4 Triangulation
The concept of triangulation suggests that this broadness is a strength; that greater confidence can be placed in research findings based on the points of agreement between the results of diverse methods. Triangulation involves applying a range of different perspectives on an issue in order to reduce uncertainty about it. The greater the differences between the perspectives, the more the confidence that can be placed in any agreement between their findings in a particular case. Triangulation therefore relies on an understanding of the sources of bias embodied in the different methods – their strengths and weaknesses as vehicles for engaging with the research context.
Triangulation is originally a term from navigation or cartography. A navigator can pinpoint their position by taking bearings on three features whose identity and position are known. Theoretically the lines should intersect one another at a single point, but in practice there are almost always errors in measurement. Plotting lines on a map on these bearings from the location of the three features will produce a triangle. The interior of the triangle is the best available approximation for the navigator’s position. The reason that three bearings are taken, rather than two is because the third line gives an idea of the magnitude and direction of the errors introduced by the other two.
The metaphor carries over to research, where the term triangulation is used in two different senses. In survey and interview methods, triangulation refers to asking the same question in different ways, or cross-checking responses against one another. In semi-structured interviewing, for example, it can be helpful to point out any perceived discrepancies between replies to different questions, and ask the interviewee to explain them or to clarify any misunderstandings. The other sense refers to the use of multiple perspectives in order to converge on an understanding of a particular issue or research question. The different perspective may be embodied in different methods, data, theories, disciplines and/or researchers. Multidisciplinary research can involve some or all of these elements.
In both cases, information from one perspective is compared to one or more other perspectives. The comparison helps to reduce the uncertainty inherent in taking a particular perspective (whether that is asking a question in a particular way, employing a particular methodology or adopting a particular theoretical standpoint), in as much as the different views agree. It should be noted that even though navigational triangulation generally relies on taking three bearings, it is not necessary to use three questions or three different data types etc. The main idea is to compare between as many divergent perspectives as is reasonable, whether that is two or twelve.
In the next reading, Roe discusses the use of triangulation in policy research on sustainable livelihoods. Sustainable livelihoods for the world’s population are arguably necessary in order to reconcile conservation of the environment with social and economic development giving rise to sustainable development. The example he gives of a triangulation is one based on different theoretical perspectives of sustainable development, and the interconnections between them. Thus, he adopts an approach which draws on the perspectives and methods of more than one traditional academic discipline, by taking an interdisciplinary or multidisciplinary approachFootnote 1. But much of what he says can be applied to triangulation on different methods too as discussed in unit 4.
Activity 35
Read pp. 35-37 (starting ‘Complex policy analaysis’ up to ‘Starting the complex policy analysis’) of Roe (1999). Why does Roe argue that elements of a triangulation be as different as possible?
Comment
Roe argues that the greater the divergence between the viewpoints upon which a triangulation is based, the greater the confidence that can be placed in its results. Divergent perspectives are ones which introduce opposing biases, and these will tend to cancel one another out. Note that he does not claim that triangulation brings a researcher nearer to the truth. Convergence between different viewpoints is merely taken as a good starting point for pursuing an issue further. In Roe’s terms, bias has been reduced (but not eliminated).
3 Research design