Professional Documents
Culture Documents
To cite this article: Peter J. Rousseeuw & Christophe Croux (1993) Alternatives to the Median Absolute Deviation, Journal of
the American Statistical Association, 88:424, 1273-1283
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of
the Content. Any opinions and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of information. Taylor and Francis shall not be
liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities
whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of
the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Alternatives to the Median Absolute Deviation
Peter J. ROUSSEEUWand Christophe CROUX*
~~~ ~ ~~ ~
In robust estimation one frequently needs an initial or auxiliary estimate of scale. For this one usually takes the median abso-
lute deviation MAD, = 1.4826 med, { I xi - med,x, I } , because it has a simple explicit formula, needs little computation time, and
is very robust as witnessed by its bounded influence function and its 50% breakdown point. But there is still room for improve-
ment in two areas: the fact that MAD, is aimed at symmetric distributions and its low (37%) Gaussian efficiency. In this article
we set out to construct explicit and 50% breakdown scale estimators that are more efficient. We consider the estimator S,
= 1.1926 medi {med, I xi - xjl } and the estimator Qn given by the .25 quantile of the distances { 1 x, - x, I ; i < j } . Note that S, and
Q, do not need any location estimate. Both S. and Qn can be computed using O(n log n) time and O ( n ) storage. The Gaussian
efficiency of S. is 58%, whereas Q. attains 82%. We study S, and Qnby means of their influence functions, their bias curves (for
implosion as well as explosion), and their finite-sample performance. Their behavior is also compared at non-Gaussian models,
including the negative exponential model where S, has a lower gross-error sensitivity than the MAD.
KEY WORDS: Bias curve; Breakdown point; Influence function; Robustness; Scale estimation.
1. INTRODUCTION The MAD has the best possible breakdown point (50%,twice
as much as the interquartile range), and its influence function
Although many robust estimators of location exist, the
is bounded, with the sharpest possible bound among all scale
Downloaded by [Carnegie Mellon University] at 11:22 17 January 2015
terval (around the median) that contains 50% of the data (or Table 1. Average Estimated Value of MAD,, Sn, On,
50% of the probability), which does not seem to be a natural and SO, at Gaussian Data
approach at asymmetric distributions. The interquartile Average estimated value
range does not have this problem, as the quartiles need not
be equally far away from the center. In fact, Huber (1 98 1 , n MAD, Sn On Son
-
p. 1 14) presented the MAD as the symmetrized version of 10 ,911 .992 1.392 .973
the interquartile range. This implicit reliance on symmetry 20 .959 ,999 1.193 ,990
is at odds with the general theory of M-estimators, in which 40 .978 .999 1.093 ,993
60 .987 1.001 1.064 .996
the symmetry assumption is not needed. Of course, there is 80 ,991 1.002 1.048 .997
nothing to stop us from using the MAD at highly skewed 100 .992 .997 1.038 .997
distributions, but it may be rather inefficient and artificial 200 ,996 1.ooo 1.019 ,999
03 1.ooo 1.000 1.ooo 1.ooo
to do so.
NOTE: Based on 10,oOO Sam@eS foC each n.
2. THE ESTIMATOR S,
The purpose of this article is to search for alternatives to
the MAD that can be used as initial or ancillary scale esti- (rousse@wins.uia.ac.be),and it has been incorporated in
mates in the same way but that are more efficient and not Statistical Calculator (T. Dusoir, fbgj23@ujvax.ulster.ac.uk)
slanted towards symmetric distributions. Two such estima- and in Statlib (statlib@stat.cmu.edu).
Downloaded by [Carnegie Mellon University] at 11:22 17 January 2015
tors will be proposed and investigated; the first is To check whether the correction factor c = 1.1926 (ob-
S,,=cmedi{medjIxi -xjl}. (2.1) tained through an asymptotic argument) succeeds in making
S,, approximately unbiased for finite samples, we performed
This should be read as follows. For each i we compute the a modest simulation study. The estimators in Table 1 are
median of { I xi - x, I ; j = 1 , . . . ,n } . This yields n numbers, the MAD,, S,,, the estimator Qn(which will be described in
the median of which gives our final estimate S,,. (The factor Section 3), and the sample standard deviation SD,,. Each
c is again for consistency, and its default value is 1.1926, as table entry is the average scale estimate on 10,000 batches
we will see later.) In our implementation of (2.1) the outer of Gaussian observations. We see that S,, behaves better than
median is a low median, which is the order statistic of rank MAD,,in this experiment. Moreover, Croux and Rousseeuw
[ (n+ 1)/2], and the inner median is a high median, which (1 992b) have derived finite-sample correction factors that
+
is the order statistic of rank h = [n/2] 1. The estimator render MAD,,, S,,, and Q,, almost exactly unbiased.
