Spurious countermeasure effects and a reversal in the sign of the trend variable can be explained as artefacts of the mis-specification of the economic variable in models of Victorian crash data