Data Viz KitRights  |  Data  |  Questions  |  Charts  |  Hazards  |  Improvements

What are the correlations?

In seeking to determine a course of action or intervention that may yield a particular outcome, it may help to start by trying to identify correlations. In statistics, a correlation is a number that describes the degree of relationship between two variables.

A strong correlation means that a change in one data set closely tracks to a change in another data set. A weak correlation indicates that changes in the data sets are not very closely related. While correlation cannot, on its own, reveal causation, it can be useful in exploring possible variables for deeper causal analysis.

Scatter-plots are particularly effective for illustrating correlations. With one variable along the x-axis and another along the y-axis the existence and degree of correlation can be visualized, and the coherence of clusters and trends can be observed.

A challenge for human rights advocates is that correlation may be influenced by additional or "hidden" forces. See the page on Spurious Correlation in the section on visualization hazards.