Equally, the way in which political routines produce or reinforce bias, and most instances of how decision-making is affected by anticipated reactions, can be investigated using conventional research methods. More problematic is how to study power when it is exercised to shape people's preferences. This, the third dimension of power, is at once the most important and the most difficult aspect of power to research yet we would argue that despite the problems the effort is worth making, and we have pointed to some ways in which this might be done.