from the literature can and should eventually supplement this list, but the focus in Step 2 is on the team's personal experience and subsequent hunches about variables that could affect the variable of interest.
Step 3: Organize the Predictor Variables by Similarity to Form a Structural Model
Have the team organize into groups all predictor variables that seem to be similar to one another. For now, you will simply separate them into columns or write them on separate sheets of paper. For example, one group of variables that may be found to predict falls may be “patient‐related,” such as the patient's age, level of mobility, the different diagnoses the patient is dealing with, and so on. These could all be listed in a construct under the heading “patient‐related variables.” You might also create a construct for “staff‐related variables,” such as “was walking with a staff member,” “staff member’s level of training in ambulating patients,” “the staff member was new to this type of unit,” and so on.
Step 4: Rank Predictor Variables Based on How Directly They Appear to Relate to the Variable of Interest
First, rank the individual variables within each construct, determining which of the variables within each construct appear to relate most directly to the variable of interest. These will be thought of as your most influential variables. Knowledge from the literature of what relates to the variable of interest is welcome at this point, but you should continue to give extra credence to the clinical experience of the team and what their personal hunches are regarding the relevance of each predictor variable. Once the predictor variables within each construct are ranked, then rank each overall construct based on how directly the entire grouping of predictor variables appears to relate to the variable of interest.
Step 5: Structure the Predictor Variables into a Model in Order to Visually Communicate Their Relationship to the Variable of Interest
It will eventually be important to get others on board who are interested in studying and improving the variable of interest. This can be done by noting the variable of interest in a circle in the middle of a blank page and then grouping all the predictor variables around the outcome variable so the visual looks a bit like the hub of a bike wheel with each predictor variable connected to the outcome variable by a spoke. Figure 1.1 is an example of a structural model that is ready to be converted to a measurement model.
For ease of understanding, items in the same construct would be the same color, and items in each color would then be arranged with those considered most influential positioned closest to the variable of interest. Selecting the most influential variables is important because you may decide you have only enough time or resources to address some of the variables. If this is the case, select those variables that are perceived to be the most influential.
A collection of three worksheets used by a neurosurgical nursing care unit for a study on workload, showing the progression from a full list of predictor variables to a workable model (Steps 2–5), can be found in Appendix A. These worksheets visually represent the conversion of a structural model to a measurement model. While Figure 1.1 offers a visual representation of a model with multiple constructs, showing the variables arranged by rank, Appendix A shows a slightly different visual representation, which is the representation typically used by the author of this chapter. Readers are encouraged to try both methods of representing the constructs to see which is more useful in understanding and visually communicating the information.
Figure 1.1 Variable of interest surrounded by constructs of predictor variables, arranged by rank.
Step 6: Evaluate if and/or Where Data on the Predictor Variables Is Already Being Collected (AKA, Data Discovery)
Investigate whether data on any of the predictor variables in your model is already being collected in current databases within your organization. Where you find that data is already being collected, you will use the existing data. You may find that data related to the variables of interest is being collected in more than one place, which will provide an opportunity for consolidation, making your data management process more efficient and standardized.
Step 7: Find Ways to Measure Predictor Variables Not Currently Being Measured
If there are important variables not being measured, it will be necessary to develop ways to measure them. If influential variables are left out of the study, the model will remain mis‐specified (wrong).
If influential variables are left out of the study, the model will remain mis‐specified (wrong).
Step 8: Select an Analytic Method
This work will likely be overseen by an in‐house or consultant mathematician, statistician, or data analyst. Consider types of analytics beyond linear methods or qualitative descriptive methods, which are the most common methods currently used in healthcare. For example, if the dataset is very large and complex and it is not clear how to sort the predictor variables in a linear method such as regression analysis, try Pareto mathematics where outliers are examined. Pareto mathematics looks at the highs and lows in the dataset to create a profile of success factors. In his book Where Medicine Went Wrong: Rediscovering the Path to Complexity, Dr. Bruce West asserts that Gaussian mathematics throws out the extreme values despite the fact that these extreme values often provide the most valuable information because they provide a profile of the biggest successes and the biggest failures (2007).
Another type of analysis to consider is constructal theory, a method derived from physics, which is the study of “flow” (Bejan & Zane, 2012). If employees are able to talk about what makes their work flow or what makes their work pause, constructal theory will allow for their comments to be themed and addressed operationally. It is the experience of this author that if employees can talk about their workflow, and data can then be arranged for them in themes, the employees get excited about working on productivity because it is readily apparent to them that the overall aim is figuring out how to do more of what works well and less of what does not. Constructal theory makes productivity, or the lack of it, visible. Gaussian mathematics provides insight into linear processes, while analyses like Parato mathematics and constructal theory provide insight into more dynamic/complex and unknown processes, respectively. Pairing the analytic method with the variable of interest is important to achieving insight into the operations of work and associated outcomes.
The overall aim is figuring out how to do more of what works well and less of what does not.
For your most complex models, you may want to consider using a machine learning problem designed to let the computer tell you the rank order of predictor variables as they relate to your variable of interest. This is suggested on the condition that you never allow the machine to have the “final say.” Machines function without regard for theory and context, so they are not able to tell a story capable of deeply resonating with the people whose work is being measured. It is tempting to be lazy and not do the work of carefully
