Which test of association can I use?

#1
Which test of association can I use if I want to know if there is a significant association between a continuous independent variable (age or BMI) and a dichotomous dependent variable (bleeding after surgery: yes/no)?

Also, which test of association can I use if I want to know if there is a significant association between a continuous independent variable (age or BMI) and a continuous dependent variable (length of hospitalization stay)?

Thanks in advance.
 

staassis

Active Member
#2
Which test of association can I use if I want to know if there is a significant association between a continuous independent variable (age or BMI) and a dichotomous dependent variable (bleeding after surgery: yes/no)?
The default choice is Mann-Whitney test. If the sample size is substantial in each of the two groups ("yes" and "no"), you can also use the independent samples t-test. The two tests are likely to agree.

Also, which test of association can I use if I want to know if there is a significant association between a continuous independent variable (age or BMI) and a continuous dependent variable (length of hospitalization stay)?
If the relationship between the two continuous variables is linear you can use the t-test for Pearson's correlation. If the relationship is nonlinear but monotonic, you can use asymptotic tests for Spearman's ρ and Kendall's τ.
 
Last edited: