Most current efforts in IA are model building exercises. However, IA is seen as more than just a model building exercise. It is also seen as a methodology that can be used for gaining insight over an array of environmental problems spanning a wide variety of spatial and temporal scales. For example, integrated assessments of acid rain are concerned with smaller geographical scales and shorter time horizons than integrated assessments of climate change. Integrated assessments are also being carried out for even smaller geographical units such as river basins and watersheds.
IA is also seen as a process, where a model provides an organizing framework for conducting research. According to this view, the IA model plays the role of ensuring consistency and points to substantive areas where more information is required, i.e., researchers use the model as a guiding tool for research prioritization. The actual research program emerges iteratively from the insights that the model provides and from investigations in the substantive domains of the sub-components. There is currently a trend to move away from defining IAs purely in terms of models towards the methodological and process definitions.
The results from IA models are used by practitioners in a number of different ways. For the sake of discussion it is useful to classify the approaches into two different categories. The first approach is the ``truth machine'' approach, which treats model results of forecasted global outcomes with a high degree of confidence and makes strong claims regarding the nature of climate change impacts and the relative benefits of different policy options. The second approach views the IA model as a heuristic tool and is more modest; here less confidence is placed on the actual model results. Instead, quantitative model outputs are used to make qualitative judgments about impacts (e.g., large, small), abatement (e.g., high, low) and dynamics (e.g., now, later), and about the interplay of processes in the model. These qualitative judgments have been termed ``insights''. Embedded in the notion of an insight is the idea that qualitative judgments are more robust than quantitative forecasts. While this may be true in many cases, it is necessary to point out that qualitative judgments are inextricably tied to quantitative model outputs and hence to underlying model assumptions, and may have similar shortcomings. In other words, use of the model as a heuristic tool relies on having some level of confidence in the model as a truth machine because qualitative judgments must always be inferred from quantitative model output. Thus the distinction between IA models as truth machines and heuristic tools is not a clean one. This blurring of roles is also brought into focus when questions regarding the validation of IA models are examined in section 5. Additionally, some qualitative judgments by their very nature rely more heavily on model forecasts than others, making some insights less robust than others.