Correlation vs Causation

Essential questions 

How to define correlation and causation?

What is the relationship between correlation and causation in experimental lab research? 

Controlled experimental study is used to test a hypothesis regarding the effect of treatment on experimental outcome, compared to the control group. This is often the most reliable way to obtain information about the causal relationship between the treatment and outcome. It is not simply that one causes another but the certainty that nothing else causes the effect that is ultimately important in experiments. This requires a researcher to maintain control over all known factors that can affect the outcome of the experiment.

Besides discovering relationship between cause and effect, a researcher is interested in learning how much cause is necessary to produce certain effect. That is how the independent variable affects the dependent variable. This may include testing the varying degrees of independent variable or removing it altogether.

Numeric raw data that are collected in laboratory experiment, are used to complete a statistical analysis of the relationships between the independent and dependent variables.

Correlation coefficient is usually calculated to indicate the strength of association between the two variables. The statistical analysis is completed to rule out random events that would contribute to the outcomes. Correlation coefficient of a zero indicates no association between the two sets of measured variables. Positive correlation coefficient indicates the increase of one experimental value that corresponds with the increase of another one. Negative correlation coefficient is the product of the decrease of one experimental value with the increase of another one. 

Correlation coefficient can only be used for the two sets of values that demonstrate linear relationship. Strong correlation between the two values may or may not be the indication of one be the cause of another. Spurious correlations can be mistaken for causations, and experimental design should be created in the way that minimizes the possibility of such outcomes. Spurious correlation between two experimental variables often is caused by the third parameter that is either unknown or not controlled at the time of experiment.

Causation establishes the cause-effect relationship between the independent and dependent variables and is not simply induced from the correlation between the two. It is not easy to prove with absolute certainty that the relationship between the two variables that are closely correlated is causation. Often it requires multiple experimental approaches before such conclusion can be drawn. For more details about correlation and causation in studies please watch the following:

https://www.youtube.com/watch?v=4EXNedimDMs

Even if the experimental research does not yield causation outcomes, correlation of experimental variables provides valuable scientific evidence. Once correlations are confirmed, every possible causative relationship must be systematically explored. However, by itself correlation cannot be used as evidence for a cause-and-effect relationship between the treatment and outcome. 

Suggested readings

Opgen-Rhein, R., Strimmer, K. (2007). From correlation to causation networks: a simple approximate learning algorithm and its application to high-dimensional plant gene expression data. BMC systems biology, 1, 37. doi:10.1186/1752-0509-1-37Ali Z., 

Geiger, S. M, Grossman, P., Schrader, U. (2019). Mindfulness and sustainability: correlation or causation? Current Opinion in Psychology, 28, 23-27. https://doi.org/10.1016/j.copsyc.2018.09.010.

Satake E. B. (2015) Statistical Methods and Reasoning for the Clinical Sciences Evidence-Based Practice. Ist ed. San Diego: Plural Publishing, Inc.