Data Analysis > Correlation and Regression > Correlation for ordinal variables

Spearman’s Rank Correlation and its context

Spearman correlation coefficient is applied where one or both the variable are qualitative in nature.

 

Where D_i is the difference of ranks of ith data point.

The t-test formula to test the hypothesis,  no correlation,

 

cor(mtcars$mpg, mtcars$hp, method = "spearman")  # Ranked correlation

cor.test(mtcars$mpg, mtcars$hp, method = "spearman")

 

Kendall’s Tao

We use Kandall’s Tao in the following situation:

• When we have both variables are on an Ordinal scale,
• When data don’t follow a normal distribution
• Small samples or data with many tied ranks
• We have suspect a consistent increasing or decreasing relationship, but not necessarily linear

 

cor(mtcars$gear, mtcars$carb, method = "kendall")

cor.test(mtcars$gear, mtcars$carb, method = "kendall")

 

Prev

Linear Regression

Next

Chi-Square Test of Independence

Feedback

Submit