Answers are grounded in this paper and cite their sources.
Regression analysis and the use of correlation coefficients have proven to be useful tools in the analysis of experimental data. But unless the proper precautions are taken, erroneous conclusions can easily be reached by their use. Several data patterns are provided to demonstrate this. Examples yielding unrepresentatively high, as well as low, correlations are discussed.
No takes yet. Share an insight, caveat, or question.
Marcus et al. (1976) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: