+ Reply to Thread
Results 1 to 7 of 7

Thread: clustered data

  1. #1
    Points: 8,343, Level: 61
    Level completed: 65%, Points required for next Level: 107

    Posts
    278
    Thanks
    14
    Thanked 7 Times in 7 Posts

    clustered data




    Hello all,

    I have a binary outcome (success \ failure). I have 20 subjects, half of them gave 2 samples, the others just one, so I have 30 data points.

    All data points without any exceptions were success (1). So my point estimate of the success rate is clearly 100%. Now I want to calculate a CI for the success rate, mainly to see what the lower limit is, I want to be able to say that with 95% confidence the success rate is higher than....(80%, 85%, whatever comes up).

    The problem is, as you can see, is the clustering, I can't use n = 30 because I then ignore the correlation, I feel it is a waste to use n = 20. Is there a way to calculate the variance of a single sample proportion while taking into account the cluster ?

    Thanks !

  2. #2
    Points: 616, Level: 12
    Level completed: 32%, Points required for next Level: 34

    Posts
    27
    Thanks
    0
    Thanked 3 Times in 3 Posts

    Re: clustered data

    I dont believe you can calculate fit statistics (confidence limits) when you have all your outcomes as succes. Anyways, have you considered using a generalized linear mixed model?

    Morten

  3. #3
    Points: 8,343, Level: 61
    Level completed: 65%, Points required for next Level: 107

    Posts
    278
    Thanks
    14
    Thanked 7 Times in 7 Posts

    Re: clustered data

    Yes, I thought about it, but SAS say that the outcome variable should be binary, not a constant ! :-(

    I can do CI when all my outcomes are success, using the common formulas, like Clopper-Pearson and others, I already did it. However, this way I use n = 20, while I have n = 30. And I find it to be a waste of information

  4. #4
    Points: 616, Level: 12
    Level completed: 32%, Points required for next Level: 34

    Posts
    27
    Thanks
    0
    Thanked 3 Times in 3 Posts

    Re: clustered data

    Hmm, I forgot about the exact methods but still, if you use any continous approximation I think you will end up with a variance estimate of 0 and hence cannot compute confidence limits.

    One (not very elegant) method could be to use GLMM with 1 failure and all 30 results (10 paired) and if the confidence limits are narrower than in the clopper-pearson with 20 succeses you can state that the confidence limit is narrower than the one found in the GLMM.

    I am sorry I can't be of any more help

    Morten

  5. #5
    Points: 8,343, Level: 61
    Level completed: 65%, Points required for next Level: 107

    Posts
    278
    Thanks
    14
    Thanked 7 Times in 7 Posts

    Re: clustered data

    On the contrary Morten, you are of great help !

    I found a reference in a book with an equation for this clustered problem, and guess what - you were right :-) I got variance of 0 because they used continuous approximation...

    I like your idea about GLMM, will give it a try, I have nothing to lose, if it doesn't work, I'll take the conservative way and use n=20

    thanks, your help was significant ! :-)

  6. #6
    Omega Contributor
    Points: 38,406, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    7,002
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: clustered data

    How did you determine success failure? Was it an interpretation?
    Stop cowardice, ban guns!

  7. #7
    Points: 8,343, Level: 61
    Level completed: 65%, Points required for next Level: 107

    Posts
    278
    Thanks
    14
    Thanked 7 Times in 7 Posts

    Re: clustered data


    No, not at all, it was scientific. The test was just too easy...

+ 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