One of the biggest problems I see with most paranormal groups is how they collect their data. Many investigators wander around with a single EMF meter and get excited when they observe a spike in their reading. Many also will be quick to conclude that the spike was paranormal in nature with nothing more to support it. Some go on to claim that because they collected an EVP at the same time as the EMF anomaly, the EVP substantiates their EMF readings.

First, let’s deal with the single meter issue. By using a single meter you are only sampling one data point. Not much can be proven with a single data point, and drawing conclusions from a single data point can easily lead to false conclusions.

Example:
Let’s say a car manufacturer produces 100,000 cars of a particular model. Out of the 100,000 cars, 0.1% has a manufacturing defect. You purchase one of these cars and find that it’s defective. Now, because the car you purchased is defective, you conclude that the model of car is a lemon, or worse yet, you decide that the manufacturer produces junk cars. Does this seem like a fair assessment? After all, you can show people that the car you have is defective.

This is how much of the data is collected and analyzed by many paranormal investigators. Let’s go back to the single EMF meter. What most investigators do is very similar to the example above. They get a spike on an EMF meter and think that it’s something significant or that they can draw conclusions from that one spike. About all that can truly be concluded by their evidence so far is that there was a spike. With only a single meter they can’t even determine if the spike is confined to a small area, as in a particular room in a building, or if the spike is much larger, perhaps covering an area of several hundred acres. Determining that a single EMF spike is paranormal in nature is just as irresponsible as concluding that the model of car in the above example is defective.

For investigators to be able to start making any sort of reasonable conclusions from their data they must first look at their collection methods. Does the method create a single data point or does it produce multiple data points? If your data collection only produces a single data point, then it’s time for a change unless you wish to spend your time collecting useless data. In the instance of EMF, multiple meters need to be employed. The more meters, the more data points. The more data points, the more accurate your conclusions are likely to be, especially if you are trying to determine the area affected by an EMF spike.

Once again let’s look at the car example. Let’s say we buy 2 more cars so that we now have a total of three. Again, the two additional cars are defective. Odds are against us getting three defective cars, but it is possible. Now, while we’ve expanded our sample size, effectively increasing its size by 100%, overall we still have a very small sample in comparison to the total number of vehicles that exists. Because we coincidentally purchased two more defective cars, we mistakenly drew the conclusion that all cars of that model are lemons. But look at what happens when we purchase a total of 1,000 cars. Even if we did somehow manage to purchase all the defective vehicles (total of 100 out of 100,000 at a 0.1% defect rate) we would only find a 10% defect rate, not the 100% defect rate that we had previously assumed. This example with a larger sample size shows us that the car is not a lemon as we had initially concluded in our original samples and illustrates, in general, how easy it is to make an inaccurate judgment based on insufficient data.

So as can be seen, one data point does not provide sufficient evidence and surely does not give an investigator enough data from which to draw any sort of conclusions that would be considered acceptable by any peer group in the scientific community.