Assessment of the Significance of Sample Serial Correlation by the Bootstrap Test

This article proposes to use the computer-based bootstraptechnique to test the significance of the lag-k sampleserial correlation coefficient (r <Subscript>k</Subscript>) of a time series.The bootstrap approach is free of sample distribution types, freeof restrictive assumptions, and easy to understand and implement. Monte Carlo simulation indicates that the bootstrap test has thesame power as the conventional parametric test for assessing thesignificance of a lag-one serial correlation coefficient (r <Subscript>1</Subscript>). The usefulness of the bootstrap test is demonstrated by applying it to assess the significance of r <Subscript>k</Subscript> of the annual mean flow data of twenty pristine river basins in Ontario, Canada. Copyright Kluwer Academic Publishers 2002