+ Reply to Thread
Results 1 to 6 of 6

Thread: Alternatives to ANCOVA? Unequal slopes.

  1. #1
    Points: 388, Level: 7
    Level completed: 76%, Points required for next Level: 12

    Posts
    1
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Alternatives to ANCOVA? Unequal slopes.




    Hello, and thank you for your help on my question.

    My goal is to compare multiple groups to determine if a certain "variable of interest" is different between different groups (in the example below: blue, red, and orange). Normally, this statistical test could be solved with an ANOVA, but the "variable of interest" is influenced by a co-variate (in the example below this is speed). In this case, an ANCOVA would be the appropriate test, but the data I am working with has two groups that are influenced by speed (blue and red) and one variable that is not influenced by speed (orange). Obviously this violates the assumptions of an ANCOVA, but I am unclear as what to do in order to compare the groups to each other.

    I appreciate any advise on this issue. Thank you so much.

    -Michael

    Name:  Example.jpg
Views: 147
Size:  67.5 KB

  2. #2
    Devorador de queso
    Points: 95,940, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Posting AwardCommunity AwardDiscussion EnderFrequent Poster
    Dason's Avatar
    Location
    Tampa, FL
    Posts
    12,937
    Thanks
    307
    Thanked 2,630 Times in 2,246 Posts

    Re: Alternatives to ANCOVA? Unequal slopes.

    You can always fit a linear model with terms for group, velocity, and the group*velocity interaction. This basically allows all the groups to have their own line (with different slopes) but still makes an equal variance assumption.
    I don't have emotions and sometimes that makes me very sad.

  3. #3
    Super Moderator
    Points: 31,766, Level: 100
    Level completed: 0%, Points required for next Level: 0
    bugman's Avatar
    Posts
    2,255
    Thanks
    290
    Thanked 324 Times in 265 Posts

    Re: Alternatives to ANCOVA? Unequal slopes.

    Quote Originally Posted by Dason View Post
    ... but still makes an equal variance assumption.
    But this could be overcome using gls couldn't it?
    The earth is round: P<0.05

  4. #4
    Points: 2,462, Level: 30
    Level completed: 8%, Points required for next Level: 138

    Posts
    200
    Thanks
    20
    Thanked 48 Times in 43 Posts

    Re: Alternatives to ANCOVA? Unequal slopes.

    Exactly, e.g. using the nlme package in R you can nicely combine various random structures (e.g. random intercept and slope models) with various variance structures in the context of GLS, for example coupling the variance to a continuous predictor or allowing the variance to vary between levels of a categorical predictior.

  5. #5
    Devorador de queso
    Points: 95,940, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Posting AwardCommunity AwardDiscussion EnderFrequent Poster
    Dason's Avatar
    Location
    Tampa, FL
    Posts
    12,937
    Thanks
    307
    Thanked 2,630 Times in 2,246 Posts

    Re: Alternatives to ANCOVA? Unequal slopes.

    Of course there are ways to overcome this. I don't see it as a "problem" that needs to be overcome unless there is evidence against the idea of the variances being equal. But I did want to make sure that it was clear that by using a linear model in this way there still is a default assumption of equal variance.
    I don't have emotions and sometimes that makes me very sad.

  6. #6
    Super Moderator
    Points: 31,766, Level: 100
    Level completed: 0%, Points required for next Level: 0
    bugman's Avatar
    Posts
    2,255
    Thanks
    290
    Thanked 324 Times in 265 Posts

    Re: Alternatives to ANCOVA? Unequal slopes.


    Yep, got that Dason. I wanted clarification for my own selfish needs.
    The earth is round: P<0.05

+ 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