The "Directionally Correct" Fallacy, Part IV: Random Guesses or the Confidence in Probability
When healthcare executives use the phrase “directionally correct,” they are invariably, if unwittingly, using information about past events as a reference point for making decisions about the future. That information about past events is, in addition to being historical, often incomplete or otherwise flawed, wherein lies the “directionally correct” fallacy. Like the clinical concept of Original Antigenic Sin, in which the immune system learns to respond to a virus based upon the specific virus it first encountered, healthcare executives are inclined to make strategic decisions based upon their memory of a strategy that worked in the past. What’s past is not prologue in healthcare, and so history can never be used to predict the future with 100% accuracy, which is the reason that probability theory exists.
Probabilistic decisions are based on probability, that is “the extent to which an event is likely to occur, measured by the ratio of the favorable cases to the whole number of cases possible.”
The issue is not one of degree, but kind. “Directionally correct” decisions can occasionally be correct, or at least not fatally flawed, but they should never be viewed as “evidence-based.” Why? Because “directionally correct” decisions are fundamentally based on insufficient evidence, whether in terms of relevance or detail, and insufficient analytic rigor.
In contrast, probabilistic predictions developed from comprehensive, longitudinal data sets and advanced data science and engineering capabilities are truly “evidence-based.”, These predictions provide a transparency that healthcare executives can, by definition, utilize with confidence to analyze potential outcomes from strategies and tactics.
Of course, evidence-based decisions are not infallible. As famed statistician George Box famously quipped, “All models are wrong, but some are useful.” Importantly, Box was not referencing a projection developed in Excel, but statistical modeling, which is defined this way:
“Statistical modeling is the use of mathematical models and statistical assumptions to generate sample data and make predictions about the real world. A statistical model is a collection of probability distributions on a set of all possible outcomes of an experiment.”
In the new health economy, healthcare executives cannot afford to be satisfied with “directionally correct” – hope is not a strategy. Instead, healthcare executives should focus on two decision categories:
- decisions for which there is insufficient data to inform that decision, and
- decisions based on probabilistic predictions.
The former might involve future policies, technologies, or pandemics. The latter are the strategic and tactical decisions, whether strategic, operational, financial, or clinical, that stakeholders in the health economy expect executives and boards to get right.