S,, can be seen as an analog of Gini’s average difference (Gini In the following theorem we prove that the finite sample
19 12), which one would obtain when replacing the medians breakdown point of S,, is the highest possible. We use the
by averages in (2.1). Note that (2.1) is very similar to formula replacement version of the breakdown point: For any sample
(1.2) for the MAD, the only difference being that the med, X = { x, , . . . ,x,,} , the breakdown point of S,, is defined by
operation was moved outside the absolute value. The idea
to apply medians repeatedly was introduced by Tukey (1977) &n*(sn,
X ) = min{ E n + ( s n , X I , & i ( S n X
, ) } , (2.2)
for estimation in two-way tables and by Siege1 (1982) for where
estimating regression coefficients. Rousseeuw and Bassett
( 1990) used recursive medians for estimating location in large en+(&, X ) = min
data sets.
Note that (2.1) is an explicit formula; hence S,, is always
uniquely defined. We also see immediately that S,, does be-
have like a scale estimator, in the sense that transforming
+
the observations xi to axi b will multiply S,, by 1 a I. We
and
:{
c,(S,,, X ) = min - ;inf S,,(X’)= 0
X‘
tribution function G. If we denote g G ( x )= medy 1 x - Y 1, (A more general “chain rule” for influence functions was
then given by Rousseeuw and Croux 1992.) Let us consider the
special case where F = a. We will use the notation q
S ( G ) = c med g G ( X ) (2.3) = W 1 ( 3 / 4 ) .From the proof of Theorem 2, it follows that
X
41 = -4, q 2 = q, and g k ( q ) = - g k ( - q ) . Because we have
is the desired functional. If Gnis the empirical distribution
function, then S( Gn)= Sn. @ ( x+ g d x ) ) - + ( x - g d x ) ) = .5
Let us consider a location-scale model F8,*(x ) = F( ( x - O ) / for all x , it follows that
a), where F is called the model distribution. We want our
estimator to be Fisher-consistent, which means that S( F8,J 4(q - c-l) - 4(q + c - l )
gk(q) =
= d for all 8 and all u > 0. It is easily verified that S is Fisher- +
4 ( q c-l) 4 ( q - c-1) .+
consistent if we take
This yields the expression
c = I/(med g ~ ( J 3 ) , (2.4)
X
c-l))
Downloaded by [Carnegie Mellon University] at 11:22 17 January 2015
[
w x ; s, F ) = c F(q2) - F(q1) - Z(q1 < x < q 2 )
- c-l)
- f(q2)sgn(Iq2 -XI
2(f(q2 + C - I ) -A42 - c-l))
+ f(4r)sgn(lq, - X I - C - I )
2(f(q1 + c-l) -f(a - c - l ) ) I ’ -4 -2 0 2
for this by a slight increase in the gross-error sensitivity and at the breakdown point ( c + 4) we find B + (E, @) --* co and
in the required computation time. B - ( c , @) + 0.
We carried out a simulation to verify this efficiency gain
at finite samples. For each n in Table 2, we computed the
variance var,(S,,) of the scale estimator S,,over m = 10,000
samples. Table 2 lists the standardized variances
n var, ( sn)
/ (ave,( Sn) ) 2 , (2.10)
where ave,(S,,) is the average estimated value as given in
Table 1. (It was argued by Bickel and Lehmann (1976) that
the denominator of (2.10) is needed to obtain a natural mea-
sure of accuracy for scale estimators.) The results show that
the asymptotic variance provides a good approximation for
(not too small) finite samples, and that S,, is more efficient
than MAD, even for small n .
