+ Reply to Thread
Results 1 to 5 of 5

Thread: considering covariates in a non parametric test

  1. #1
    Points: 1,885, Level: 25
    Level completed: 85%, Points required for next Level: 15

    Posts
    56
    Thanks
    25
    Thanked 0 Times in 0 Posts

    considering covariates in a non parametric test




    Hello, I have the following problem:
    I have to analyze two variables, both continuous (they are results of body tests).
    Since I have cases only and I tested the variables to check the distribution (which was not normal) I ran a spearman correlation. Actually this was the approach chosen even by former studies studying similar things.

    my tutor asked me to adjust this for covariates like sex, disease duration and age...

    But this changes the approach completely. I don't know any non parametric test that allows to do that for continuous variables. Any suggestion?

  2. #2
    TS Contributor
    Points: 17,779, Level: 84
    Level completed: 86%, Points required for next Level: 71
    Karabiner's Avatar
    Location
    FC Schalke 04, Germany
    Posts
    2,542
    Thanks
    56
    Thanked 640 Times in 602 Posts

    Re: considering covariates in a non parametric test

    Run a multiple linear regression and check the distribution of
    the residuals. The dependent variable doesn't need to be
    normally distributed, only the residuals. And if the sample size
    is large enough, then departures form normality usually don't
    matter much. Maybe you want to make a little search for linear
    regression assumptions.

    With kind regards

    K.

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

    Rerun (04-12-2014)

  4. #3
    Points: 1,885, Level: 25
    Level completed: 85%, Points required for next Level: 15

    Posts
    56
    Thanks
    25
    Thanked 0 Times in 0 Posts

    Re: considering covariates in a non parametric test

    I checked the distribution of the independent variable using the Shapiro-wilk/francia and sktest and it is not normally distributed!
    That is why I did not go ahead with multiple linear regression. Is my assumption wrong?

    thank you very much

  5. #4
    TS Contributor
    Points: 17,779, Level: 84
    Level completed: 86%, Points required for next Level: 71
    Karabiner's Avatar
    Location
    FC Schalke 04, Germany
    Posts
    2,542
    Thanks
    56
    Thanked 640 Times in 602 Posts

    Re: considering covariates in a non parametric test

    I checked the distribution of the independent variable using the Shapiro-wilk/francia and sktest and it is not normally distributed!
    There's no need for an independent variable in regression to be
    normally distributed, only the model residuals. And if the sample size
    is large enough, then departures form normality usually don't
    matter much. Maybe you want to make a little search for linear
    regression assumptions. http://pareonline.net/pdf/v18n11.pdf
    for example.

    With kind regards

    K.

  6. The Following User Says Thank You to Karabiner For This Useful Post:

    Rerun (04-12-2014)

  7. #5
    Super Moderator
    Points: 13,151, Level: 74
    Level completed: 76%, Points required for next Level: 99
    Dragan's Avatar
    Location
    Illinois, US
    Posts
    2,014
    Thanks
    0
    Thanked 223 Times in 192 Posts

    Re: considering covariates in a non parametric test


    Another viable alternative you could consider is a Bootstrap Regression technique. Of course, you would need the appropriate software (S-Plus, R) to perform this technique.

    http://cran.r-project.org/doc/contri...tstrapping.pdf

+ Reply to 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