0
$\begingroup$

I recently performed an experiment to implement Gamification (a simple points system) into a web application to see if the points had an effect on the users in terms of usage.

We have data on how much time users spend daily in the application, 1 month before Gamification was implemented, and 1 month after. The users selected for the experiment were sampled into 3 different groups:

  1. Control group (users who did not use gamification) - 300 users
  2. Points group (users who used the points system) - 300 users, and
  3. Performance group (top 10% points earners) - 156 users

What statistical tools can I use to evaluate this data?

All I have done now is to look at how the total time for each group changed from the first month to the next, and got:

  1. Control group: +1%
  2. Points group: +28%
  3. Performance group: +49%

However, this is after I've removed what seemed to be outliers (new users appearing in the middle of the experiment etc.). Prior to this, it looked like this:

  1. Control group: +41%
  2. Points group: +105%
  3. Performance group: +110%

Is this enough proof that the points did, in fact, have an impact on usage?

$\endgroup$
3
  • $\begingroup$ Just checking that the 3 groups contain the same people before and after. $\endgroup$
    – Peter Flom
    Jun 1, 2014 at 9:50
  • $\begingroup$ @PeterFlom yes. For each group, there are the same people before and after. :) $\endgroup$
    – Mickel
    Jun 1, 2014 at 10:11
  • 1
    $\begingroup$ See this question and its answers. Does that answer your question? If not, please say what else you are asking about $\endgroup$
    – Peter Flom
    Jun 1, 2014 at 10:44

0

Browse other questions tagged or ask your own question.