The experiments that led to the conclusion that intention effects are real are conceptually simple. The underlying premise behind the experiments is that a process should be run under different conditions with all but one variable being held constant. For example, in a regular experiment, a group of plants might be grown in identical rooms under identical environmental conditions, but half of the plants would be given fertilizer and another half would not. By waiting a few months and allowing the plants to grow, the experimenter can look for differences in the populations and determine whether or not fertilizer had an impact on plant growth.

In the case of the study that Jahn’s student wanted to conduct, the constant being evaluated would be the output of a Random Event Generator, and the condition that would be varied was the intention of a human operator. In one condition, which we will refer to as the "calibration condition", the device would be left to itself to run without any operators attempting to influence it. In the other condition, which we will refer to as the "active intention condition", an operator is provided feedback and asked to influence the outcome of the device using only his or her will.

To be able to accurately compare the outcome under these conditions, we must first think back to the design and behavior of the Random Event Generator. As was mentioned before, the Random Event Generator used in these experiments is designed to produce a truly random output that can take on two states. Let us now refer to them as 1 and 0, which correspond to the idea of heads and tails respectively. If the device behaves as we expect it to, about 50% of the outputs that it produces should be 1s and the other 50% should be 0s. At this point, it is important to remember that we are not saying that the output will always consist of perfect 50/50 mix, but rather that it should follow a known statistical distribution with these properties.

If this sounds confusing, imagine that you have a fair coin. Even though any given coin flip has a 50% chance of landing on heads or tails, you can never be 100% certain of the next outcome. If you flip the coin 10 times, you might find that you have 3 heads and 7 tails, or perhaps 6 heads and 4 tails. In either case, these outcomes would be well within your expectation. Now imagine that you flip the coin 100 times; you wouldn't be surprised if you found 54 heads and 46 tails, but you might think that something strange was going on if the coin produced 80 heads and 20 tails. If you flipped the coin a million times and it produced 800,000 heads and 200,000 tails, you would almost certainly feel as though the coin was not behaving as it should.

This exact logic applies to the Random Event Generator. The device is designed to produce output that is statistically perfect and unpredictable, and when run in a calibration mode, its output is shown to be exactly what we would expect from a balanced coin. The interesting result is that when subject to the influence of human intention for a large number of outcomes, the researchers found that the devices produces results more like that of an unbalanced coin. More specifically, when operators were trying to produce a "heads" outcome, it was found that the generator produced more 1s that would be expected by chance or calibrations. When the operators shifted their intention to "tails", the opposite occurred.