method of moments

E665237

The method of moments is a statistical technique for estimating distribution parameters by equating sample moments to theoretical moments.

Jump to: Statements Referenced by

Statements (48)

Predicate Object
instanceOf parameter estimation technique
statistical estimation method
advantage can be applied when likelihood is intractable
does not require full likelihood specification
appliesTo parametric statistical models
probability distributions
approximateDate late 19th century
basedOn equating sample moments to theoretical moments
canEstimate mean parameter
scale parameters
shape parameters
variance parameter
exampleUse estimating parameters of the Poisson distribution
estimating parameters of the binomial distribution
estimating parameters of the gamma distribution
estimating parameters of the normal distribution
field mathematical statistics
statistics
generalization generalized method of moments
historicalOrigin introduced by Karl Pearson NERFINISHED
limitation higher-order moments can be unstable in finite samples
moment equations may have multiple solutions
moment equations may have no real solution
output parameter estimates
property can yield biased estimators
estimators are consistent under regularity conditions
may be less efficient than maximum likelihood estimates
often easier to compute than maximum likelihood estimates
solves system of equations formed by matching moments
relatedTo generalized method of moments
least squares estimation
maximum likelihood estimation
requires existence of required population moments
step compute sample moments from data
express theoretical moments as functions of parameters
set sample moments equal to theoretical moments
solve resulting equations for parameters
typicalInput assumed parametric family of distributions
random sample
usedFor estimating distribution parameters
usedIn actuarial science
applied probability
econometrics
engineering
usesConcept moment of a random variable
population moments
sample moments
theoretical moments

Referenced by (1)

Full triples — surface form annotated when it differs from this entity's canonical label.

Karl Pearson notableWork method of moments