Toolbox: Overarching Issues

Avoiding bias

Bias can be introduced by the type of questions researchers ask and who they talk to, and when interviews or surveys are conducted. The way that questions are asked, the behaviour of interviewers and their gender or background (etc.) can influence responses. In addition, the way that data is analysed or presented can introduce bias (Gosling and Edwards, 1995:39-41). Ways to minimise bias include the careful training of researchers, setting of objectives and indicators, and the triangulation of information.

It is important to remember that impacts considered significant will differ by gender, class and other dimensions of social difference, in addition to being influenced by their role in the project or programme. Aggregating these views into an ‘objective truth’ may be impossible. In addition, where donors or other ‘outsiders’ specify the selection of impact assessment indicators or the use of specific methods of data collection or analysis important unanticipated benefits or changes may be missed, or the intervention may be identified as having a narrower impact than it did in reality.