+ Reply to Thread
Results 1 to 6 of 6

Thread: Pearson or Spearman?

  1. #1
    Points: 1,340, Level: 20
    Level completed: 40%, Points required for next Level: 60

    Posts
    19
    Thanks
    9
    Thanked 1 Time in 1 Post

    Pearson or Spearman?




    Hello all

    I know you're supposed to observe a scatterplot first in order to see if the shape is monotonic or linear to determine which test to use...but this graph has me baffled. Can anyone advise on which test to use based on this scatterplot? Thank you.
    Attached Images  

  2. #2
    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: Pearson or Spearman?

    I have always been taught that the key to decide if you use pearson or spearman's is not if the results are linear, but if the data is interval (pearson) or ordinal (spearman) - although you can if you want use spearman for interval data. It is true that pearson's r won't work with non-linear data (and yours looks non-linear to me), but I ran across disagreement on whether spearman's deals well with it or not. Some said that the data must be roughly linear to use spearman, others that it only had to have a monotonic not linear relationship.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  3. #3
    TS Contributor
    Points: 14,811, Level: 78
    Level completed: 91%, Points required for next Level: 39
    Miner's Avatar
    Location
    Greater Milwaukee area
    Posts
    1,171
    Thanks
    34
    Thanked 405 Times in 363 Posts

    Re: Pearson or Spearman?

    Wikipedia has some examples of a nonlinear, monotonic curve with a Spearmans correlation of 1.

  4. #4
    Points: 842, Level: 15
    Level completed: 43%, Points required for next Level: 58

    Posts
    9
    Thanks
    3
    Thanked 1 Time in 1 Post

    Re: Pearson or Spearman?

    That data does not look very linear to me :-). For this scatterplot, I'd expect a higher correlation using Spearman than when using Pearson, but I'd be a bit skeptical on how to interpret the results. It looks almost like two datasets to me: one up to 2000 on the y-axis, which is spread out over the x-axis, and one around 100% BMI.

    Please ignore if this is not helpful, but do you need to correlate these exact two quantities? Why do the BMI percentages heap up so strongly around 100% BMI?

  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: Pearson or Spearman?

    I think, in agreement I believe with the post immediately above me, that the data approaches some limit at which point the previous distribution becomes fundamentally different. I am not sure it even makes sense to analyze what is almost a constant at that point. You might comment out that section of the data and just analyze the rest of it.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  6. #6
    Points: 1,340, Level: 20
    Level completed: 40%, Points required for next Level: 60

    Posts
    19
    Thanks
    9
    Thanked 1 Time in 1 Post

    Re: Pearson or Spearman?


    Thanks for your responses guys, it was a big help.

+ 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