Whereas the influence function describes how the esti-
mator reacts to a single outlier, the bias curve tells us how
much the estimator can change (in the worst case) when a
fraction e of the data is contaminated. Bias curves were briefly
mentioned by Hampel et al. (1986, p. 177), but their full
potential was not realized until the work of Martin and Za- 0.0 0.1 0.2 0.3 0.4 0.5
mar (1989, 1991), Martin, Yohai and Zamar (1989), and epsi l on
He and Simpson (1993). (4
In the case of scale estimators, we distinguish between an
increase of the estimate (explosion) and a decrease (implo-
sion). Fore > 0, we define 3== { G; G = (1 - E ) F +c H } ,
where H ranges over all distributions. Then the explosion
bias curve of S is defined as
B + ( E F)
, = SUP S(G)
CE3.
Note that the influence function and the bias curve are We first show that the finite sample breakdown point of
asymptotic concepts. For finite samples, we have confirmed Q,, attains the optimal value.
the above results by computing the correspondingsensitivity Theorem 5 . At any sample X = { x l , . . . ,x,,} in which no
curves and empirical bias curves. two points coincide, we have &,+(en, X) = [(n + 1)/2]/n
3. THE ESTIMATOR Q,
and e;( Q,,,X ) = [ n / 2 ] / n ;thus the breakdown point of Q,,
is
A drawback of MAD,, and S,, is that their influence func-
cn*(Qn, X ) = [n/21/n.
tions have discontinuities. In the location framework, the
sample median has the same drawback, unlike the Hodges- We now give the asymptotic version of our estimator. Let
Lehmann (1963) estimator X and Y be independent random variables with distribution
G . Then we define
(3.1)
Q ( G ) = d H ; ' ( t ) with HG = f ( I X - Y l ) . (3.4)
which possesses a smooth influence function (see, for ex-
Note that Q ( G n )is not exactly the same as Q,, (where we
ample, Hampel et al. 1986, p. 1 12). Therefore, the Hodges-
take an order statistic among (2") elements instead of n 2 ) ,
Lehmann estimator might be viewed as a "smooth version"
but asymptotically this makes no difference. Representation
Downloaded by [Carnegie Mellon University] at 11:22 17 January 2015
of the median.
(3.4)belongs to the class of generalized L-statistics, as intro-
An analogous scale estimator, mentioned by Shamos duced by Serfling ( 1 984). Because X - Y is symmetric, we
(1976, p. 260) and Bickel and Lehmann (1979, p. 38), is could also write
obtained by replacing the pairwise averages by pairwise dis-
tances, yielding
Q ( G ) = d K E ' ( i ) with KG = L ( X - Y ) (3.5)
med{ ! x i - x i \ ; i < J } . (3.2)
instead of (3.4). Finally, we obtain
This resembles definition (2.1) of S,,, where separate medians
were taken over i and j , whereas (3.2) uses an overall median
over (1) pairs. We propose to multiply (3.2) by 1.0483 to
achieve consistency for the parameter u of Gaussian distri-
1 s
Q ( G ) = inf s > 0;
I
G ( t + d - ' s ) d G ( t ) 2 5 / 8 . (3.6)
In Figure 1 we plotted the function ZF(x; Q, a). It turns Table 3. Simulation Results for Cauchy Data,
out that the gross-error sensitivity y*( Q, a) = 2.069 is larger Based on 70,000 Samples for Each n
than that of the MAD and S. Note that this influence func- Average value Standardized variance
tion is unbalanced: The influence of an inlier at the center
of the distribution is smaller (in absolute value) than the n MAD. S" Qn MAD, S" Qn
-3 -1 1
which generates the location-scale family
Figure 3. Influence Functions of the MAD, S, and Q at the Cauchy Dis-
tribution. Fe,Ax) = (1 - exp(-(x - e)/u))Z(x 2 e),
Rousseeuw and Croux: Alternatives to the MAD 1279
which we will call the shifted exponential model. Note that Table 4. Simulation Results for Exponentially Distributed Data,
Based on 70,000Samples for Each n
the location parameter 8 is not at the center of the distri- ~~~ ~ ~
bution, but rather at its left boundary. To estimate u by Average value Standardized variance
MAD,, S,, or Q,, we need to set b = 2.0781,c = 1.6982,
and d = 3.4760.Figure 4 displays the influence functions of n MA0, S” 0” MAD, S” Q”
