Look Before You Leap: Make Strong Inferences

 “Draw inferences from data, observations or other evidence and recognize when firm inferences cannot be drawn.” 

What is data?  “Factual information (as measurements or statistics) used as a basis for reasoning, discussion, or calculation.”  It is useful to keep this in mind when assessing claims.  Observations are, “act(s) of recognizing and noting … fact(s) or occurrence(s) often involving measurement with instruments.”  Evidence can be described as, “an outward sign.”

What data, observations, and evidence have in common is that they are things that demonstrate a particular state of affairs without requiring inferential reasoning.  In other words, they are things that stand for other things.  To make sense of evidence, which I shall use as shorthand to describe data, observations, and evidence, inferential reasoning is necessary.  In other words, evidence only means something if we connect the dots between multiple pieces of evidence and what those pieces stand for.  For example, in the forensic setting DNA evidence simply means that a biologic sample from a particular person was found at a particular location.  This in itself does not allow us to draw an inference that the person whose DNA is found at the scene of a crime committed the crime.  Instead, we use inferential reasoning based on additional factors to conclude that the DNA evidence means that the person being accused of the crime and whose DNA was found at the scene committed the crime.  Typically, forensic experts and attorneys make the inference based on one of two possible factual scenarios.  First, that the suspect’s DNA is the only DNA present other than the victim’s and there is no connection between the two persons.  Second, the suspect has a connection to the victim but is the only person whose DNA was present and who did not have a plausible alibi.  In essence, the DNA places the person at a location and the background information allows us to infer that the person was at the location at a particular time i.e., when the crime was being committed.  The evidence is important, but only insofar as strong inferences can be drawn from it.

In the context of claims, the disconnect between evidence and inference arises frequently.  For example, we often assume that adverse employment actions precipitate claims when there is some proximity between event and claim.  Essentially, we assume that an adverse employment event triggers an emotional response in the claimant that causes her to want to punish the employer for the adverse event.  Our assumption is based on a socio-culturally transmitted understanding of human psychology and behavior.  What we do not in fact know is whether there is an actual link between our assumptions about how people respond to bad news on the job and specific behavior, in this case making a worker’s compensation claim.  In short, we do not know in fact if upset workers make more claims.  

The problem with making weak inferences from evidence is that we have little idea as to the validity of the inference; hence, we could be making strategic claim handling decisions based on what amounts to little more than a superstition masquerading as a fact.  Instead of connecting an adverse employment event and a subsequent claim, we should note the two occurrences and investigate to determine if there is additional evidence to make the inference strong.  Some factors that would help us determine the strength of our inference would include:

  • The severity of the injury (broken leg v. back strain based wholly on subjective complaints);
  • Proximity of injury to adverse employment event in time (obviously the closer in time, the stronger the inference);
  • Admissions to third parties or lack thereof (“I’m going to do something about this!”);
  • Witnesses or video evidence of the injury;
  • Frequency of type of injury among employees doing the same work;
  • Whether injury occurred before adverse employment event but was only reported afterwards;
  • Patterns of injury reports (is the report, in spite of the adverse employment event, consistent with how the claimant or other employees usually reports injuries?);
  • The personnel file (is the employee generally happy with good reviews?  Does she have a bad attitude and have poor reviews?)
  • Severity of the adverse employment action (dismissal v. verbal warning).

So if we have an employee who is written up for punching out too early and a week later falls off a platform and breaks his leg, no one could reasonable suggest the two occurrences are related.  On the other hand, if an employee is suspended without pay for a week because he doctored a time card and claims an unwitnessed low back injury on his first day back after having been overheard by two co-workers shouting, “You’ll never get away with this!” at his supervisor when leaving after being suspended, the adverse employment action and the injury appear to have a relationship that is more than coincidental.

The point of this discussion is that we need to evaluate what inferences the evidence allows us to make and whether those inferences are strong or weak.  To the extent possible, strategic decisions should only be made based on strong inferences.  In addition, we should explore weak inferences to determine if we are likely to find evidence in an investigation that will strengthen the inference.  It is only in the context of strong or firm inferences that we can make rational decisions the outcomes of which are predictable.  Otherwise our decisions will not be based on reason and will have unpredictable outcomes, which bears all the hallmarks of decision making based on superstition rather than fact.


log in to comment

Back to Blog
Recent Posts