Anomalous Evidence and Proof of Paranormal Activity

Documentation of anomalous data in paranormal surveillance

During basic paranormal field research we collect site information, interview claimants, and at times conduct surveillance of the client and their home in an attempt to document the experiences that they claim to have had.  While reviewing surveillance data we often encounter anomalous events that cannot be explained except through speculation.  Most of the time these events cannot be explained because we simply do not have enough recorded information to piece together the actual cause.  Even if the event can be explained it may be directly related to the client’s claims and needs to be documented.

The term paranormal basically means that the event is beyond any natural explanation, which in other terms means that all known, possible causes have been investigated and found to be invalid.   Of course, the opposite of a paranormal event is an explainable event, and when investigating anomalous activity we could also be attempting to determine if it is an explainable rather than paranormal experience. 

I need to state a disclaimer here.  The research of paranormal claims does not include supernatural explanations.  This is primarily because any supernatural explanation is subjective and primarily based upon a person’s beliefs.  Belief is important, but as different people often have different beliefs we cannot support any one supernatural theory over another.  Any suggestion of supernatural beliefs by an investigator will likely influence the beliefs of the client and ultimately contaminate the research data.

The attitude of an investigator toward supernatural explanations should be vague and favoring the acceptance of the client’s beliefs. In any case, the Client’s beliefs on their experiences need to be documented as they are extremely relevant to the analysis of the claims.  If the client does not know what they believe, it should be noted as such and left as is without any suggestions by the investigator.

There is a controversial space between an explainable event and what can be called paranormal.  One simple approach usually states that any event where a natural explanation cannot be proven is considered paranormal. The opposite approach that states any event where a paranormal explanation cannot be proven should be considered explainable.  Between both of these simple approaches is a large space of unknowns that keep the event from being classified as either extreme.

These unknowns could often come from lack of information; data that was not collected at the time or simply the fact nobody can know everything about all the possible explanations that could be applied.  The size of this unknown area between plausible explanations is the real factor that determines how confident we can be about our analysis of anomalous evidence.

So what do we do about the anomalous data?  One popular method for reporting the normality of an event is called odds against chance.  Basically, if an event has an expected chance of occurring, then the odds against that chance is a measurement of how unlikely the actual occurrence is in comparison.  The problem is that paranormal events do not have as many controlled variables to make this an easy calculation.

Odds  =  ( # Occurrences – (Expected Occurrence  ) ÷ √ [ Expected Occurrence × ( 1 - Chance ) ]

This is the difference between actual and expected results compared to the standard deviation of the chance. 

Standard deviation = √ [ Expected Occurrence × ( 1 - Chance ) ]   

While this is not the normal  statistical form of standard deviation, I found this variation to be pretty useful for the application.   This represent the amount of variation from the expected results that is considered to be normally within chance.  Any actual results that vary more than this amount may be paranormal.

While this method works fine for statistical studies ( E.G. Psychical testing ) it becomes much more complex when applied to paranormal experiences.  There is the problem of determining the chance of occurrence for an event that normally should not be occurring In the first place.  Maximum possible occurrences and expected occurrence is also a problem because it is really dependent upon the nature of the event.

We do know from experience that there are quite a few ways that we can get unexpected results; no system of measurement is completely accurate and foolproof.  So an inherent chance of error can be applied to any form of evidence, this chance being higher or lower depending upon the actual nature of the evidence we have collected. 

Rather than trying to determine the actual chance of an event occurring naturally, we can apply a general assessment of the evidence.  When listening to a person talk about paranormal experiences we often here a number of considerations used to justify the legitimacy of paranormal claims.  It basically starts off with a simple claim that could probably be explained, and then the story expands with numerous descriptions that make the event seem more unlikely.  This continues until the event seems so improbable that everyone agrees that it must be paranormal.

I try to simulate this by describing the evidence by four basic criteria:

Quality, Relevance, Support, and Significance.

Quality is probably the least important characteristic, but good quality evidence really helps people believe that what they are seeing or hearing is authentic. Mumbled recordings and blurry pictures are not really that convincing when it comes to paranormal evidence.