~~ ~
these estimators at F. Note that the influence functions were 10 0.925 1.002 1.536 2.152 2.121 1.677
also computed at negative x although F has no mass there, 20 0.959 0.981 1.246 2.182 2.065 1.521
40 0.983 0.991 1.120 2.167 2.006 1.481
because outliers can occur to the left of the unknown 19. 60 0.986 0.993 1.078 2.172 1.989 1.420
The influence function of the MAD looks very different 80 0.988 0.993 1.056 2.153 1.924 1.395
from those in Figures 1 and 3, because this time F is not 100 0.993 0.996 1.046 2.122 1.911 1.373
200 0.996 0.998 1.024 2.173 1.919 1.381
symmetric. Note that ZF(x; MAD, F) = 0 for negative x a, 1 .ooo 1 .ooo 1 .ooo 2.135 1.822 1.343
(due to the compensating effect of the outlier at x on the
location median) and that it has three jumps for positive x
+
(at ln(2) - b-I, at ln(2), and at ln(2) b-I). The influence
functions of S and Q are roughly U-shaped, as in Figures 1 V ( 5,F) = I . But the extreme efficiency of 8, is offset by its
and 3.Surprisingly, the gross-error sensitivity of S is smaller extreme sensitivity to any outlier xi < 8. One possibility for
than that of the MAD. Indeed, constructing a more robust estimator of 8 would be to use
Downloaded by [Carnegie Mellon University] at 11:22 17 January 2015
Theorem 8. At every data set { xI,. . . ,x,,} ,it holds that 6. EXTENSIONS AND OUTLOOK
is not recommended.
MAD,,, both asymptotically and for finite samples, while If we also allow scale estimators that measure dispersion
their gross-error sensitivities and bias curves are almost as around a location estimate, possible alternatives would be
good. Another advantage is that they do not presuppose a
symmetric model distribution, but can be considered as
nonparametric measures of spread. (The estimator LMS,
shares this advantage, but is less efficient.) The only price
we pay is a small increase of the computation time, which and
is O( n log n ) for S,, and Q,,as compared to O ( n )for MAD,,
but today's computers can easily afford this.
One could also compare S,, and Q,, with the class of M-
estimators. The latter can be more efficient, and their com- which also have a 50% breakdown point. At symmetric dis-
putational complexity is similar (or even lower, if we restrict tributions, (6.3) and (6.4) are asymptotically equivalent to
their algorithm to a fixed number of iteration steps). But M- S,, and Q,,;however, simulations have indicated that at finite
estimators are defined implicitly, whereas S,, and Q,, have samples these estimators do not perform better than S,, and
an explicit formula, which guarantees uniqueness of the es- Q,,, so we see no reason to switch to either (6.3) or (6.4). A
timate and is more intuitive. Moreover, there is no need to possible application of (6.3) would be to divide it by S,, to
select a function x or to choose tuning constants. obtain a test for symmetry (analogously, (6.4) can be divided
Another class is that of one-step M-estimators, which have by Q,,). This is similar to a proposal of Boos ( 1982) using
an explicit formula and can also attain a high efficiency and Gini's average deviation.
a high breakdown point. But like fully iterated M-estimators, Recently, Rousseeuw and Croux (1 992) proposed several
they are not location-free because they need an ancillary other explicit scale estimators with high breakdown point.
location estimator. And, also like fully iterated M-estimators, A promising estimator is
their computation requires an initial high-breakdown scale
estimator such as MAD,, S,,, or Qnin the first place. Finally,
it should be noted that although k-step M-estimators inherit
the breakdown point of the initial estimator, it turns out that
their actual contamination bias does increase (Rousseeuw It was proved that T,, has a 50% breakdown point, a contin-
and Croux 1993). uous influence function, an efficiency of 52%, and a gross-
Choosing between S,, and Qncomes down to a tradeoff error sensitivity of 1.4688.
