Evaluation of time-series data sets using the Pearson product-moment correlation coefficient.

The Pearson product-moment correlation has been used by researchers to compare time series data sets to assess the temporal similarities. Computer generated data, vertical ground reaction force (VGRF) data and hybrid data (constructed by combining features of computer generated and VGRF data) were used to investigate the influence of timing and amplitude differences on the effectiveness of this technique. Under a specific set of conditions the correlation coefficient is a valid and reliable indicator of temporal similarity. Deviations from these conditions, however, result in interactive effects between timing and amplitude components with subsequent reductions in the value of the coefficient. Although GRF data were evaluated, the results apply equally to other types of curves as well. The correlation coefficient is easy to use and can be used to evaluate the entire curve as opposed to discrete data points. Its usefulness is jeopardized, however, since it can be influenced by timing and amplitude differences as well as the characteristics of the curves being analyzed. A high coefficient is always indicative of temporal similarity but a lesser value does not guarantee a lack of temporal similarity.