What type of correlation coefficient should I be using?

Vicki Lawrence, MS, PHD

Statistics Consultant

o   Pearson correlation coefficients are used to determine the linear correlation between two continuous variables.  The Pearson correlation coefficient can range from -1 to 1, with negative values indicative of an inverse relationship between the two variables, and positive values indicative of a positive relationship between the two variables.  The square of the Pearson correlation coefficient is called the proportion of variation, and represents the proportion of variation in one variable that is attributable to variation in the second variable.

o   What happens, though when we do not expect linear relationships, or we do not have continuous (interval/ratio) variables?  The interpretation of the Spearman coefficient is only with regard as to whether a relationship exists, but does not indicate whether the association is linear.  The Spearman coefficient has a range of -1 to 1, with negative values indicative of an inverse relationship between the two variables, and positive values indicative of a positive relationship between the two variables.

DR. VICKI LAWRENCE is an academic researcher with a background in epidemiology and biostatistics.  She studies social correlates of mental and physical chronic health inequities, and has provided statistical support on multiple studies with various investigators.  She has worked with clinicians, research investigators, and tutored multiple graduate students in public health, epidemiology, social work, medicine, education, and nursing.

Network Coordinator

Network Coordinator at Book Editing Associates
Formed in 1998, Book Editing Associates is a one-stop shop for writers who need professional book editors, tested proofreaders, published ghostwriters, and publishing consultants.

Network Coordinator

Formed in 1998, Book Editing Associates is a one-stop shop for writers who need professional book editors, tested proofreaders, published ghostwriters, and publishing consultants.