between advantages that are hard to compare. Although Qn In cluster analysis one often needs a measure of dissimi-
is more efficient, in most applications we would prefer S,, larity between two groups A and B, which is based on the
because it is very robust, as witnessed by its low gross-error interobject dissimilarities. (For a discussion of such measures
sensitivity. Another advantage of S,, is that its simplicity see Kaufman and Rousseeuw 1990, sec. 5.5.) Our estimators
makes it easier to compute. S,, and Qncan be extended to this situation, yielding
We expect the application potential of the new robust scale
&(A, B ) = media{medjd(i,j)} (6.6)
estimators to be mainly in the following three uses: (1) as a
data analytic tool by itselc (2) as an ancillary scale estimate and
to make M-estimators affine equivariant and as a starting
~ Q ( AB, ) = { d ( i , J ) ; i E A , j E B}(k), (6.7)
value for the iterative computation of M-estimators; and (3)
as an objective function for regression analysis, as will be where k is approximately (#A)( #B)/4. (Both measures are
described in the next section. equivariant for monotone transformations on the d( i, j ) ;
Rousseeuw and Croux: Alternatives to the MAD 1281
hence they can still be applied when the dissimilarities are = c lomed,himed,Ix, - x,l. We first show that E ; I [n/2]/n.
on an ordinal scale.) Note that dQis symmetric in the sense Construct a contaminated sample X'by replacing the observations
that dQ(A,B ) = dQ(B,A ) , whereas ds is not, but this can xz, . . . , x[,,,~~+~ by xi. Then we have himed,Ix: - x; I = 0
be repaired by working with dk(A, B ) = min{ d s ( A , B ) , for [ n/2] + 1 observations x i , hence S,,(X') = 0. Moreover, ti
> [n/2]/n. Indeed, take any sample X'where fewer than [n/2]
dS(B,A)}.
observations are replaced. Because X was in general position,
Both dsand dQcan be seen as robustifying existing meth-
himed,I x: - x; I 2 min,,,I x, - x,1/2 = 6 > 0 for all i; hence
ods. Indeed, if the medians in (6.6) were replaced by averages,
S,,(X') > cd.
or if the order statistic in (6.7) was replaced by an average, Further, E ; I [( n + 1)/2]/n. Construct a sample X'by replac-
then we would recover the well-known average linkage cri- ing X I by -q,,)+ L , xz by x(") + 2 L , . . . , and XI(,,+I)/ZI by x(,,)
terion. On the other hand, replacing the medians by minima + +
[(n 1)/2]L, with L > 0. Then himed,)~:- x(l] 2 L for all i .
(or equivalently, setting k in (6.7) equal to 1) produces the Letting L tend to infinity will then inflate the estimator S,, be-
single linkage criterion. Moreover, replacing the medians by +
yond all bounds. On the other hand, t: > [(n 1)/2]/n. Take any
maxima (or setting k = ( # A ) ( # B )in (6.7)) yields complete +
sample X ' where fewer than [ ( n 1)/2] observations are replaced.
linkage. We can thus think of d's and dQ as intermediate If X: belongs to the original sample, then himed,I xi - XI I
between single linkage and complete linkage, which are often 5 I x(,,)- x ( ~I. )Because this holds for at least half the points, we
considered to be too extreme. For instance, the latter criteria find t h a t S , ( X ' ) I c l ~ ( , ) - x ( ~ , <
l CQ.
do not have an asymptotic limit (because usually the smallest Remark. This proof also implies that if we replace the outer
dissimilarity tends to 0 as n + 00, and the largest dissimilarity median by an order statistic of rank k 5 h = [ n/2] +
1 , then the
Downloaded by [Carnegie Mellon University] at 11:22 17 January 2015
tends to a),whereas (6.6) and (6.7) do. estimator will keep the same breakdown points.
