Multiple Correspondence Analysis

Hervé Abdi1 & Dominique Valentin

1 Overview
Multiple correspondence analysis (MCA) is an extension of corre-
spondence analysis (CA) which allows one to analyze the pattern of
relationships of several categorical dependent variables. As such,
it can also be seen as a generalization of principal component anal-
ysis when the variables to be analyzed are categorical instead of
quantitative. Because MCA has been (re)discovered many times,
equivalent methods are known under several different names such
as optimal scaling, optimal or appropriate scoring, dual scaling,
homogeneity analysis, scalogram analysis, and quantification me-
thod.
Technically MCA is obtained by using a standard correspon-
dence analysis on an indicator matrix (i.e., a matrix whose entries
are 0 or 1). The percentages of explained variance need to be cor-
rected, and the correspondence analysis interpretation of inter-
point distances needs to be adapted.
1
In: Neil Salkind (Ed.) (2007). Encyclopedia of Measurement and Statistics.
Thousand Oaks (CA): Sage.
Address correspondence to: Hervé Abdi
Program in Cognition and Neurosciences, MS: Gr.4.1,
The University of Texas at Dallas,
Richardson, TX 75083–0688, USA
E-mail: [email protected] http://www.utd.edu/∼herve

1
 H. Abdi & D. Valentin: Multiple Correspondence Analysis

2 When to use it
M CA is used to analyze a set of observations described by a set of
nominal variables. Each nominal variable comprises several lev-
els, and each of these levels is coded as a binary variable. For ex-
ample gender, (F vs. M) is one nominal variable with two levels.
The pattern for a male respondent will be 0 1 and 1 0 for a female.
The complete data table is composed of binary columns with one
and only one column taking the value “1” per nominal variable.
M CA can also accommodate quantitative variables by recod-
ing them as “bins.” For example, a score with a range of −5 to +5
could be recoded as a nominal variable with three levels: less than
0, equal to 0, or more than 0. With this schema, a value of 3 will be
expressed by the pattern 0 0 1. The coding schema of MCA implies
that each row has the same total, which for CA implies that each
row has the same mass.

3 An example
We illustrate the method with an example from wine testing. Sup-
pose that we want to evaluate the effect of the oak species on barrel-
aged red Burgundy wines. First, we aged wine coming from the
same harvest of Pinot Noir in six different barrels made with two
types of oak. Wines 1, 5, and 6 were aged with the first type of
oak, whereas wines 2, 3, and 4 were aged with the second. Next,
we asked each of three wine experts to choose from two to five
variables to describe the wines. For each wine and for each vari-
able, the expert was asked to rate the intensity. The answer given
by the expert was coded either as a binary answer (i.e., fruity vs.
non-fruity) or as a ternary answer (i.e., no vanilla, a bit of vanilla,
clear smell of vanilla). Each binary answer is represented by 2 bi-
nary columns (e.g., the answer “fruity” is represented by the pat-
tern 1 0 and “non-fruity” is 0 1). A ternary answer is represented
by 3 binary columns (i.e., the answer “some vanilla” is represented
by the pattern 0 1 0). The results are presented in Table 1 (the same
data are used to illustrate STATIS and Multiple factor analysis, see
the respective entries). The goal of the analysis is twofold. First

2
 H. Abdi & D. Valentin: Multiple Correspondence Analysis

we want to obtain a typology of the wines and second we want to

know if there is an agreement between the scales used by the ex-
perts. We will use the type of type of oak as a supplementary (or
illustrative) variable to be projected on the analysis after the fact.
Also after the testing of the six wines was performed, an unknown
bottle of Pinot Noir was found and tested by the wine testers. This
wine will be used as a supplementary observation. For this wine,
when an expert was not sure of how to use a descriptor, a pattern
of response such .5 .5 was used to represent the answer.

