0
$\begingroup$

I have a set of data where I have a number of observations over the course of a year per individual. Generally speaking, I want to know the average activeness of the individuals that participated in this project. However, not everyone completed the full year. Therefore, I have opted to use the participation days as weights to evaluate the activeness not over the whole year but over the course of participation. However, some individuals have been so extremely active that I have to consider these to be outliers. Therefore, I want to look at the median instead of the mean of the poplulation activeness.

What I have thought of so far is taking the average (mean) activeness per individual, so the mean number of observations divided by the number of participation days, and just take the median of these. However, I am not sure if that is the right way to get to the right value.

If anyone could weigh in on this and maybe give me some pointers I would be extremely thankful!

Improve this question
$\endgroup$
5
  • $\begingroup$ en.wikipedia.org/wiki/Weighted_median. $\endgroup$
    – user2974951
    Commented Jun 18 at 9:52
  • $\begingroup$ Questions about code are off topic here. But there seems to also be a statistics question here. If you edit your question to emphasize that, then it is less likely to be closed. $\endgroup$
    – Peter Flom
    Commented Jun 18 at 10:03
  • $\begingroup$ @PeterFlom thanks for the feedback. Would StackOverflow be a more appropriate forum to ask for specific help to generate this in Excel? Will edit the question $\endgroup$
    – P.Weyh
    Commented Jun 18 at 10:24
  • $\begingroup$ I don't know. I rarely visit StackOverlow. $\endgroup$
    – Peter Flom
    Commented Jun 18 at 10:46
  • $\begingroup$ Taking a median activeness seems reasonable. Taking a mean activeness would also be reasonable. If you provide a model of treatment effects, we might be able to say which statistic is more likely to show a significant effect of the treatment. But so far nothing in the question provides a framework for deciding which is “the right” statistic to calculate. $\endgroup$
    – user225256
    Commented Jul 4 at 5:14

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.