Abstract: Gasoline price rises in China show synchronization with the occurrence times of global M7.0+ earthquakes in recent years, attracting attentions from earthquake scientists and related personnel. Through statistical analysis of these two event sequences during 2003—2015, which are physically irrelevant, we found a significant correlation between them, which seems to be contrary to our common sense. The reason of this paradox is that these two sequences are pre-selected, if the population is sufficient, we can always find some samples highly-correlated with each other, even if these samples have no causal relation at all. The results imply that, some anomalies, which can pass statistical tests and are found retrospectively useful for earthquake prediction, may be not resulted from earthquakes, since they are pre-selected. In practical, such correlation between irrelevant processes can be easily found in the current big-data era, therefore, it is essential to investigate the physical mechanism behind and perform perspective forecasting tests in addition to doing correlation analysis.

Key words: Global earthquakes, Gasoline price rises in China, Statistical correlation, Empirical prediction