Relationship Between Random Correlation Coefficients and Statistical Significance Under a True Null Hypothesis

• In Event: Statistical Grab Bag

Abstract

Big data sets in the 21st century have thousands of variables and millions of observations and statistical analyses are done with computer intensive algorithms. Nonetheless, classical statistical significance is still relevant when working with small sample sizes. R.A. Fisher stated: “it is with small samples, less than 100, that the practical research worker ordinarily wishes to use the correlation coefficient.” This paper demonstrates that statistical significance and substantive significance are both important for correlational analysis with small sample sizes. Without statistical significance, there will be many more unreliable or un-replicable Pearson correlation coefficients appearing in the research literature.

Author

• Eugene Komaroff, Keiser University
  Presenting Author