+ Reply to Thread
Results 1 to 7 of 7

Thread: How to rank long lists in Spearman's Rank Correlation?

  1. #1
    Points: 1,408, Level: 21
    Level completed: 8%, Points required for next Level: 92

    Posts
    26
    Thanks
    9
    Thanked 0 Times in 0 Posts

    How to rank long lists in Spearman's Rank Correlation?




    Hi there,

    I have 2 lists of 1000 items and ranked them based on criteria A or B. When I want to compare the Spearman correlation at N=20 , should I use the rank values when all items exist or when only the 20 items exist?

    For example using the 1000 items, I would get a ranking such as
    Criteria A, Criteria B
    1, 3
    2, 200,
    3, 202
    ..
    100, 333

    When I want to calculate pearman's Rank Correlation, should I keep the 200 and 202 or re-rank the list so no value is higher than 20? Re-ranking the lists when N=20 , means that I dropped items that were higher in the list, so is that correct?

    Many thanks !
    Last edited by tnaser; 03-15-2013 at 12:13 PM.

  2. #2
    Omega Contributor
    Points: 38,334, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,998
    Thanks
    398
    Thanked 1,185 Times in 1,146 Posts

    Re: How to rank long lists in Spearman's Rank Correlation?

    Lost me! No idea what you are writing about. Perhaps an example may help convey your question.
    Stop cowardice, ban guns!

  3. #3
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: How to rank long lists in Spearman's Rank Correlation?

    I dont understand how you compare list of 1000 and a subset of 20 regardless of which correlation you use.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  4. #4
    Points: 1,408, Level: 21
    Level completed: 8%, Points required for next Level: 92

    Posts
    26
    Thanks
    9
    Thanked 0 Times in 0 Posts

    Re: How to rank long lists in Spearman's Rank Correlation?

    Sorry for the confusion , I will try to explain more. For example lets's say we are ranking universities based on 2 criterias A and B. And we want to see if there is correlation between the 2 ranking. Assume we have 1000 university.

    1- Rank using Criteria A ( let's call it list A)
    University - Rank
    University A - 1
    University B - 2
    University C - 3
    ...
    University ZZ - 1000

    2- Rank using Criteria B could be like ( let's call it list B)
    University - Rank
    University A - 1
    University B - 2
    University ZZ - 3
    ...
    University C - 1000

    3- When ranking both lists , C will get rank of 1000.
    University - Rank A - Rank B
    University A - 1 - 1
    University B - 2 - 2
    University C - 3 - 1000
    ...
    University ZZ - 1000 - 3

    4- Now the question, if I want to compare the top 20 universities, what value should University C get using criteria B ? would it get 1000 since there are better universities above it in the initial rank , or I just ignore that fact and rank the list of the 20 universities that existed in list A using criteria B ?

    If I rank the 1000 universities and then select the top 20 universities from both sets, some universities may exist in list A but not list B and vise versa.


    Hope this clarify my question and many thanks for your help
    Last edited by tnaser; 03-16-2013 at 10:52 AM.

  5. #5
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: How to rank long lists in Spearman's Rank Correlation?

    How are you determining the rank for C? Adding the rank of A + B, adding a weighted A + B, multipling times them? Either way you are no using Spearman Rho. You are just finding some way to judge the best universities using some criteria from two different dimensions (A and B).

    If you want to use Spearman rho to compare A & B you would just run a correlation between them. Spearman rho would throw out (I believe) pairs where a value was missing. So if A or B did not assign a value to say the 997 school it would not be used in the spearman rho.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  6. #6
    Points: 1,408, Level: 21
    Level completed: 8%, Points required for next Level: 92

    Posts
    26
    Thanks
    9
    Thanked 0 Times in 0 Posts

    Re: How to rank long lists in Spearman's Rank Correlation?

    There is no rank C , only University C , the 3rd list is just to compare criteria A and B . University C got rank 3 in list A and rank 1000 in list B. So the question was when I want to compare the top 20 universities, should I give University C
    1- rank 1000 - OR
    2- NA , and then just remove it since it is not in the top 20 - OR
    3- remove the universities above university C in the 1000 rank and assume that University C got rank 20

    Thanks

  7. #7
    Points: 1,408, Level: 21
    Level completed: 8%, Points required for next Level: 92

    Posts
    26
    Thanks
    9
    Thanked 0 Times in 0 Posts

    Re: How to rank long lists in Spearman's Rank Correlation?


    Any advice?

+ 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