Monotone likelihood ratio
that is, if the ratio is nondecreasing in the argument .
If the functions are first-differentiable, the property may sometimes be stated
For two distributions that satisfy the definition with respect to some argument x, we say they "have the MLRP in x." For a family of distributions that all satisfy the definition with respect to some statistic T(X), we say they "have the MLR in T(X)."
- 1 Intuition
- 2 Families of distributions satisfying MLR
- 3 Relation to other statistical properties
- 4 Uses
- 5 References
The MLRP is used to represent a data-generating process that enjoys a straightforward relationship between the magnitude of some observed variable and the distribution it draws from. If satisfies the MLRP with respect to , the higher the observed value , the more likely it was drawn from distribution rather than . As usual for monotonic relationships, the likelihood ratio's monotonicity comes in handy in statistics, particularly when using maximum-likelihood estimation. Also, distribution families with MLR have a number of well-behaved stochastic properties, such as first-order stochastic dominance and increasing hazard ratios. Unfortunately, as is also usual, the strength of this assumption comes at the price of realism. Many processes in the world do not exhibit a monotonic correspondence between input and output.
Example: Working hard or slacking off
Suppose you are working on a project, and you can either work hard or slack off. Call your choice of effort and the quality of the resulting project . If the MLRP holds for the distribution of q conditional on your effort , the higher the quality the more likely you worked hard. Conversely, the lower the quality the more likely you slacked off.
- Choose effort where H means high, L means low
- Observe drawn from . By Bayes' law with a uniform prior,
- Suppose satisfies the MLRP. Rearranging, the probability the worker worked hard is
- which, thanks to the MLRP, is monotonically increasing in (because is decreasing in ). Hence if some employer is doing a "performance review" he can infer his employee's behavior from the merits of his work.
Families of distributions satisfying MLR
Statistical models often assume that data are generated by a distribution from some family of distributions and seek to determine that distribution. This task is simplified if the family has the monotone likelihood ratio property (MLRP).
A family of density functions indexed by a parameter taking values in an ordered set is said to have a monotone likelihood ratio (MLR) in the statistic if for any ,
- is a non-decreasing function of .
Then we say the family of distributions "has MLR in ".
List of families
|Family||in which has the MLR|
|Normal||if known, observations|
If the family of random variables has the MLRP in , a uniformly most powerful test can easily be determined for the hypothesis versus .
Example: Effort and output
Example: Let be an input into a stochastic technology – worker's effort, for instance – and its output, the likelihood of which is described by a probability density function Then the monotone likelihood ratio property (MLRP) of the family is expressed as follows: for any , the fact that implies that the ratio is increasing in .
Relation to other statistical properties
has a monotone non-decreasing likelihood ratio in the sufficient statistic T(x), provided that is non-decreasing.
Most powerful tests: The Karlin–Rubin theorem
Monotone likelihood functions are used to construct uniformly most powerful tests, according to the Karlin–Rubin theorem. Consider a scalar measurement having a probability density function parameterized by a scalar parameter θ, and define the likelihood ratio . If is monotone non-decreasing, in , for any pair (meaning that the greater is, the more likely is), then the threshold test:
- where is chosen so that
is the UMP test of size α for testing
Note that exactly the same test is also UMP for testing
Median unbiased estimation
Monotone likelihood-functions are used to construct median-unbiased estimators, using methods specified by Johann Pfanzagl and others. One such procedure is an analogue of the Rao–Blackwell procedure for mean-unbiased estimators: The procedure holds for a smaller class of probability distributions than does the Rao–Blackwell procedure for mean-unbiased estimation but for a larger class of loss functions.
Lifetime analysis: Survival analysis and reliability
If a family of distributions has the monotone likelihood ratio property in ,
- the family has monotone decreasing hazard rates in (but not necessarily in )
- the family exhibits the first-order (and hence second-order) stochastic dominance in , and the best Bayesian update of is increasing in .
But not conversely: neither monotone hazard rates nor stochastic dominance imply the MLRP.
Let distribution family satisfy MLR in x, so that for and :
Integrating this expression twice, we obtain:
|1. To with respect to
integrate and rearrange to obtain
|2. From with respect to
integrate and rearrange to obtain
First-order stochastic dominance
Combine the two inequalities above to get first-order dominance:
Monotone hazard rate
Use only the second inequality above to get a monotone hazard rate:
The MLR is an important condition on the type distribution of agents in mechanism design. Most solutions to mechanism design models assume a type distribution to satisfy the MLR to take advantage of a common solution method.
- Casella, G.; Berger, R.L. (2008), Statistical Inference, Brooks/Cole. ISBN 0-495-39187-5 (Theorem 8.3.17)
- Pfanzagl, Johann. "On optimal median unbiased estimators in the presence of nuisance parameters." The Annals of Statistics (1979): 187–193.
- Brown, L. D.; Cohen, Arthur; Strawderman, W. E. A Complete Class Theorem for Strict Monotone Likelihood Ratio With Applications. Ann. Statist. 4 (1976), no. 4, 712–722. doi:10.1214/aos/1176343543. http://projecteuclid.org/euclid.aos/1176343543.
- Page 713: Brown, L. D.; Cohen, Arthur; Strawderman, W. E. A Complete Class Theorem for Strict Monotone Likelihood Ratio With Applications. Ann. Statist. 4 (1976), no. 4, 712–722. doi:10.1214/aos/1176343543. http://projecteuclid.org/euclid.aos/1176343543.