Introduction

The hypothesis of a god’s existence is sometimes said to be untestable. An omnipotent god could create a universe for us and leave no evidence indicating its existence. Such a god, or other supernatural characters, could even manufacture evidence contradicting god theories, such as an unimaginably vast universe or fossils.

For these reasons, both theists and agnostics sometimes argue the existence of god can only be proven, and cannot be conclusively disproven.

One cannot prove a negative, the saying goes; but what, then, is a null hypothesis?

In the scientific method, the alternative hypothesis is a statement about the existence of a relationship between two things (e.g. smoking causes cancer). Experiments often test the alternative hypothesis by attempting to disprove the null hypothesis (e.g. there is no relationship between smoking and cancer).

Scientists’ starting point is skepticism of the claimed relationship. To overcome our skepticism, evidence must be gathered that would be highly improbable if the null hypothesis is correct.

For example, if there is no relationship between smoking and cancer, why in a random sample of 1,000 smokers and 1,000 non-smokers do a large percentage of smokers get cancer and a much smaller percentage of nonsmokers get cancer? This and the results of dozens of other studies create a case for rejection of the null hypothesis.

The question of god can be approached similarly. The alternative hypothesis is that god is causally related to phenomena. The null hypothesis is that god has nothing to do with phenomena. We can disprove the null hypothesis if it is possible to create two treatment conditions: a control group of observations and a group of observations the proposed god would necessarily have to be involved with.

What conditions would necessarily draw in the involvement of a god? This depends on the properties of the god being tested.

Humans have believed in thousands of gods throughout history, and the differences between those theories are the properties associated to each god. These properties include the name of the god, methods to summon the god’s attention, the god’s exchange terms, its relationship to other gods or supernatural forces, gender, powers, level of awareness, personality, history, and objectives. To test for the effects of a god, it is necessary to pick a god, because a proposed god’s properties must determine its only potentially measurable effects.

A proposed god whose properties include an interest in convincing humanity of its non-existence might be impossible to test for. Such a god could simply confound any experiment.

However, I propose we can test for a god with the following properties:

a) Omniscient,

b) Omnipotent,

c) Wants humans to discover it.

Most sects of the Abrahamic religions (Judaism, Christianity, and Islam), and many of the Dharmic religions (Hinduism, Jainism, Buddhism, and Sikhism) propose a god with exactly these properties. This experiment seeks to test for the existence of such a god. In doing so, we test the actual specific belief of most people, rather than a philosophical construct with undefined properties.

In operational terms, omniscience means the god, if it exists, is aware of the experiment and experimenters. Omnipotence means if the god wants to intervene in the experiment, it is able. Finally, the god’s desire for humans to discover it provides the incentive for the god to intervene in an experimental setting.

Method

An empty Styrofoam 12-ounce cup, weighing less than half a gram, was placed on a level table. Windows, doors, fans, and the building’s HVAC system were shut off to ensure air currents did not affect the foam cup. The experimenter observed the table and the cup from a seated position approximately 3 meters away from the table. The experimenter made no physical movements that could have affected the cup.

In both the control and experimental conditions, the cup was observed for one hour. In the control condition, no intervention occurred. In the experimental condition, the experimenter recited the following statement every five minutes:

“God, I personally guarantee I will believe you exist if you tip over that cup in the next five minutes.”     

Results

The cup tipped over zero times in either the control or experimental conditions. As a result, no statistical testing was necessary or possible.

Discussion

The experiment was designed to identify evidence against the null hypothesis: that an omniscient, omnipotent god with a desire for humans to discover it has nothing to do with phenomena.

If the cup had tipped over significantly more often in the experimental condition than in the control condition, evidence would suggest the null hypothesis is incorrect. Given the lack of tip-overs in either condition, we cannot reject the null hypothesis.

The study is limited to god constructs featuring all three of the properties described above. The most parsimonious explanation for the null result is simply: the god does not exist. However, the following alternative explanations and rebuttals are provided.

• God has unknown motivations to avoid detection by humans, though no such motivation is described in the most popular holy texts.

• God cannot interact with our reality on a regular basis, for example, so that a deterministic plan remains unchanged. Note how there would seem to be a limitation on god’s omnipotence if it could not both intervene and get the deterministic result it desires.

• God finds experiments such as this offensive, laughable, or unworthy of attention. Yet, it would seem that if a god wanted humans to be aware of its existence, the most direct and effective pathway toward that goal would be to interact with us via our most successful knowledge development technology, the scientific method. A reliance on one of our least reliable techniques – faith – would seem a bad choice to achieve the goal of human belief.

The author encourages replication of this experiment, but doubts additional trials are needed to elicit a response from an omniscient, omnipotent god who wants to respond.