Another application is to regression. The approach put The fact that eE(S,,, X ) 5 [n/2]/n for any affine equivariant
forward in (Rousseeuw 1984) was to construct affine equi- scale estimator was proved by Croux and Rousseeuw (1 992a).
variant high-breakdown regression estimates by minimizing Proof of Theorem 3. We use the notations
a high-breakdown scale estimate of the residuals rl , . . . , r,,.
gl = ((gF))\]-m,x~])-'and g2 = ((gF)lh,co[)-'
For instance, minimizing the scale estimator (5.1) yields the
LMS regression estimator. In the same vein, S-estimators Condition I ensures that these functions are well defined. With
(Rousseeuw and Yohai 1984) were obtained by minimization G F ( u )= P ( g F ( Y )i u ) , we have S ( F ) = cGF'(1/2) = 1, using
of an M-estimator of scale based on a smooth function p . +
(2.3).Let us denote F. = ( 1 - e ) F &Ax.Differentiating the expres-
The high-breakdown scale estimators introduced in this ar- sion GFc(CIS(F,)) = 1 / 2 yields
ticle could be used in the same way. It seems that Q,, is - acFc(c-')l
preferable to S,, in this regard, because of its smooth influence a& ILO
function. Therefore, we may consider the regression esti- . (A.1)
G;( c-I )
mator
Using Conditions 1 and 2 , we obtain for u in a neighborhood of
8, = argmin8Q,,(rl, . . . , rn). (6.8) c-' that
We will refer to 8, as the least quartile difference (LQD) G F ( u )= P ( g i ( u ) I Y s g z ( u ) ) = F ( g z ( u ) )- F(gi(u)-).
estimator. The LQD belongs to the larger class of generalized
Condition 2 yields g,(c-') = q1 and g 2 ( c - ' ) = qz;hence
S-estimators (Croux, Rousseeuw, and Hossjer 1993). The
asymptotic efficiency of the LQD turns out to be 67.170,
which is much higher than that of S-estimators.
A final extension is to scale estimation in a linear model. Condition 3 allows us to assume that x is different from the points
Existing estimators of the error dispersion are based on the ql ? c-' and q2 f c-' (in which the influence function will not be
residuals from some regression fit. But it is possible to gen- defined). For E sufficiently small, we may write
eralize the location-free estimators s,,and Q,, to scale esti- GF,(U) = Fr(gZ.r(u)) - Fc(gl,c(u)) ('4.3)
mators that are regressionfree, in the sense that they do not for a11 u in a small neighborhood of c-I . We choose a neighborhood
depend on any previous estimate of the regression parame- such that gFc stays decreasing in a neighborhood of qi and increasing
ters. In the simple linear model, we may consider triangles near qz. We may also assume that gl,c(c-')and gZ,Jc-l) are lying
formed by data points and compute their vertical height in that neighborhood. Differentiation of (A.3) yields
h( i,j, k ) . This leads to several scale estimators (Rousseeuw
and Hubert 1993), obtained from quantiles or repeated me-
dians ofthe h( i , j , k ) .When attention is restricted to adjacent
triangles, this also yields a test for linearity.
APPENDIX: PROOFS
Evaluating in C - I gives
Due to lack of space, some proofs and computations have been
omitted and referred to a technical report (Rousseeuw and Croux
1991).
Proof of Theorem 1. Let X be a sample in general position
and denote E: = E:(S,,, X ) and E ; = E;(S,,, X ) , where S,,
1282 Journal of the American Statistical Association, December 1993
By definition of gFc,we have When x < gG,(X), Equation (A.9) entails g C o ( x )= g - ( x ) . There-
fore, &,,( x) is symmetric and increasing on the positive numbers.
(1 - c){F(c-l + g2.Jc-I)) - F(gz.,(c-’) - c-’)} This implies that
+ E { A ~ ( c -+’ & . , ( c - ’ ) )- AX(g2,.(c-’)- c-’)} = 1/2.
B - ( E ,F ) = S ( G o )= cgGo(medlY I )
Differentiating yields
Combining (A.I), (A.2), (A.4), (A.5), and (A.6) gives the desired to F and YI , Y2 according to G, then we can rewrite (A. I 1) as
equation (2.6).
