Testing the ‘standard’ model of stochastic choice under risk