Excessive Video Game Playing Indicates Poor Satisfaction With Life

Last year, I did an informal experimental study about Starcraft 2, a highly complex real time strategy game. Your success in Starcraft heavily relies on multitasking (in it’s every-day use). Multitasking in this sense means that you need to keep a complex mental loop for actions which need to be performed in a rapid succession. Professional-level players normally perform up two 300 actions per minute (actions = button presses).

The study had two main goals. First, to investigate the relationship between playing Starcraft and multitasking abilities. Since I used a flash game and used self-report measures, the methodolgy in this part could potentially skew the results.

Second, and this part was not announced, I read a lot of complaints from people who play a lot of Stracraft 2 in forums. I wanted to check whether these qualitative statements had some quantitative foundations. In specific, I investigated whether the amount of playing SC2 is correlated with  general satisfaction with life.

In total, 70 participants took part in the study. For investigation of multitask abilities I used the flash game Multitask from Kongregrate. The participants played 10 times and entered their results manually into a google form in which I continued the study. Life satisfaction was measured using the widely employed Satisfaction with Life Scale by Ed Diener. After that, participants filled out questions about their playing behavior, most importantly how many hours per week the play, and additionally demographical data.


From the initial 70 participants, 20 had to be excluded from the study. Three because of missing data and 17 due to outlier detection (mostly because they reported several multitask results below 20 seconds whichmeans that they didn’t really play). That means, that 50 participants were included in the final analysis.


The longest time reported was 262 seconds. The average of all participants was 70 seconds. I didn’t find any connection between “hours of SC2 played” and multitasking ability. Linear models also revealed no interaction with league (which is an in-game indication of a player’s skill). Either, playing more doesn’t improve one’s multitasking abilities or it improves multitasking abilities which are not measured by the Multitask game I used or the results are skewed because I relied on manipulable self-report measures and participants did not report the correct results.

[image loading]

(Looks messy, doesn’t it?)

I found a trend (p = .081) that master’s players (the highest in-game league) lasted on average 20 seconds longer in the multitasking game compared to the other leagues. This is an indication that multitasking abilities such as measured by the Multitask game are really reflected in Starcraft 2 skill. As expected, age significantly correlates with multitask ability (p = .003; r = -.400). Those in their early 20s were best. (see What do the numbers mean? below)

Life Satisfaction

Playing Starcraft correlates negatively with general satisfaction with life (p = .001; r = -.441). The more you play, the less satisfied you are. Don’t play so much! To give you an idea, most players in my data set played around 15 hours per week (M = 14.60; SD = 11.9), the minimum was around 1 hour and the maximum was around 40 hours. Also, not surprisingly, the older you get, the less time you invest in SC2 in total (p = .006; r = -.465).

What do the numbers mean?

The p-value is the estimated probability of how likely we found a difference between two groups although there is none. For a p-value of .03 this would mean that if we generated results completely random, in 3% of the cases, we would find this “difference”.

The r-value is an estimate of how big the correlation between the two things is we measure. In the example above, we look at the variance in the satisfaction with life between the participants. Now we overlay this variance with the variance from the “hours of Starcraft played” and check how much of the variance in the one data set can be explained by the other. A value of .441 means that 44% of the variance in the satisfaction with life scores can be explained by the variance in the “hours of Starcraft played” data. Given that one’s satisfaction with life can potentially influenced by thousands of things (like social life, sleep, achievements, job, etc) such a high correlation is astonishing!

I think that the connection between amount of hours of weekly play and life satisfaction is interesting, I don’t believe that playing Starcraft causes one to be less satisfied with one’s life. The more probable explanation is that, if you are less satisfied with your life, lack social inclusion or have problems at home, you tend to retract yourself from society, because it’s the source of your problems, and approach alone-activities such as playing video games extensively.

You may also like...

5 Responses

  1. Tim says:

    Interesting finding!

    Have you considered how deep the age element might go as a confounding variable? It’s possible that younger participants might report less satisfaction with life simply because of their youth. IMO, those in their later 20s and 30s tend to have settled down emotionally a bit more (on average) and also settled in better financially, etc., while teens and early 20s have had less time to kind of find their way in the world.

    Maybe I’m being anti-teen? 🙂

  2. Miips says:


    A very interesting survey! I was wondering though exactly which tests did you use to get your p-values? I was also wondering how does the data generally look, for example in the case of hours played in a week it seems like you would have a very skewed distribution. Could you perhaps give your readers both the mean and median in each case to understand the data better? Also aren’t p-values over 0.05 generally considered to be statistically insignificant? At least my statistics professors advice us to use extreme caution if we are claiming anything based on such high p-values..

    • msauter says:

      I will try to get into this data set again after I analyzed the current study. The p-values were mostly derived from pearson correlations. You are correct, we need to be extremely cautious with such high p-values and I hope I phrased it in a way that you didn’t hink I wanted to imply otherwise.

  1. 24 February 2015

    […] 24th February 2015 msauter Leave a comment […]

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.