+ Reply to Thread
Results 1 to 5 of 5

Thread: Advise on choosing statistical method

  1. #1
    Points: 803, Level: 15
    Level completed: 3%, Points required for next Level: 97

    Posts
    4
    Thanks
    5
    Thanked 0 Times in 0 Posts

    Thumbs down Advise on choosing statistical method




    Hi. I'm a medicine student and I'm relatively new to statistics. I'm looking at some data and I can not figure out which statistical method I should use.

    The setup is the following:

    We are investigating disease X on Y number of patients.

    Doctor 'A' does a set of 8 clinical tests (I, II, III, IV, V, VI, VII, IIX) on each patient. Each test is either positive or negative (symptom or no symptom)

    Then doctor 'B' does a different clinical test (IX) that is either positive or negative.

    None of the above mentioned tests are 100% sensitive.

    I'm interested in knowing how strong the correlation is between doctor 'A' and doctor 'B's findings. Which method would you recommend using for the above mentioned setup? Any help would be much appreciated. Thank you

  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: Advise on choosing statistical method

    I'm interested in knowing how strong the correlation is between doctor 'A' and doctor 'B's findings.
    You could calculate 8 Phi-coefficients.

    With kind regards

    K.

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

    orangemasa (11-11-2013)

  4. #3
    Points: 4,939, Level: 44
    Level completed: 95%, Points required for next Level: 11
    Martin Marko's Avatar
    Posts
    53
    Thanks
    10
    Thanked 5 Times in 4 Posts

    Re: Advise on choosing statistical method

    Quote Originally Posted by orangemasa View Post
    I'm interested in knowing how strong the correlation is between doctor 'A' and doctor 'B's findings. Which method would you recommend using for the above mentioned setup?
    Apart from Phi, you can try polychoric correlations - in R, package polycor, call for polychor().

  5. The Following User Says Thank You to Martin Marko For This Useful Post:

    orangemasa (11-11-2013)

  6. #4
    Points: 803, Level: 15
    Level completed: 3%, Points required for next Level: 97

    Posts
    4
    Thanks
    5
    Thanked 0 Times in 0 Posts

    Re: Advise on choosing statistical method

    Thank you. How do I calculate the significance (i'm not even sure if this is the right term)? I have tried to find an example of how it is done (preferably a step-by-step guide) but I have not found one yet.

  7. #5
    Points: 501, Level: 10
    Level completed: 2%, Points required for next Level: 49

    Posts
    12
    Thanks
    1
    Thanked 3 Times in 3 Posts

    Re: Advise on choosing statistical method


    I am still a little puzzled. Are you looking at agreement between the doctors? I think not as A and B are not doing the same test. Then why the interest in A vs B; are you not looking for the best test to predict disease?
    The proposed methods above are indeed correct for the correlation between all 8 tests and test IX, but I am not sure that is what you are looking for.
    Best, Martin

+ 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