(I - E ) 2 ~ * 2 ( d - l ~ ( ~ ) )
Proof of Theorem 4. Let E be any value in 10, [. We obtain 4 + E ( I - E ) { I +P(IX-Y1I s ~ - ’ Q ( G ) ) } + E ~ P ( ( Y I - Y ~ )
B + ( c ,F) as the limit of S ( G , ) , where G, = (1 - c ) F &Ax,and +
x, goes to infinity. Let Y be distributed according to G,. Then I d - l Q ( G ) ) z 5/8. (A.12)
g c , ( x ) is the smallest positive value for which Note that each term in (A. 12) is increasing in Q( C). To maximize
G , ( x + g c , ( x ) ) - G,(x - g c , ( x ) ) + P(Y = x - g c , ( x ) ) 2 .5. Q ( G ) , we have to minimize P ( I X - YII I d - l Q ( G ) ) and
P(( Yl - Y2)5 d-IQ( G)). These terms approach their lower bounds
Substituting G, yields (0 and 4) when both location and scale of H tend to infinity. (For
( I - c ) { F ( x+ g c . ( x ) ) - F ( x - g c ” ( x ) ) } instance, consider a sequence of Gaussian distributions with location
n and standard deviation n .) Therefore, (A. 12)yields formula (3.10).
+ & { Axn(x+ gG,(X)) - Axn(x- gC,(x))1 For the implosion bias curve, we have to maximize P( I X - Yl I
+ EI(X - g c , ( x ) = x,) 2 .5. 5 d - I Q ( G ) ) and P ( ( Y l - Y,) I d - ’ Q ( G ) ) .When we choose H
= &, we attain the maximal values 2F(Q(G)) - 1 and 1. Com-
Because E < .5, for each M > 0 we can choose no such that for all
bining this with (A.12), we obtain Equation (3.1 I ) , which deter-
n > no
mines B - ( E ,F).
SUP ( x + g c , ( x ) ) < xn (‘4.7)
IxlcM
[Received September 1991. Revised June 1992.1
and
inf gC,(x) 2 SUP gG,(x), (A.8) REFERENCES
IxlrM Ixl<M
making use of the properties of F. Using (A.7) yields g + ( x ) Andrews, D. F., Bickel, P. J., Hampel, F. R., Huber, P. J., Rogers, W. H.,
and Tukey, J. W. (1972). Robust Estimates of Location: Survey and Ad-
= g G , ( x )for all I X I < M and for all n > 4. If we choose M such vances, Princeton, NJ: Princeton University Press.
that Bickel, P. J., and Lehmann, E. L. (1976), “Descriptive Statistics for Non-
parametric Models 111: Dispersion,” The Annals of Statistics, 4, 1139-
1158.
(1979). “Descriptive Statistics for Nonparametric Models IV:
Spread,” in Contributions to Statistics, Hiijek Memorial Volume, ed. J.
then, using (A.8) and the symmetry and monotonicity of &.(x)
JureEkovB, Prague: Academia, pp. 33-40.
on 1-M, M [ (from Eq. (2.12)), we obtain Boos, D. (1982), “A Test for Asymmetry Associated With the Hodges-
Lehmann Estimator,” Journal of the American Statistical Association,
S ( G , ) = cgGJmedlYI), 77,647-65 1.
Croux, C., and Rousseeuw, P. J. (1992a), “A Class of High-Breakdown
which does not depend on n (provided that n > a).Therefore,
Scale Estimators Based on Subranges,” Communications in Statistics,
Equations (2.1 1) and (2.12) together determine It+(&,F). Theory andMethods, 21, 1935-1951.
For the implosion bias curve, we have B - ( E ,F ) = S(Go),with (1992b), “Time-Efficient Algorithms for Two Highly Robust Esti-
Go = ( 1 - E ) F+ E&. Now g c o ( x )is the smallest positive solution mators of Scale,” in Computational Statistics, Volume 1, eds. Y . Dodge
of and J. Whittaker, Heidelberg: Physika-Verlag, 41 1-428.
Croux, C., Rousseeuw, P. J., and Hossjer, 0. (1993), “Generalized S-
(1 - & ) { F ( X+ g c 0 ( x ) )- F ( x - g c o ( x ) ) } Estimators,” technical report 93-03, Universitaire Instelling Antwerpen,
Belgium.
+ & { & ( x + gCo(x)) - & ( x - gGo(x))} Gini, C. (1912), “Variabiliti 6 mutabilita, contributo allo studio delle dis-
+ EI(x - gGo(x)= 0 ) 2 .5. (7.9) tribuzioni e delle relazioni statistiche,” Studi Economico-Giuridici della
R. Universitd di Cagliari, 3, 3- 159.
It follows that x 5 g G o ( x and
) that Griibel, R. (1988), “The Length of the Shorth,” The Annals of Statistics,
16, 619-628.
x = gGo(x) iffx 2 F-I
(-/2. Hampel, F. R. (1974), “The Influence Curve and its Role in Robust Esti-
mation,” Journal of the American Statistical Association, 69, 383-393.
