Correlation: Not all correlation entails causality

Abstract

The concept of correlation entails having a couple of observations (X and Y), that is to say, the value that Y acquires for a determined value of X; the correlation makes it possible to examine the trend of two variables to be grouped together. We know that, with increasing age, blood pressure figures also increase, therefore, if we want to answer a research question like “what is the connection between age and blood pressure?” the relevant statistical test is a correlation test. This test makes it possible to quantify the magnitude of the correlation between two variables, but it is also helpful for predicting values. If these variables had a perfect correlation, the value of the variable Y could be deduced by knowing the value of X. Because of these advantages, the correlation is one of the most frequently used tests in the clinical setting since, in addition to measuring the direction and magnitude of the association of two variables, it is one of the foundations for prediction models, such as linear regression model, logistic regression model and Cox proportional hazards model.