Retire Statistical Significance? Reevaluation of the Employment Effects of the 2006 World Cup

Abstract

Almost all econometric analyses of mega sporting events show nonsignificant results, i.e., they fail to reject the null hypothesis at the usual α levels of significance. Our contribution takes up the recent critique against the usual handling of the term of significance and qualifies the common interpretation that from a statistical point of view, the effects of mega sporting events are not significant. For this purpose, we use employment data from the World Cup 2006 in Germany. We determine empirical power functions of the relevant t-tests by means of a simulation study with autocorrelated and heteroscedastic error terms. In addition, we consider equivalence tests. Despite the relatively large goodness of fit of the model, it is not possible to reject the hypothesis of zero employment effects of the World Cup. At the same time, our simulation experiments show that even a fundamental raise/decline in unemployment in the World Cup venues of 6 to 7 percentage points would not be recognized in 50% of all cases