Relevance is important to our analysis as it reflects upon the claims that are being investigated. It can also be considered when the evidence is directly in response to the investigators actions. Basically, relevance is any coincidence that makes you question whether or not the event could happen by chance. The more unlikely the coincidence is, the higher the relevance to a paranormal claim.

Support is the amount of other anomalous events that occurred around the same time as the evidence. Data such as temperature or EMF fluctuations are examples of supporting data. Paranormal evidence is more plausible when multiple events occur around the same time because a chain of unexplained activity has a much lower chance of occurring naturally than a single event. The supporting data does not have to be significant, but does need to be confirmed as unexplained.  If there is only one piece of evidence then the support value is 1, personal experiences count as 0 unless they are corroborated in which case they are 1.

Significance is an estimate of the commonality of the event. If the event could happen easily by natural means, and is expected to be a pretty common occurrence, then the significance is low. If the event has no known causes, very few natural causes, or natural causes that have been proven to be inapplicable to the event then the significance is high.

These four characteristics could be used to determine the odds against chance for the evidence to occur naturally.

Each characteristic is rated from 1 to 5; 

1 is Horrible Quality, Totally Irrelevant,  No Other Support, or Insignificant.

2 is Poor Quality, Somewhat Irrelevant,  1 Other Support, or Minor Significance.

3 is Average Quality, Possible Relevance,  2 Other Support, or Possibly Significant.

4 is Good Quality, Somewhat Relevant,  3 Other Support, or Significant.

5 is Excellent Quality, Very Relevance,  4+ Other Support, or Very Significant.

I use the following relationships to fill in the Odds formula above.  

# of Occurrences is the amount of data that supports the evidence = Support

The Max Occurrences is the total amount of evidence = 5 ÷ Significance

The Chance of the event occurring is modified by the relevance and the quality = 1 ÷ [ Relevance × ( Quality + 2.067 ) ]

The Expected Occurrence is calculated from the other values = Chance × Max Occurrence 

The constants used in these equations are there to set the range of expected odds to best evaluate the Event.

Substituting these into the Odds against Chance equation, and combining as needed, we get :


                   [ Support - 5 ÷ (Significance × Quality × ( Relevance + 2.067 ) ) 


      √ { [ 5 ÷ (Significance ×  Quality × ( Relevance + 2.076 ) ) ] × [ 1 - 1 ÷ ( Quality × ( Relevance + 2.076 ) ) ] }


This formula will give a maximum value of 30 and a minimum value of -0.6. 

More significance, support, relevance and quality will result in higher odds against.

This is read: 

Odds of chance of 1 to (the calculated Number) against in (Max Occurrence)  

A 1 to 30 is highly improbably, a 1 to 0 is exactly chance, a 1 in 1 against is normal, a 1 to -0.6  very common.

I would then classify the odds against into the following probability ( 5-9-9-5 ) distribution:

4.99 and below (Above 83.3% Common) = Paranormally Inadmissible Evidence

5 to 13.99 (83.3-53.3% Common)= Paranormally Questionable Evidence

14.00 to 22.99 (53.3-23.3% Common) = Paranormally Significant Evidence

23 and above (Below 23.3% Common) = Paranormally Plausible Evidence

Having a complete record of your analysis provides a firm basis for consistency and can be used to back up any conclusions you reach about the evidence collected. It also provides a way to correlate data from multiple investigations to recognize patterns and establish hypotheses about paranormal experiences.

On this scale, 1:1 is 96.7% common, while 1:15 is 50% common and 1:28 is only 6.7% common. 

This % probability is calculated by  100 x (1-(# against ) ÷ 30)  

Classification of evidence, regardless of proof, helps us identify what may be or may not be paranormal activity in future investigations. All evidence, when properly classified and documented, adds to our understanding of the paranormal, can bring us closer to repeatable working models, and helps expand our knowledge of the unknown.

However, putting the technicality and scientific aspects aside, there is no reason that we cannot enjoy the prospects that any compelling evidence found could be paranormal in nature. This is the reason most people get involved in paranormal investigation in the first place


*** Note:  previous versions of this article and evidence reports I have done in the past may vary from the information given here.  The method is the same, but the scale of the calculations might have changed slightly as I try the formulas out in the field.