7P13SR- ONE- SHEET5
MN7P13 Building Business Insights Workshop 2: Disaggregation of problem structure and solution drivers
Dr. Stephen Hills
Disaggregation of problem structure and solution drivers
The seven-steps process
How do you define a problem in a precise way to meet the decision maker’s needs?
How do you disaggregate the issues and develop hypotheses to be explored?
How do you prioritize what to do and what not to do?
How do you develop a workplan and assign analytical tasks?
How do you decide on the fact gathering and analysis to resolve the issues, while avoiding cognitive biases?
How do you go about synthesizing the findings to highlight insights?
How do you communicate them in a compelling way?
Step 2: Disaggregate the issues
Types of logic trees
Early in the process we start with factor/level/component trees to help us define basic problem structure.
Later in the process we move to hypothesis trees, deductive logic trees or decision trees, depending on the nature of the problem, so to drive analysis or action.
MECE: Logic trees should have branches that are…
| Mutually Exclusive | The branches of the tree don’t overlap, or contain partial elements of the same factor or component. The core concept of each trunk or branch of the problem is self-contained, not spread across several branches. |
| Collectively Exhaustive | Taken as a whole, the logic tree contains all of the elements of the problem, not just some of them. Missing parts could lead to missing solutions to the problem. |
2. Disaggregation of problem structure and solution drivers (25%, 2000 words)
Use an initial logic tree (i.e., factor/lever/component or inductive logic) to break the problem into component parts or issues to illustrate and define the basic structure of the problem (e.g., causes of the problem).
This should be evidence-based, using a combination of credible industry and academic literature, evidence and theory, covering the problem generally and the problem in the context of your client.
Provide a fully-referenced commentary of the logic tree, concluding with a summary of the insights gained. It is expected that this logic tree will have three layers.
Using the basic problem structure logic tree as a guide to locate further industry and academic literature, evidence and theory on the problem component parts of issues, produce a more complete logic tree (i.e., deductive logic, hypothesis or decision) of the drivers of the problem solution, which help us to see potential pathways to solve the problem (e.g., hypothesised solutions).
Provide a fully-referenced commentary of the logic tree, concluding with a summary of the insights gained. It is expected that this logic tree will have four layers.
Use an initial logic tree (i.e., factor/lever/component or inductive logic) to break the problem into component parts or issues to illustrate and define the basic structure of the problem (e.g., causes of the problem). This should be evidence-based, using a combination of credible industry and academic literature, evidence and theory, covering the problem generally and the problem in the context of your client. Provide a fully-referenced commentary of the logic tree, concluding with a summary of the insights gained. It is expected that this logic tree will have three layers.
Problem disaggregation
Any problem of real consequence is too complicated to solve without breaking it down into logical parts that help us understand the drivers or causes of the problem.
We need to take the problem apart in a way that helps us to see potential pathways to solve it.
Taking the problem apart to see all of its parts clearly also allows us to determine what not to work on:
The problem components or issues that are too difficult to change (i.e., that can be actively managed).
The problem components or issues that don’t impact the problem sufficiently.
Factor/lever/component logic tree
Structures for seeing elements of a problem clearly.
Schemas that provide a visual mental map of the different levels of a problem.
Clear logic of relationships linking component parts of the problem to each other.
Factor/lever/component logic tree
At the start of this step, when you are able to state your problem clearly but don’t yet have a detailed understanding of it, you should employ the simplest kind of logic tree.
Start with the most obvious elements that make up a problem - components that can help focus data gathering.
A logical first disaggregation can usually be achieved with a small amount of Internet research.
Literature reviews and theoretical frameworks
A literature review (e.g. Google Scholar search) of the facets of the problem will provide insight into the different ways that a problem can be broken up.
Using knowledge from the literature you can develop a theoretical framework of the drivers or causes of the problem.
From this we can develop hypotheses of pathways to the solution – our evidence-based predictions of potential pathways to a solution that we can go on to test.
Using the basic problem structure logic tree as a guide to locate further industry and academic literature, evidence and theory on the problem component parts of issues, produce a more complete logic tree (i.e., deductive logic, hypothesis or decision) of the drivers of the problem solution, which help us to see potential pathways to solve the problem (e.g., hypothesised solutions). Provide a fully-referenced commentary of the logic tree, concluding with a summary of the insights gained. It is expected that this logic tree will have four layers.
Hypothesis logic tree
After a literature review and other in-depth research, it is possible to refine a logic tree and transition from a simple factor/lever/component logic tree to a hypothesis tree – predictions of solutions that need to be tested.
Deductive logic trees
Appropriate for when you have a very clear idea of the problem structure, which is logically or mathematically coherent.
Use deductive reasoning (a.k.a. top-down reasoning) that argue from general rules or principles to conclusions via more specific data and assertions.
General statement: All LMU MBA students need a minimum of a 2.2 for an honours degree (or equivalent) to enter the programme.
Specific observation: Priyanka is a LMU MBA student.
Deductive conclusion: Priyanka has a minimum of a 2.2 for an honours degree (or equivalent).
Deductive logic trees are constructed similarly, with a problem statement that may sometimes be expressed in quantities, and branches that are typically logically or mathematically complete, so that the components add up to the desired objective of the problem statement.
You can use this kind of tree when you know a lot about the logical structure of a problem and especially when the cleaving frame is inherently mathematical.
Pacific Salmon Case: From initial component tree to refined hypothesis tree
Initial factor/lever/component logic tree
A rudimentary factor/lever/component logic tree was developed to get a hold of the problem – to get a grasp of all the elements and relationships that defined the problem space.
For several days (not weeks) undertook readings about the Salmon problem and talked to experts in salmon conservation.
Just enough initial research to generate a first-cut tree, which would then act as a guide to make further research more efficient.
Factor/lever/component logic tree
The big levers that affect salmon and the secondary tertiary layers of the problem without judgment of importance or magnitude of levers or which could be actively managed (i.e., affected by the grant funding).
Hypothesis logic tree
Better organized.
Mutually exclusive & collectively exhaustive.
Focuses analysis on both specific regions & intervention types.
Initial hypotheses to push for some early outcomes (i.e., achieve some traction).
Nursing Case: Deductive logic tree
Case: Improving nursing-related patient outcomes
Focuses attention on the key drivers of nursing numbers and skill levels, moving from general rules or principles to conclusions via more specific data and assertions.
Data and analysis were used to determine which levers were most powerful in improving patient outcomes and which were cost-effective to address.
After 12 years of investment, more than 4.5k registered nurses were added, nursing school curriculums improved, bloodstream infections and readmission rates reduced and 1k lives saved a year from sepsis.
Case: Improving nursing-related patient outcomes
Nurses provide at least 90% of patient care in hospitals.
Over 100k lives a year are lost in the USA from mistakes in patient care in hospitals.
There is a substantial shortage of nurses, resulting in more patients per nurse (or fewer nurses per patient).
For each patient added per nurse, mortality rates increase.
Deductive logic tree: Improving nursing outcomes
This problem is suited to a deductive logic tree because it is logically complete:
Increasing number of skilled new nurses
Improving skills and practices of current nurses
…adds up to the desired outcome.
Conclusions
Conclusions
Problem disaggregation provides us with manageable chunks to work on and allows us to begin to see the structure of the problem.
Start with simple factor/compoenent/lever logic trees when you are starting out and don’t know a lot.
Use those to guide your research then move on to more complete logic trees using hypothesis, deductive and decision trees.
Your logic tree structures should be both mutually exclusive (i.e., no overlapping branches) and collectively exhaustive (i.e., no missing branches).