the probability of god test explained

Is there a God?

What this page is about

The probability of God is a test in which you answer ten questions about the universe, the world and humanity, from which a statistical calculation of the probability of God existing shows the logical conclusions of your opinions.

This page outlines the background information to that test. (This is a revised version of the original test firat posted in 2008.)

The questions

For millennia, people have debated whether God exists, using a range of arguments that have been refined over that time. Most philosophical and many popular books on this subject analyse much the same set of arguments.


For this test, I have used the following books and websites to determine the main topics:


These references led me to consider the following arguments and hence the questions used.

  1. Whether the concept of God does or doesn't make sense.
  2. God as the cause of the universe.
  3. God as designer of the universe.
  4. Evil and suffering caused by the natural world.
  5. The origin of human life, with free will, consciousness, purpose and the ability to reason.
  6. The origin of ethics.
  7. Evil and suffering caused by human beings.
  8. Why can't we detect God?
  9. Religious experience and miracles.
  10. Has God revealed himself through a world religion?

Bayesian probability

The idea for a test, and the mathematics for this probability calculation, came from Stephen Unwin's book, "The Probability of God".

The approach to probability we are mostly familiar with is based on the frequency of occurrence of an event. For example, if you roll a die many times, each of the faces will come up with a frequency of about 1/6, so that is the probability of throwing a six, or any other number. But for some calculations, we don't have a number of trials from which to calculate a frequency - for example, if we wanted to estimate the probability that human life would evolve on earth, or the probability that a nuclear reactor will melt down.

Bayesian probability is often used in risk assessment. It starts with an initial estimate of an occurrence, and then considers how new information changes that probability. An estimate is made of how likely the new information would be if the occurrence is true, and if the occurrence is false. For example, if one is estimating the probability that a murder suspect is guilty, and then some new evidence is received (say the results of a chemical analysis of soil on the suspect's shoe), then one can ask what is the probability of this result if the suspect is guilty and if the suspect is innocent, and the Bayesian equation allows calculation of a new probability.

It is clear that the probability of God's existence suits this type of analysis. We can make an initial assessment of how sensible the concept of God is (see below, prior probability), then examine how this probability changes as we consider different pieces of evidence - for example, how likely is it that human beings would evolve (i) on the assumption that God exists, and (ii) on the assumption that God doesn't exist.

The Equation

The Bayesian equation (for the comparison of two hypotheses, H1 and H2) is

Prafter = (Prbefore x Pr of X if H1 is true) /

(Prbefore x Pr of X if H1 is true) + ((100% - Prbefore) x Pr of X if H2 is true)

Assigned probabilities

To facilitate making choices, I have chosen a range of probabilities for each question, each with a simple verbal description.

Question 1 concerns prior probability (see below), so I have used an assessment of the philosphical coherence of the concept of "God". The 7 possible choices are 5%, 10%, 25%, 50%, 75%, 90% and 95%.

For the remaining questions, which require the assessment of the probability of some fact being more or less likely to be true if God actually exists, compared to the case if God doesn't actually exist, I have used the probability ratios of 0.2, 0.4, 0.625, 1, 1.6, 2.5, 5. (For the test to be unbiased, it is necessary that the opposite pairs of numbers be reciprocals.)

The numbers have a smaller range than Unwin uses in his book. I initially tried using his wider range but this tends to give extreme values for the final probability (either close to 0 or 1). This both uninteresting and implies a higher degree of certainty than we can reasonably claim.

So for this revision, I have used the smaller range, where certainty is defined as beling 5 times more likely that the alternative.

Prior probability

This is probably the most contentious and difficult part of the Bayesian assessment (see for example papers by Robert Prevost, Gordon McCabe, Drew Thomas and Gabe Czobel).

Author Stephen Unwin, who has a relevant background in risk assessment, argues that the initial probability should be taken as 50/50, based on the fact that we really don't know how likely or unlikely it is that a God might exist. Christian philosopher Richard Swinburne justifies a similar starting point based primarily on the argument that the God explanation is "simpler" than alternatives.

However many sceptics argue that the prior probability should be set much lower. Some argue that the concept of God is incoherent or meaningless. But perhaps the most common argument is that "propositions that postulate existence have a far less than 50 percent chance of being correct" (Larry Ford quoted by Victor Stenger - Ford suggests 1 in a million rather than 50/50). I don't know any reason to accept that argument more than any other (it would seem that there are as many real beings as there have ever been imagined beings, not a million times more imagined ones as Ford seems to imply), but it is nevertheless a common position.

This wide variation in opinion suggests that our choice of prior probability reveals as much about our own bias as about anything objective. One view is that we should accept that this estimate of prior probability is subjective, but since the whole thing is subjective, not scientifically factual, this doesn't matter. I have taken this view, and so offer those taking the test a range of prior probabilities to choose from.

can we take this seriously?

I do not pretend that we can make accurate assessments of these probabilities, but all we are doing is make assessments of what each of us believes. At the very least it is fun, and of interest. At best, it gives us a useful insight into what we believe about God, and the factors that most influence what we believe, and thus allows us to further consider the evidence.

Return to the Probability of God test.