The practice of testing multiple combinations of variables with several statistical tests. Too many tests run without proper statsitical corrections can increase the chance of spurious findings, or those that are statistically significant just by chance alone.
