Roughly 600 informal sector entrepreneurs in Dar es Salaam, Tanzania, were surveyed in order to compare incomes of project beneficiaries with those of a suitable control group. The aim was to carry out a cost-benefit analysis and ultimately to gauge whether the project intervention could be justified in terms of increased incomes. However, the act of measurement significantly affected the results. It strongly mattered who asked the questions and how the questions were actually worded. Furthermore, project participation should not be considered a truly exogenous variable. In all probability it is systematically correlated with stated income and with the degree of truthfulness in the interviewee's response. In the light of these problems, a variation on willingness to pay is suggested as a more appropriate monitoring and evaluation indicator.