The analytical formulas describing the behaviour of dielectric elastomer actuators (DEAs) are based on hyperelastic strain energy density functions. The analytical modelling of a DEA will only lead to meaningful results if the dielectric elastomer can be accurately represented by the chosen hyperelastic model and if its parameters are carefully matched to the elastomer. In the case of silicone elastomers, we show that the strain energy density of a thin elastomeric membrane depends on the maximum deformation the membrane was previously submitted to (Mullins effect). We also show that using model parameters coming from an uniaxial pull-test to predict the behaviour of the elastomer in an equi-biaxial configuration leads to erroneous results. We have therefore built a measurement setup, which allows testing thin elastomeric membranes under equi-biaxial stress by inflating them with a pressure source. When modelling a DEA under equi-biaxial stretch, the measurement data can be used directly, without the need of an hyperelastic model, leading to voltage-stretch prediction closer the the measured stress-stretch behaviour of the dielectric membrane.