Triangulation

What: draw on multiple lines of evidence to justify your conclusion.

Use when: individual sources are considered useful but be thought to have some weaknesses. Multiple different sources are available that can cover each others' weaknesses. This is a common approach in social science (see Wikipedia article on Triangulation)

Why: using multiple independent lines of evidence is intended to compensate for the individual biases of the separate sources, and therefore increase confidence in the conclusion. It is natural to frame the conclusion to emphasise this, if this method was used in the analysis. Note that this differs from Demonstrate-robustness, which tests the effect of different assumptions rather than drawing on different evidence. By relying on evidence, rather than assumptions, a triangulated claim is in a sense stronger than a robust one. However, in practice, triangulation may be used even to support uncertain conclusions (e.g. qualified as Degree-of-belief), such that the relative value of the two frames may vary.

How to implement: use multiple lines of evidence in the analysis, and refer to each of them in writing about it. There are various ways of structuring the presentation of the evidence. One line of evidence might be used as the focus, with others introduced in a supporting capacity. Alternatively, the range of sources of evidence can be introduced upfront, and then expanded on in sequence.

How to recognise:

Multiple warrants and grounds all supporting the same conclusion. In principle, they would be independent, but it is sufficient that the supporting arguments differ in any way, e.g. different data or different methods.

• Triangulation is distinguished from New-view-on-existing-result in that the former uses multiple lines of evidence to support a new result, whereas the latter uses multiple lines of evidence (existing and new results) to support an existing result.

• Use of triangulation might also indicate the conclusion is robust. One distinction from Demonstrate-robustness is whether the conclusion is supported by multiple lines of evidence or multiple assumptions. Robustness is however considered a stronger statement. Triangulation indicates that the conclusion is supported more than once, robustness indicates that there is no counter-argument to the conclusion.

Examples of Triangulation, showing multiple lines of evidence in italics for primary claim in bold, ordered by decreasing clarity:

• Similarity of this result and an existing one

  • "The simulations suggest that ... a poorly cemented borehole (k(c) 10 mD) could yield methane leakage rates at the base of an aquifer ranging from 0.04 m(3)/d to more than 100 m(3)/d ... These values are compatible with surface casing vent flows reported for shale gas wells in the St. Lawrence Lowlands (Quebec, Canada)" 2014WR016146. The lines of evidence are the simulations and reported values. (Also Uncertainty-quantification because of use of a range of leakage rates)

• Using data and theory

  • "Simulations show that ... once tidal oscillations are included, as the degree of heterogeneity increases, the combined effect of heterogeneity and tidal oscillations on mixing and spreading of the interface reduces. This is explained by the fact that an increase in the log-permeability variance induces an increase in both the effective permeability and the spatial connectivity, which implies a more uniform hydraulic response to tidal forcing and, as a result, the degree of mixing decreases." 2014WR016068. The lines of evidence are the simulations, and a description of the physical basis by which the result would be obtained.

• Multiple methods

  • "A calculation of the travel times through a channel-island complex shows travel times through the islands to be at least 3 times their channel counterparts. A dye release experiment also indicates that travel times in islands are much longer that those within channels as dye remained in the island for the 3.8 day duration of the experiment." 2014WR016149. The lines of evidence are calculation of travel times and a dye release experiment. Note that the first sentence is also a separate claim on its own, but its classification is not of interest for this example.

  • "Our results showed that DOC concentration significantly decreased (p< 0.001) while TDN concentration significantly increased (p< 0.001) in runoff water from pioneer to mature forests, which in turn resulted in a decrease in DOC flux and an increase in TDN flux, ... An increase in TDN flux from pioneer to mature forests was consistent with the previous finding that mature forest was nitrogen saturated while pioneer forest was nitrogen limited." 2015WR016912. The lines of evidence are statistical analysis of concentration between pioneer and mature forests, and an existing result about level of nitrogen in the two forest types. Both lines of evidence include other claims not discussed here.

• More complex cases

  • "In this paper, relatively simple, but science-based, methods are presented that allow researchers, engineers, and water managers to obtain first-order estimates of essential process parameters in estuaries ... These apparently simple relationships are assumed to result from the capacity of freely erodible water bodies to adjust themselves to external drivers and to dissipate the free energy from these drivers as efficiently as possible. Thus, it is assumed that these systems operate close to their thermodynamic limit, resulting in predictable patterns that can be described by relatively simple equations. Although still much has to be done to develop an overall physics-based theory, this does not prevent us from making use of the empirical laws that we observe in alluvial estuaries. 2015WR016936. The triangulated claim is that the methods are useful, and the multiple lines of evidence are: empirical data illustrating the relationships, a hypothesised physical justification for why these relationships might be occurring, and a pragmatic argument that there is no physical basis for rejecting this simple approach. This is a complex example with several frames, as well as multiple claims, which are not discussed here.