The knowledge used to deconstruct a problem comes from a number of sources; and it is important to recognize that, although the product of the analysis is a heuristic, it is still informed by empirical sources. Sources for this knowledge are captured with the acronym LODES: Literature, Observation, Dialogue, Experience, and Survey. Each of these sources can be used to amass information for deconstructing a problem (see Figure 2).
Of these sources, deconstruction naturally begins endogenously, with the researcher’s personal experience and existing schema about the problem \citep{deGrave}. This initial understanding will inevitably affect choices about further sources. To this, the remaining exogenous sources such as literature, observation, dialogue, and survey may be added. It should be noted that, of the exogenous sources, literature will likely not provide contextually situated information, but will provide broader perspectives into the problem; whereas observation, dialogue, and survey will provide contextually-grounded details but may lack a broader perspective on the problem. Given the relative strengths and weaknesses of each source, a combination of literature review and in situ data collection is recommended.
Throughout this phase, the researcher should keep two considerations in mind. First, this is not an exhaustive study of the problem, it is merely a pause to gain further clarity before committing to a research initiative. Concise and targeted investigation of the problem using a few of the LODES sources should prove sufficient. Second, it is important to remember that this phase of the analysis is highly iterative as the researcher moves between collecting new information, synthesis of that information, and updating schema. Through this iterative process, new knowledge intermingles with existing schema and opaque layers of the problem begin to give way to clearer understanding \citep*{Clarke}.

Prioritize the Drivers

Once a problem has been deconstructed, the identified drivers should be prioritized. The purpose of analyzing a problem is to identify its root causes so the problem can be more effectively addressed; and effectively addressing a problem involves, in part, making informed decisions about where to focus efforts to achieve maximum effect with minimum expenditure of resources. Prioritization helps the researcher make these decisions by scoring each of the identified drivers on three criteria: Impact, Influence, and Time.
Impact (IM) estimates the potential effect that a driver has on the overall problem. Although all drivers are expected to have some impact on the problem, some will have more effect than others. Influence (IN) designates the perceived amount of control the researcher has over a given driver. Action research is conducted to affect some desired change; for this to happen, the driver must be tractable within the influence of the researcher. Time (T) estimates how long it will take to address the driver. Drivers contain varying degrees of inertia; and some simply take more time to address. If the goal of efficiency in research is accepted, then the time it takes to address a driver becomes an important consideration.
For each criterion, a value of 1-10 is applied, where one indicates very little , five indicates moderate , and ten indicates considerable . Once assigned, the values are used to calculate an Overall Priority Rank (OPR) for each driver. The calculation for this is: OPR = (IM + IN) – T, where IM is the assigned value for the Impact criterion, IN is the assigned value for the Influence criterion, and T is the assigned value for the Time criterion.
This formula summarizes the criteria values assigned to each driver and provides a relative numerical ranking (OPR range -8 to 19) that can be used for decisions about prioritization. Problems addressed in action research are situationally-dependent and often complex \citep*{Berwick}; and the values applied to these criteria are sensitive to researcher interpretation and contextual factors \citep*{Polkinghoime}. Variation would be expected if these were generated for the same problem by different researchers or for different locations.

Plan for Research

The final step of the analysis process is to plan for upcoming research to address the problem by developing a theory of action. Solutions to problems are often proposed and implemented without a clear understanding of the situation or larger strategic vision \citep*{Williamsa}. An effective theory of action provides this strategic vision by establishing a process for addressing the problem, and a rationale that is situationally grounded in the problem framing and identified contextual factors. This is not a hypothesis nor an explication of methods; rather, a theory of action provides a broad argument for why a problem should be addressed in a particular way given its current framing and deconstruction.
This analysis process is grounded in the belief that research is more actionable when addressing problems that have been deconstructed into their root causes. Prioritization of these identified root causes (i.e., drivers) points to those areas of the problem that should be addressed first; therefore, priority should be given here to drivers with relatively larger OPRs. A well-developed theory of action summarizes all steps of the analysis process and can serve as the nexus for follow-on studies targeting each of the identified drivers.