Multivariate Behrens–Fisher problem

In statistics, the multivariate Behrens–Fisher problem is the problem of testing for the equality of means from two multivariate normal distributions when the covariance matrices are unknown and possibly not equal. Since this is a generalization of the univariate Behrens-Fisher problem, it inherits all of the difficulties that arise in the univariate problem.

Notation and problem formulation

edit

Let   be independent random samples from two  -variate normal distributions with unknown mean vectors   and unknown dispersion matrices  . The index   refers to the first or second population, and the  th observation from the  th population is  .

The multivariate Behrens–Fisher problem is to test the null hypothesis   that the means are equal versus the alternative   of non-equality:

 

Define some statistics, which are used in the various attempts to solve the multivariate Behrens–Fisher problem, by

 

The sample means   and sum-of-squares matrices   are sufficient for the multivariate normal parameters  , so it suffices to perform inference be based on just these statistics. The distributions of   and   are independent and are, respectively, multivariate normal and Wishart:[1]

 

Background

edit

In the case where the dispersion matrices are equal, the distribution of the   statistic is known to be an F distribution under the null and a noncentral F-distribution under the alternative.[1]

The main problem is that when the true values of the dispersion matrix are unknown, then under the null hypothesis the probability of rejecting   via a   test depends on the unknown dispersion matrices.[1] In practice, this dependency harms inference when the dispersion matrices are far from each other or when the sample size is not large enough to estimate them accurately.[1]

Now, the mean vectors are independently and normally distributed,

 

but the sum   does not follow the Wishart distribution,[1] which makes inference more difficult.

Proposed solutions

edit

Proposed solutions are based on a few main strategies:[2][3]

Approaches using the T2 with approximate degrees of freedom

edit

Below,   indicates the trace operator.

Yao (1965)

edit

(as cited by [6])

 

where

 

Johansen (1980)

edit

(as cited by [6])

 

where

 

and

 

Nel and Van der Merwe's (1986)

edit

(as cited by [6])

 

where

 

Comments on performance

edit

Kim (1992) proposed a solution that is based on a variant of  . Although its power is high, the fact that it is not invariant makes it less attractive. Simulation studies by Subramaniam and Subramaniam (1973) show that the size of Yao's test is closer to the nominal level than that of James's. Christensen and Rencher (1997) performed numerical studies comparing several of these testing procedures and concluded that Kim and Nel and Van der Merwe's tests had the highest power. However, these two procedures are not invariant.

Krishnamoorthy and Yu (2004)

edit

Krishnamoorthy and Yu (2004) proposed a procedure which adjusts in Nel and Var der Merwe (1986)'s approximate df for the denominator of   under the null distribution to make it invariant. They show that the approximate degrees of freedom lies in the interval   to ensure that the degrees of freedom is not negative. They report numerical studies that indicate that their procedure is as powerful as Nel and Van der Merwe's test for smaller dimension, and more powerful for larger dimension. Overall, they claim that their procedure is the better than the invariant procedures of Yao (1965) and Johansen (1980). Therefore, Krishnamoorthy and Yu's (2004) procedure has the best known size and power as of 2004.

The test statistic   in Krishnmoorthy and Yu's procedure follows the distribution   where

 

References

edit
  1. ^ a b c d e Anderson, T. W. (2003). An Introduction to Multivariate Statistical Analysis (3rd ed.). Hoboken, N. J.: Wiley Interscience. p. 259. ISBN 0-471-36091-0.
  2. ^ Christensen, W. F.; A.C. Rencher (1997). "A comparison of type I error rates and power levels for seven solutions to the multivariate Behrens–Fisher problem". Communications in Statistics - Simulation and Computation. 26 (4): 1251–1273. doi:10.1080/03610919708813439.
  3. ^ a b Park, Junyong; Bimal Sinha (2007). Some aspects of multivariate Behrens–Fisher problem (PDF) (Technical report).
  4. ^ Olkin, Ingram; Jack L. Tomsky (1981). "A New Class of Multivariate Tests Based on the Union-Intersection Principle". The Annals of Statistics. 9 (4): 792–802. doi:10.1214/aos/1176345519.
  5. ^ Gamage, J.; T. Mathew; S. Weerahandi (2004). "Generalized p-values and generalized confidence regions for the multivariate Behrens--Fisher problem and MANOVA". Journal of Multivariate Analysis. 88: 177–189. doi:10.1016/s0047-259x(03)00065-4.
  6. ^ a b c Krishnamoorthy, K.; J. Yu (2004). "Modified Nel and Van der Merwe test for the multivariate Behrens-Fisher problem". Statistics and Probability Letters. 66 (2): 161–169. doi:10.1016/j.spl.2003.10.012.
  • Rodríguez-Cortés, F. J. and Nagar, D. K. (2007). Percentage points for testing equality of mean vectors. Journal of the Nigerian Mathematical Society, 26:85–95.
  • Gupta, A. K., Nagar, D. K., Mateu, J. and Rodríguez-Cortés, F. J. (2013). Percentage points of a test statistic useful in manova with structured covariance matrices. Journal of Applied Statistical Science, 20:29-41.