0

I would like to investigate the correlation of visually and automated assesed amount of emphysema (%) in CT-scans.

visual: categorical data (visual score 1-4, score 1= 1-25%, score 2= 26-50%, score 3=51-75%, score 4= 76-100%)

automated: interval scaled data (1-100%)

Since the data is not parametric, I thought about to use a spearman correlation test.

Is this correct? Is there any other test that would fit better?

captcoma
  • 155
  • 13
  • What do you mean with “the data is not parametric”? – aivanov Mar 09 '18 at 00:05
  • Visual appears to be an ordinal variable. It's probably not best to use it as a categorical variable. – Sal Mangiafico Mar 09 '18 at 01:26
  • The data is not normally distributed. And yes, visual is an ordinal variable. – captcoma Mar 09 '18 at 06:37
  • 1
    You might look into polyserial correlation. "[polyserial correlation... between a quantitative variable and an ordinal variable, based on the assumption that the joint distribution of the quantitative variable and a latent continuous variable underlying the ordinal variable is bivariate normal.](https://www.rdocumentation.org/packages/polycor/versions/0.7-9/topics/polyserial) – Sal Mangiafico Mar 09 '18 at 13:43
  • Also, consider Kendall correlation. – Sal Mangiafico Mar 09 '18 at 13:45
  • polyserial correlation works very well. this is exactly what I was looking for. thanks – captcoma Mar 09 '18 at 22:09
  • 1
    ***data*** are never "parametric" or "nonparametric". ***analyses*** are. – Alexis Sep 26 '18 at 16:18

0 Answers0