ESP Test Statistics Explained for Beginners (2026)

How to read ESP test statistics without turning them into a verdict

ESP tests often produce a number—a percentage, a streak, a z-score, or a p-value—and that number can feel more scientific than the experiment that produced it. The first question is not “how impressive is it?” but “what was measured, compared with what baseline, and over how many trials?” A statistic cannot repair an unclear protocol or a selected result.

Begin with the unit of analysis. One guess, one session, one participant, and one repeated experiment are different observations. If a participant makes twenty guesses, treating them as twenty independent people inflates the apparent evidence. If the best run is selected from dozens of runs, the chance of finding a memorable pattern is higher than the headline suggests. Write the denominator before interpreting any percentage.

Four questions for a result

1. What was predicted? Define the target before the reveal and preserve every miss.
2. What was the baseline? State how a random result would be generated, including the number of options and selection rules.
3. How many opportunities were available? Count all trials, not just the final or best run.
4. What decision follows? A useful result may guide the next controlled test; it should not become a promise about a person, illness, or future event.

Keep the raw log beside the summary. A PSI GYM-style log can help you record the protocol, but no application can decide whether your stopping rule was honest. If the result is unexpected, do not immediately add more trials until it passes. First check for transcription errors, protocol drift, and a selected subset.

What a p-value does not mean

A p-value is not the probability that a supernatural explanation is true, nor is it a measure of practical importance. It depends on the model and the assumptions behind the data. A small value can be distorted by multiple testing, a badly chosen baseline, or selective reporting. Explain the result in plain language and keep the calculation attached to the raw observations.

Frequently asked questions

Is a high hit rate proof of ESP?

No. The protocol, chance baseline, number of trials, and selection rules must be examined before making that claim.

Why keep misses?

Misses make the dataset complete and help reveal whether the result depends on cherry-picked trials.

Can a calculator prove paranormal ability?

No. A calculator can process a specified model, but it cannot validate the experiment or the interpretation behind it.

What should I do with an unexpected result?

Preserve the complete log, check the protocol and arithmetic, and do not keep repeating the test solely to obtain a better result.

A pre-registration checklist for ESP statistics

Before looking at the results, write the hypothesis, number of trials, chance baseline, stopping rule, and exclusion policy. This simple pre-registration does not make a study scientific, but it makes it harder to change the story after a surprising score. A PSI GYM-style log should preserve every prediction, including misses, rather than showing only the best run.

Separate three questions: Did the score differ from the stated baseline? Is the result precise enough for the sample? Does the protocol have a plausible alternative explanation? None of those questions automatically answers whether ESP exists. If you calculate many subsets until one looks unusual, that is a selection problem; label it as exploratory and do not call it confirmation.

Review the raw data before the shareable summary. Check transcription, clock errors, missing trials, and the order of reveal. If the study involved another person, obtain consent and avoid publishing names or responses. The reality-check tracker can hold the facts and interpretations separately.

What to write after a result

1. Report the whole experiment, not a favorite streak.
2. State what was measured and what was not.
3. List alternative explanations and limitations.
4. Choose one next test only if it is safe and genuinely different.

A score can inspire a better question, but it cannot diagnose a person or guarantee a supernatural result. If repeated testing creates anxiety, sleep loss, or pressure to achieve a result, pause the practice.