Rousseeuw and Croux: Alternatives to the MAD 1283
Hampel, F. R., Ronchetti, E. M., Rousseeuw, P. J., and Stahel, W. A. (1986), Rousseeuw, P. J., and Croux, C. (199 I), “Alternatives to the Median Absolute
Robust Statistics: The Approach Based on Influence Functions, New York: Deviation,” Technical Report 9 1-43, Universitaire Instelling Antwerpen,
John Wiley. Belgium.
He, X., and Simpson, D. G. (1993), “Lower Bounds for Contamination (1992), “Explicit Scale Estimators with High Breakdown Point,” in
Bias: Globally Minimax versus Locally Linear Estimation,” The Annals L,-Statistical Analysis and Related Methods, ed. Y. Dodge, Amsterdam:
OfStatistics. 21, 314-337. North-Holland, pp. 77-92.
Hodges, J. L., Jr., and Lehmann, E. L. (l963), “Estimates of Location Based (1993), “The Bias of k-Step M-estimators,” submitted to Statistics
o n Rank Tests,” Annals of Mathematical Statistics, 34, 598-61 1. and Probability Letters.
Hossjer, O., Croux, C., and Rousseeuw, P. J. (1993), “Asymptotics of a Rousseeuw, P. J., and Hubert, M. (1993), “Regression-Free and Robust
Very Robust Scale Estimator,” technical report 93- 10, Universitaire In- Estimation of Scale,” technical report 93-24, Universitaire Instelling An-
stelling Antwerpen, Belgium. twerpen, Belgium.
Huber, P. J. (1981), Robust Statistics. New York: John Wiley. Rousseeuw, P. J., and Leroy, A. M. (1987), Robust Regression and Outlier
Johnson, N. L., and Kotz, S. (1970), Continuous Univariate Distributions- Detection, New York: John Wiley.
I , New York: John Wiley. (1988), “A Robust Scale Estimator Based on the Shortest Half,”
Kaufman, L., and Rousseeuw, P. J. (1990), Finding Groups in Data: An Statistica Neerlandica, 42, 103- 1 16.
Introduction to Cluster Analysis, New York: John Wiley. Rousseeuw, P. J., and Yohai, V. J. (1984), “Robust Regression by Means
Martin, R. D., Yohai, V. J., and Zamar, R. H. (1989), “Min-Max Bias of SEstimators.” in Robust and Nonlinear Time Series Analysis, (Lecture
Robust Regression,” The Annals of Statisfics, 17, 1608-1630. Notes in Statistics 26), eds. J. Franke, W. Hardle, and R. D. Martin, New
Martin, R. D., and Zamar, R. H. (1989), “Asymptotically Min-Max Bias York: Springer-Verlag. pp. 256-272.
Robust M-estimates of Scale for Positive Random Variables,” Journal of Serfling, R. J. (1984). “Generalized L-, M-, and R-Statistics,” The Annals
the American Statistical Association, 84, 494-50 I . of Statistics, 12, 76-86.
(1991), “Bias Robust Estimation of Scale when Location is Un- Shamos, M. I. (1976), “Geometry and Statistics: Problems at the Interface,”
known,” submitted to The Annals of Statistics. in New Directions and Recent Results in Algorithms and Complexity, ed.
Downloaded by [Carnegie Mellon University] at 11:22 17 January 2015
Rousseeuw, P. J. (1984), “Least Median of Squares Regression,” Journal J. F. Traub, New York: Academic Press, pp. 251-280.
of the American Statistical Association. 79, 87 1-880. Siegel, A. F. (l982), “Robust Regression Using Repeated Medians,” Bio-
Rousseeuw, P. J., and Bassett, G. W., Jr. (1990), “The Remedian: A Robust metrika, 69, 242-244.
Averaging Method for Large Data Sets,” Journal of the American Statistical Tukey, J. W. (1977), Exploratory Data Analysis, Reading, MA: Addison-
Association. 85. 97- 104. Weslev.