Evaluations should be independent in terms of being able to arrive at impartial conclusions. They must be seen as being credible if they are to fulfil this function. Their impartiality is derived from using appropriate scientific methodology and by being protected from political interference. Within the government, evaluation offices should be lodged in independent agencies located close to decision makers, but not in the same office as the programme being evaluated. Scientific methods that protect against various forms of bias should be derived with assistance from the professional evaluation community.