Impact assessments at the project level and other forms of poverty-oriented research have moved increasingly from single method to multi-method approaches (Herbert and Shepherd, 2001), and greater use of participatory approaches in impact assessment has expanded the toolbox (Hulme, 1997 in Herbert and Shepherd, 2001). Although sample surveys are still common, they are now often combined with participatory and other qualitative approaches, and qualitative methods (rapid appraisal, participant observation, PLA) are often used on their own, particularly for NGO implemented projects (Herbert and Shepherd, 2001). As each key method has its own strengths and weaknesses (see Table 2) they are increasingly selected for use together. As a result studies are now able to benefit from the advantages of sample surveys and statistical methods (quantification, representativeness and attribution) and the advantages of the qualitative and participatory approaches (ability to uncover approaches, capture the diversity of opinions and perceptions, unexpected impacts etc.) (Herbert and Shepherd, 2001). Which method(s) to choose will depend on the nature of the project, the type of information which is needed (or given priority), the context of the study and the availability of resources (time, money, human) (ibid.).
Triangulation is simply using different methods to research the same issue with the same unit of analysis (i.e. an in-depth unstructured interview with each member of a household on health care needs following a survey of household heads on the same topic), thus cross-checking one result against another, and increasing the reliability of the result. Contradictory results often bring up important problems with question design, as well as fundamental issues surrouding researcher understanding of a topic.