What does one do when there is more than one hypothesis that cannot be disproved? Just saying undetermined as to a fire cause is not enough. Undetermined give a false indication. It makes it sounds as if there is possibility of reaching a finding as to the fire cause. What I am say is by using words such as unresolved, undecided, or needing additional investigation would better state the reality of the status of the investigation to determine the actual fire cause.

When one uses the word undetermined, that person is only saying that he cannot make a determination, it does not mean it cannot be made by others. It makes it sound as if he is stopping his investigation without resolution. How would the person answer the question, are you saying you are not continuing to try to answer the question as to the fire cause. If the person answers he is not going forward, then it is pointed out that he has conducted an incomplete investigation.

When an investigator completes the investigation, it is usually followed by a report. In the report the investigator tells a story to explain what took place. The fire cause is the ending of the story. There are six ways to end a story. I will address the two that are the most relevant to fire investigation reports. These endings are resolved and unresolved.

A resolved ending leaves the reader with no lingering questions or loose ends, where an unresolved ending is the kind of ending that leaves the reader with more questions than answers. Which would better relate to undetermined. A resolved ending does not necessarily lead to the desired ending. The unresolved ending leaves the person with more question than answers. With unresolved the reader is expecting there is additional information to come.

There is nothing wrong with the investigator believing he does not have the knowledge or resources to continue the investigation. What would make you look better, just saying undetermined and walk away without an explanation or to say unresolved, but additional information may be obtained through these resources.

When one uses the word undetermined as a fire cause what are they trying to express to the reader of the report. Does that term tell the reader that you have not been able to prove any of the potential hypotheses wrong. Does it say some were proven wrong and others were not. Does it tell the reader that there are multiple hypotheses still possible and further investigation needs to be conducted? Why not just say the facts. Here are the hypotheses that were considered. Here are the ones that were proven wrong, and here are the ones that were not able to be proven wrong and need additional investigation. As one can see, it is my belief that the word undetermined is not appropriate for a fire cause.

You say a fire investigation is a deterministic analysis. This type of analysis is meant to reach an absolute conclusion. This type of investigation is one which aims to demonstrate is to identified faults/hazards. The fault would be the cause for the event and there is nothing else that could have caused it.

In science, ideas can never be completely proved or completely disproved. Instead, science accepts or rejects ideas based on supporting and refuting evidence and may revise those conclusions if warranted by new evidence or perspectives.

The last step of the Scientific Method is usually "draw a conclusion." In fire investigation this is the cause of the fire. It's easy to imagine that investigations that do not reach a clear conclusion must not be scientific. In fact, most scientific and investigations do not reach "firm" conclusions.

Scientific articles or reports usually end with a discussion of the limitations of the tests performed and the alternative hypotheses that might account for the phenomenon. By issuing a report that has an absolute finding as to the fire cause is one eliminating part of the process that is used in scientific research.

The nature of scientific knowledge — it's inherently tentative and could be overturned if new evidence, new interpretations, or a better explanation come along. In science, investigations that carefully analyze the strengths and weaknesses of the test performed and of the different alternative explanations are particularly valuable since they encourage others to more thoroughly scrutinize the ideas and evidence and to develop new ways to test the ideas. What better for one’s conclusion than to have others use the same date and methodology to reach the same conclusion.

To say probabilities have no place in determining the cause of a fire because fire investigation is a sole deterministic analysis is not supported by the facts as to how an investigation is conducted using science as its foundation. Because there can be no absolutes in science on can only say that the determination of the fire cause is based on the known data used at the time the conclusion was reached. Here is where probabilities come into play. What is being said is that the probability of this conclusion being correct is based on the known data. One must concede that there may be date that was not known at the time of the conclusion that could result in a different cause for the fire. If one cannot be accurate is saying the conclusion is an absolute finding then there will always be the probability it is wrong, however small that probability may be.

