Nonlinear Dynamics, Psychology, and Life Sciences, Vol. 18, Iss. 2, April, 2014, pp. 123-136
@2014 Society for Chaos Theory in Psychology & Life Sciences


Nonparametric Tests for Serial Dependence Based on Runs

Manuel Ruiz, Technical University of Cartagena, Spain
Usula Faura, University of Murcia, Spain
Matilde Lafuente, University of Murcia, Spain
Mohammed H. I. Dore, Brock University, Canada

Abstract: In this paper we propose several nonparametric independence tests for serial dependence based on runs. The tests can be applied to quantitative or categorical data. With a Monte Carlo experiment we show the size and power performance of the different statistics under linear and nonlinear data generating processes. We also compare the runs tests with well-known statistics TST(p) (Skaugh and Tjostheim, 1993), TDE(p) (Delgado, 1996) and G(m) (Matilla-Garcia and Ruiz, 2008) for serial dependence showing the runs tests perform well.

Keywords: serial dependence, nonparametric tests, runs test, symbolic dynamics