Statistics – Spearman correlation
Hi, Pearson or Spearman’s correlation?
The measurement level or normal distribution will lead you to the most correct correlation method. So if you have 2 scale variable that are both normal distributed, then you can have a look on the Pearson’s correlation. But otherwise you look at Spearman’s correlation (both ordinal and scale variables are allowed to use here).
Command: Analyze – Correlation – Bivariate
In this example I chosed both to show you the different results.
Result: Here it’s better to go for the Spearman’s correlation, because 2 of the 3 variable are not normal distributed. Both salary and worktime are very skewed distributed.
The Spearman’s correlation rank all the data, which means that even if a salary is extremely high it will not effect the correlation because the method use ranks.
As you can see the correlation coefficients are higher in Spearman’s correlation matrix compare to Pearson’s correlation matrix. Normally if the variables are normal distributed it use to be the opposite situation that Pearson is higher.