Researchers often use theories as a tool to explain how different factors or variables affect each other. But sometimes, there isn't enough data to develop a theory from scratch. In these cases, researchers can use causal search algorithms to explore possible relationships between variables. These algorithms can generate graphs that show how different factors might be connected. Researchers can then use these graphs to develop hypotheses that can be tested with new data.
In this study, I compared two causal search algorithms: the PC algorithm and the LiNGAM algorithm. We ran a simulation study to see how well these algorithms performed on different types of data. We found that both algorithms performed well on most types of data. However, for small sample sizes, it was difficult to distinguish between the two algorithms.
I also applied these algorithms to two real-world datasets. In both cases, the algorithms were able to generate graphs that were consistent with the known causal relationships.
These results suggest that causal search algorithms can be a useful tool for researchers who want to develop theories about causal relationships. However, it is important to note that these algorithms are not perfect. They can sometimes generate incorrect graphs, so it is important to carefully evaluate the results.