Main Page | See live article | Alphabetical index

Likelihood

In statistics, a likelihood function is a conditional probability function considered a function of its second argument with its first agument held fixed, thus:

and also any other function proportional to such a function. That is, the likelihood function for B is the equivalence class of functions

for any constant of proportionality α > 0. Thus the numerical value L(b) is immaterial; all that matters are ratios of the form L(b2)/L(b1), since these are invariant with respect to the constant of proportionality.

Likelihood as a solitary term is a shorthand for likelihood function.

In a sense, likelihood works backwards from probability: given B, we use the conditional probability P(A | B) to reason about A, and, given A, we use the likelihood function P(A | B) to reason about B. This mode of reasoning is formalized in Bayes' theorem; note the appearance of a likelihood function for B given A in the numerator:

For more about making inferences via likelihood functions, see also the method of maximum likelihood, and likelihood-ratio testing.

In the colloquial language, "likelihood" is one of several informal synomyms for "probability", but throughout this article we use only the technical definition.

Likelihood function of a parametrized model

Among many applications, we consider here one of broad theoretical and practical importance. Given a parametrized family of probability density functions

where θ is the parameter (in the case of discrete distributions, the probability density functions are probability "mass" functions) the likelihood function is

where x is the observed outcome of an experiment. In other words, when f(x | θ) is viewed as a function of x with θ fixed, it is a probability density function, and when viewed as a function of θ with x fixed, it is a likelihood function.

Note: This is not the same as the probability that those parameters are the right ones, given the observed sample. Attempting to interpret the likelihood of a hypothesis given observed evidence as the probability of the hypothesis is a common error, with potentially disastrous real-world consequences in medicine, engineering or jurisprudence. See prosecutor's fallacy for an example of this.

Example

For example, if I toss a coin, with a probability pH of landing heads up ('H'), the probability of getting two heads in two trials ('HH') is pH2. If pH = 0.5, then the probability of seeing two heads is 0.25.

In symbols, we can say the above as

Another way of saying this is to reverse it and say that "the likelihood of pH = 0.5 given the observation 'HH' is 0.25", i.e.,

.

But this is not the same as saying that the probability of pH = 0.5 given the observation is 0.25.

To take an extreme case, on this basis we can say "the likelihood of pH = 1 given the observation 'HH' is 1". But it is clearly not the case that the probability of pH = 1 given the observation is 1: the event 'HH' can occur for any pH > 0 (and often does, in reality, for pH roughly 0.5).

The likelihood function does not in general follow all the axioms of probability: for example, the integral of a likelihood function is not in general 1.

This is because integration of the likelihood density function L is performed over all possible values of the model parameters (in this case, pH), while integration of a probability density function f is performed over the random variables (which in this case take on the four pairs of values 'TT', 'TH', 'HT' and 'HH').

In this example, the integral of the likelihood density over the interval [0, 1] in pH is 1/3, demonstrating again that the likelihood density function cannot be interpreted as a probability density function for pH.

On the other hand, given any particular value of pH, e.g. pH=0.5, the integral of the probability density function over the domain of the random variables is 1.

See also: