This is an old study (so we will want to assess what might have changed), but it represents a classic example of how to perform an experimental study.

Procedure

1. In 'Experiment 1', a sample of 11 games was reviewed and graded by 55 student volunteer players. Players were given a 7-point scale on which to assess the games (measuring violence [content and graphics], action, lack of pauses, difficulty, enjoyment and frustration). 2 games were then selected for further research -- these represented games which scored very differently only on the 'violence' ratings. Thus we have 'highly' and 'mildly aggressive' video games ready for the next stage.

2. In 'Experiment 2', another sample of 60 student volunteers were recruited. They were then randomly assigned to one of three groups -- one group played the highly aggressive game (from Experiment 1), one the mildly aggressive, and one played a non-video game (the card game solitaire). Volunteers then rated their own feelings immediately afterwards. They did this by using a self-administered checklist which asked them to ring words best describing how they felt -- this checklist is an apparently reliable instrument which can yield information about people's anxiety, hostility and depression. The student then also filled in a 'Departmental Questionnaire' designed to get at how they felt towards psychology in general and towards this experiment. All volunteers were then thoroughly debriefed.

Additional Comments on Procedure

1. Experiment 1 is a good way to gauge violence and difficulty etc -- these are relative values of course, but it is better to let players gauge these, rather than  outside experts? However, there is an unexplained slip between ratings of violence and ratings of aggression? Perhaps they are the same thing?

2. Experiment 2 is classically designed -- random allocations and a control group. The self-administered checklist seems like a useful way to describe feelings -- and Anderson and Ford say it also helps prevent the students from guessing what the researchers wanted (which a conventional conversation might have prompted). The debriefing helped eliminate this possibility too, as did the complexity of the findings (see below) -- especially in the deliberate inclusion of measures of depression to act as a kind of test of integrity [no-one had ever suggested that video game playing causes depression, and the researchers did not seriously think it would). The Departmental Questionnaire was designed to see if students subsequently felt hostile towards the experiment -- if they did, the researchers might be able to check to see if playing aggressive games had had some sort of transfer effect on to actual human situations (none was found).

Main results

1. Hostility was recorded at [statistically] significantly higher levels for both groups who had played video games compared to the solitaire players.

2. Hostility levels were higher in the group who had played the highly aggressive game compared to the players of the mildly aggressive one -- but not significantly

3. Anxiety levels were significantly higher in those playing the highly aggressive game, compared to those in the other two groups. The sex of the player was not significant in this respect. Anxiety levels seemed unaffected by playing the mildly aggressive game

4. As expected, no connection overall was found between levels of depression and games played, but  puzzling sex differences were detected -- men were more likely to be depressed in the 'mild' game playing group, and women more than men in the other two groups -- Anderson and Ford say there is no obvious interpretation for these results, so they can be ignored.

5. The effect of the sex of the players was only marginally significant apart from these odd cases.

Comments on the Results

1. As usual in careful studies, actual differences were always tested statistically to see if they were significant (i.e. more than would be expected by chance alone)

2. I especially liked the idea of including items known NOT to be related, as a kind of test of the methodology. I also liked the careful work to detect signs of the respondents trying to guess what the researchers wanted -- 'demand artifacts' as Anderson and Ford call such results. I think they are dead right to insist that no-one could really guess at such a complex pattern of results (especially those in 4 above), and that any guessers would probably inflate the hostility they felt towards the researchers too -- and the Departmental Questionnaire found none.

3. Anderson and Ford discuss some possible psychological theories to explain their findings. I am not competent here, but they seem to favour an approach known as 'semantic priming' -- mass media affect people 'due to the priming of semantic categories (e.g. aggression) and spreading activation along associative networks to related categories' (p.392). If I understand this at all, I think it means that mass media encourage us to map the world in aggressive ways ('kill or be killed', 'winners or losers' etc?) -- games encourage this in the world of the game, but the mapping gets spread to situation outside the games as we use these categories in real life. Anderson's and Ford's study is more modest than this though, and more focused on short-term effects -- they think that 'the aggression inherent in games may cue certain cognitions that are linked, in memory, to the aggression-related effects' (p. 392), which may mean that the games do not actually create or spread aggressive categories but can 'cue' them. Anderson and Ford suspect that the games induce 'temporary world-view changes', and that players engage in 'self-observation of highly aggressive symbolic actions' [i.e. they get 'fired up' but only while they play?]. This more limited approach also explains why the cognitive categories of the self-administered checklist can be seen as reliable indications of feelings?

Self-critical Analysis
As with all good research, Anderson and Ford are quite open about the problems that remain:

1. No actual theory was rigorously tested by this study -- maybe 'semantic priming' should have been pursued more systematically?

2. Problems remain in interpreting what it actually was about the games that produced the effects -- we have the items on the 7-point rating scale in Experiment 1, but there may be other unmeasured factors. Anderson and Ford suspect that similarities between games and actual human situations might be a factor -- but this is not actually tested here.

3. Their study focused on short-term effects, but long-term ones might be more important. There may be no transfer at all between games and life, or there may be cumulative effects. No-one knows until they do the research, but 'attention to these possibilities is clearly warranted' (p. 400)

4. Other issues arise too. (a) Mild levels of hostility and anxiety can be 'good', since they can aid in learning how to cope with these feelings. (b) Various 'indirect' effects may exist, connected to the whole 'social emotional and intellectual development' [and cultural experience] of youth, video games, and video arcades.

Final comment
Clever stuff this -- but, as usual, very mixed in its findings. Nice methods on offer --especially those designed to see if the respondents are really reporting what they feel or whether they are trying to guess what they should be feeling. As usual again, it is odd to find such hard-nosed measurers resorting to the mysteries of 'long term' and 'indirect' effects at the end though (and see my comments on Belson's study here)

Now here is something for you to consider --what important changes have there been since 1987, do you think? Remember that it is not enough to list changes and assume that the consequences are self-evident. Other students have said that the games are far more graphic and violent than they were then, for example, which is correct -- but has this made them more or less likely to induce or transfer aggression? (It might be the case that more graphic violence is more effective -- but also that kids are more used to graphic violence). The same goes for statistics showing the greater use of games. Are there any important differences at all, then?

Above all, the techniques seem as strong as ever. Certainly we should inquire into any study, including recent ones, to see if there has been an advance in methodology since Anderson and Ford? Actually now there is one -- a more recent study by Anderson and Dill.