+ Reply to Thread
Results 1 to 4 of 4

Thread: Appropiate test/s for bounded data sets?

  1. #1
    Points: 32, Level: 1
    Level completed: 64%, Points required for next Level: 18

    Posts
    3
    Thanks
    2
    Thanked 0 Times in 0 Posts

    Question Appropiate test/s for bounded data sets?




    Hello! I am comparing between groups of data that have both lower and upper bounds. Most of the time my data sets qualify as being normally distributed and I have been using anova and Tukey post-hoc tests. However, in some experimental group data sets where the mean is near the lower or upper bound, the data is not normally distributed even though the control data sets are still normally distributed.

    So, I would like to be able to compare between data sets with different types of distributions; normal and skewed towards and upper or lower bound. But, I see that anova and Tukey is only for normally distributed data and kruskall-wallis/mann-whitney is for non-normally distributed data which have the same types of distribution. My data does not fit either of these. I have tried a bootstrap test and this seems to give me very liberal p-values much lower than I expect. I would like to use a more conservative test, but I am not sure what is appropriate. Preferably I would like to use something that could be easily implemented in Matlab.

    Thanks for the help!

  2. #2
    Human
    Points: 12,676, Level: 73
    Level completed: 57%, Points required for next Level: 174
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,362
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Appropiate test/s for bounded data sets?

    What about the beta distribution?

  3. The Following User Says Thank You to GretaGarbo For This Useful Post:

    phaynes (04-11-2014)

  4. #3
    TS Contributor
    Points: 5,246, Level: 46
    Level completed: 48%, Points required for next Level: 104
    maartenbuis's Avatar
    Location
    Konstanz
    Posts
    372
    Thanks
    3
    Thanked 146 Times in 123 Posts

    Re: Appropiate test/s for bounded data sets?

    Alternatively, if phaynes is only interested in differenced in conditional means, (s)he could look at the fractional logit which seems a bit more robust. A comparison of linear regression, various forms of beta regression and the fractional logit can be found here.

  5. The Following 2 Users Say Thank You to maartenbuis For This Useful Post:

    GretaGarbo (04-11-2014), phaynes (04-11-2014)

  6. #4
    Points: 32, Level: 1
    Level completed: 64%, Points required for next Level: 18

    Posts
    3
    Thanks
    2
    Thanked 0 Times in 0 Posts

    Re: Appropiate test/s for bounded data sets?


    Thanks, I think a beta distribution or fractional logit both seem good fits to my data. So, now I am wondering what the appropriate hypothesis tests would be for this data. mann-whitney seems like it could be appropriate, yes/no? But I would appreciate any other suggestions, perhaps a way to compare means?

+ Reply to Thread

           




Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats