Skip to content
Omnikalc

Index of dispersion calculator: are your counts random, clumped or regular?

Enter the number of counts, their mean and their variance to get the variance-to-mean ratio, the chi-square statistic and the p-values against a Poisson pattern.

Updated on
Your inputs

Value out of range

Value out of range

Value out of range

Results

Index of dispersion (variance ÷ mean)

1.857

Above 1: counts are more spread out than Poisson (clumped)
Chi-square statistic (n − 1) × index
35.286
Degrees of freedom
19
p-value for overdispersion (clumped counts)
0.0129
p-value for underdispersion (regular counts)
0.9871
Two-sided p-value against the Poisson model
0.0258
The counts differ significantly from a Poisson pattern at the 5 % level

Random events do not always follow the Poisson pattern. The index of dispersion calculator takes the number of counts, their mean and their variance, and gives the index of dispersion, the chi-square statistic, the degrees of freedom and the p-values for overdispersion, underdispersion and a two-sided test.

How the calculation works

  • Enter the number of counts n, for example the number of plots, time intervals or samples, with the mean and the sample variance of the counts.
  • The index of dispersion, also called the variance-to-mean ratio or the Fano factor, is variance ÷ mean. A Poisson process has an index of 1.
  • The chi-square statistic is (n − 1) × index, compared with a chi-square distribution with n − 1 degrees of freedom.
  • A large index with a small p-value for overdispersion means that the counts are clumped. A small index with a small p-value for underdispersion means that they are more regular than random.

Example

A survey counts a plant in 20 plots, with a mean of 4.2 and a variance of 7.8. The index of dispersion is 7.8 ÷ 4.2 = 1.857, and the chi-square statistic is 19 × 1.857 = 35.29 with 19 degrees of freedom. The p-value for overdispersion is 0.0129, and the two-sided p-value is 0.0258, so the plants are clumped and not randomly scattered.

Why it matters

Ecologists use the index of dispersion to tell if organisms are clumped, random or regular. Epidemiologists, quality engineers and neuroscientists use it to check if counts of events fit a Poisson model before choosing a statistical method.

Good to know

The chi-square test assumes independent counts and a mean that is not too small, typically above 1 to 3. With few counts or a very small mean, use an exact test. The variance should be the sample variance with n − 1 in the denominator.

The Poisson distribution calculator gives probabilities under the random pattern, and the negative binomial calculator covers overdispersed counts.

Frequently asked questions

What is the index of dispersion?

It is the ratio of the variance to the mean of a set of counts. For a Poisson distribution, the variance equals the mean, so the ratio is 1. A ratio above 1 means overdispersion, as in clumped patterns, and a ratio below 1 means underdispersion, as in regular patterns.

How is the test done?

Under the Poisson model, (n − 1) times the index follows approximately a chi-square distribution with n − 1 degrees of freedom. A high statistic gives a small p-value for overdispersion, and a low statistic gives a small p-value for underdispersion. The calculator gives both and a two-sided value.

What do I do if the counts are overdispersed?

A Poisson model understates the variation. Common choices are the negative binomial distribution, which adds a dispersion parameter, or a quasi-Poisson model. In ecology, overdispersion often reflects clustering, and in epidemiology it can come from differences between groups.

What is the Fano factor?

It is the same ratio, used in physics and neuroscience for the variance of counts of events such as photons or spikes in a time window divided by their mean. A Fano factor of 1 corresponds to a Poisson process.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.