For best experience please turn on javascript and use a modern browser!
You are using a browser that is no longer supported by Microsoft. Please upgrade your browser. The site may not present itself correctly if you continue browsing.

3.9 The search for causality: A comparison of different techniques for causal inference graphs

‪Jolanda J. Kossakowski‬, Lourens J. Waldorp and Han L. J. van der Maas

Keywords: Causal inference; Perturbation; Transitive reduction; Invariant causal prediction; Experimental design.

Authors on the article: 

Psychologists study the (possible) causal relations between psychological constructs, like sleep, concentration, and feelings of guilt. For example, does sleep deprivation lead to concentration problems? And could sleep deprivation be caused by increased feelings of guilt? Knowing what the cause is of something so intrusive as sleep problems may in turn lead to finding the solution to help an individual with sleep problems. If we know what causes a problem, we can help to solve it.

The type of data that is most often used to estimate causal relations between variables are observational data. These are (empirical) data in which no manipulations have taken place. Although one can use observational data to estimate some causal relations, this alone is not enough to properly estimate all relationships between variables. We also need so-called experimental data to estimate causal relations. These are (empirical) data where some perturbation or manipulation has taken place. The combination of observational and experimental data may give adequate information to properly estimate causal relations.

In our paper, we consider the conditions where estimating causal relations might work and we show how well different algorithms, namely the Peter and Clark algorithm, the Downward Ranking of Feed-Forward Loops algorithm, the Transitive Reduction for Weighted Signed Digraphs algorithm, the Invariant Causal Prediction (ICP) algorithm and the Hidden Invariant Causal Prediction (HICP) algorithm, determine causal relations in a simulation study. Results showed that the ICP and the HICP algorithms perform best in most simulation conditions. We also apply every algorithm to an empirical example to show the similarities and differences between the algorithms. We believe that the combination of the ICP and the HICP algorithm may be suitable to be used in future research.

Results that are shown in this paper may be of interest as it compares different techniques in both a simulation study and an empirical study. The paper serves as a reference for those who are interested in estimating causal relations between (psychological) variables because it gives a step-by-step tutorial on how to execute each algorithm and the reader can replicate this as all the data are online available. The algorithms that are discussed in this paper do not cater specifically to psychological data. Most, if not all, algorithms were developed in a different scientific field, which makes the results relevant for researchers from various scientific fields that are interested in estimating causal relations.

Kossakowski, J. J., Waldorp, L. J., & van der Maas, H. L. J. (2021). The search for causality: A comparison of different techniques for causal inference graphs. Psychological Methods, 26(6), 719 –742. https://doi.org/10.1037/met0000390