moderate and choose their own behaviour, even where there are risks involved. Anyone who plays video games to the extent that it does alter their personality or, hypothetically, to the point that games do make them violent, and yet keeps on playing to their detriment is either an idiot or has some serious problems and warped priorities.

If you can't moderate something, and you let something negatively effect your life, you should seek help, but to assume that everyone is like that and can't regulate their own behaviour is paternalistic.