THE RELATIVE PERFORMANCE OF BIVARIATE CAUSALITY TESTS IN SMALL SAMPLES

Causality tests have been applied to establish directional effects and to reduce the set of potential predictors, For the latter type of app lication only bivariate tests can be used, In this study we compare bi variate causality tests. Although the problem addressed is general and could benefit researchers from different fields, most attention is gi ven to marketing applications, Even though there are many alternative tests, applications in marketing have almost exclusively been based on the I-laugh-Pierce test, We compare five bivariate tests in a specifi c marketing application, The empirical results indicate that conclusio ns about causality may depend strongly on the test used, To provide ge neralizable insights about the relative performances of alternative te sts we conduct a simulation study with data characteristics that cover the range of conditions encountered by researchers who have applied c ausality tests in marketing. We find that the Granger-Wald test has th e highest power but also the greatest upward bias in alpha (the probab ility of a type I error). If causality testing is done for the purpose of selecting a good subset of the available predictors, this combinat ion of high power and high alpha may be attractive. For researchers de siring a simple test with a substantial amount of power and little upw ard bias in alpha we recommend the Granger-Sargent teal. Interestingly , neither of these Granger tests has been used in marketing, (C) 1997 a Elsevier Science B.V.