Need help with inferential statistics?

It will be a two-sided test, since your null is that the parameters are equal. How about using a two-sample t-test on the difference in times spent in play behavior? (It's better than a z test because your sample sizes may not be the same.)