One of the things supporting the finding mentioned above is the how data from an incident is evaluated. Evaluating data is one of the most important parts of the scientific method. Persons do not always think the same when evaluating the same data. It is for this reason we see different conclusion as to a fire cause when the same data is evaluated. By interjecting into the process the way different people process the same information one is removing the ability for any conclusion as to the fire cause being absolute. At this point it comes down to the weighing of the two bodies of studies and giving them a probability score. The one with the greater score is the one that is accepted. I am not saying which one is the correct finding, because unless one is there when the event takes place, there is no way to say exactly what took place. This leaves us with probabilities. One can say they do not believe that probabilities are what should be used, but the facts of life is that is exactly what is taking place when it come to forensic investigation where the parties where not present or when the cause could not be duplicated base on the exact conditions that were present at the time of the event.

A hypothesis is uniquely consistent with what when it comes to forensic investigations? You are not there at the time of the occurrence to say what took place. Recreating the exact conditions at the time and place of the event presents a different set of circumstances that can alter the results if the conditions being used to support the finding were not accurate.

Saying the application of probabilities is not a reliable methodology when applied to fire or explosion investigations is not using the scientific method leaves the question is this true. In information below is from a text book titled Reproducibility and Replicability in Science.

A commonly used formulation of hypothesis testing is based on the answer to the following question: If the null hypothesis is true, what is the probability of obtaining a difference at least as large as the observed one? In general, the greater the observed difference, the smaller the probability that a difference at least as large as the observed would be obtained when the null hypothesis is true. This probability of obtaining a difference at least as large as the observed when the null hypothesis is true is called the “p-value.”3 As traditionally interpreted, if a calculated p-value is smaller than a defined threshold, the results may be considered statistically significant. A typical threshold may be p ≤ 0.05 or, more stringently, p ≤ 0.01 or p ≤ 0.005.4 In a statement issued in 2016, the American Statistical Association Board (Wasserstine and Lazar, 2016, page 129).

Here is an example that probabilities are being used in science. I understand this is not fire investigation, but I must question why the science of fire investigation differs from other scientific inquiries.

Saying that the scientific method is a deterministic analysis is saying the result one gets from using the scientific method is an absolute fact. All the text about absolutes in science states there are no absolutes in science. This means that anyone finding cannot be considered absolutely correct. Based on this, since there can be no absolutes, then all fire cause determinations must be undetermined. If one is willing to accept a finding that is not considered by science to be an absolute, then one must consider the finding as being a probability if they are to accept the conclusion.
What we are doing in forensic fire investigation is to determine what will take place under known set of circumstances. These circumstances are based on an individual’s evaluation of what was the perceived conditions at the time of the incident. The question here is can the investigator during a post incident investigation accurately exactly state what all the precise conditions at the exact time of the incident were. My experience is this cannot be done. It is for this reason the deterministic methodology is being replaced by the probabilistic methodology. As an example of this is the following:

A deterministic methodology is a method in which the chance of occurrence of the variable involved is ignored and the method or model used is considered to follow a definite law of certainty, and not probability. Since 2000, analysis of dams has changed from a deterministic approach to a probabilistic approach.

Based on the above comment one must conclude that to use the deterministic methodology on must agree to ignore any variables. The methodology requires the investigator to follow a predetermined model based on the conclusion reach during the investigation as to the precise condition at the location of the incident.

Let’s look at the difference in the definition of a deterministic experiment and a non-deterministic experiment.

A deterministic experiment is the subset of Experiments in which there is only one possible outcome. i.e., we know before the experiment what will be the result as there is only one possibility. Random Experiment. A random experiment is the subset of experiments in which there are two or more outcomes.
Non-deterministic:
In probability theory, an experiment or trial is any procedure that can be infinitely repeated and has a well-defined set of possible outcomes, known as the sample space. An experiment is said to be random if it has more than one possible outcome, and deterministic if it has only one.

Which one of these best describes the testing of the hypothesis in the scientific method found in 921.

Jim Mazerat
Forensic Investigations Group