4 Notations
There are K nominal variables, each nominal variable has J k lev-
els and the sum of the J k is equal to J . There are I observations.
The I × J indicator matrix is denoted X. Performing CA on the in-
dicator matrix will provide two sets of factor scores: one for the
rows and one for the columns. These factor scores are, in gen-
eral scaled such that their variance is equal to their corresponding
eigenvalue (some versions of CA compute row factor scores nor-
malized to unity).
The grand total of the table is noted N , and the first step of
the analysis is to compute the probability matrix Z = N −1 X. We
denote r the vector of the row totals of Z, (i.e., r = Z1, with 1 being
a conformable vector of 1’s) c the vector of the columns totals, and
Dc = diag {c}, Dr = diag {r}. The factor scores are obtained from the
following singular value decomposition:
−1
³ ´ −1
Dr 2 Z − rcT Dc 2 = P∆QT (1)

(∆ is the diagonal matrix of the singular values, and Λ = ∆2 is the

matrix of the eigenvalues). The row and (respectively) columns
factor scores are obtained as
−1 −1
F = Dr 2 P∆ and G = Dc 2 Q∆ . (2)
The squared (χ2 ) distance from the rows and columns to their re-
spective barycenter are obtained as
n o n o
dr = diag FFT and dc = diag GGT . (3)

3
 Table 1: Data for the barrel-aged red burgundy wines example. “Oak Type" is an illustrative (supplementary)
variable, The wine W? is an unknown wine treated as a supplementary observation.

Expert 1 Expert 2 Expert 3

Oak red
Wine Type fruity woody coffee fruit roasted vanillin woody fruity butter woody
W1 1 1 0 0 0 1 0 1 1 0 0 1 0 0 1 0 1 0 1 0 1 0 1

4
W2 2 0 1 0 1 0 1 0 0 1 1 0 0 1 0 1 0 0 1 1 0 1 0
W3 2 0 1 1 0 0 1 0 0 1 1 0 1 0 0 1 0 0 1 1 0 1 0
W4 2 0 1 1 0 0 1 0 0 1 1 0 1 0 0 1 0 1 0 1 0 1 0
W5 1 1 0 0 0 1 0 1 1 0 0 1 0 0 1 0 1 1 0 0 1 0 1
W6 1 1 0 0 1 0 0 1 1 0 0 1 0 1 0 0 1 1 0 0 1 0 1

W? ? 0 1 0 1 0 .5 .5 1 0 1 0 0 1 0 .5 .5 1 0 .5 .5 0 1
H. Abdi & D. Valentin: Multiple Correspondence Analysis
 H. Abdi & D. Valentin: Multiple Correspondence Analysis

The squared cosine between row i and factor ` and column j and
factor ` are obtained respectively as:

f i2,` g 2j ,`
o i ,` = 2
and o j ,` = 2
. (4)
d r,i d c, j

2 2
(with d r,i , and d c, j
, being respectively the i -th element of dr and
the j -th element of dc ). Squared cosines help locating the factors
important for a given observation or variable.
The contribution of row i to factor ` and of column j to factor
` are obtained respectively as:

f i2,` g 2j ,`
t i ,` = and t j ,` = . (5)
λ` λ`

Contributions help locating the observations or variables impor-

tant for a given factor.
Supplementary or illustrative elements can be projected onto
the factors using the so called transition formula. Specifically, let
iT
sup being an illustrative row and jsup being an illustrative column
to be projected. Their coordinates fsup and gsup are obtained as:
³ ´ ³ ´
fsup = iT
sup 1 i T
sup G∆ −1
and gsup = j T T
sup jsup F∆
1 −1
. (6)

Performing CA on the indicator matrix will provide factor scores

for the rows and the columns. The factor scores given by a CA pro-
gram will need, however to be re-scaled for MCA, as explained in
the next section.
The J × J table obtained as B = XT X is called the Burt matrix
associated to X. This table is important in MCA because using CA
on the Burt matrix gives the same factors as the analysis of X but
is often computationally easier. But the Burt matrix also plays an
important theoretical rôle because the eigenvalues obtained from
its analysis give a better approximation of the inertia explained by
the factors than the eigenvalues of X.

5
 H. Abdi & D. Valentin: Multiple Correspondence Analysis

5 Eigenvalue correction for multiple corre-

spondence analysis
M CA codes data by creating several binary columns for each vari-
able with the constraint that one and only one of the columns gets
the value 1. This coding schema creates artificial additional di-
mensions because one categorical variable is coded with several
columns. As a consequence, the inertia (i.e., variance) of the so-
lution space is artificially inflated and therefore the percentage of
inertia explained by the first dimension is severely underestimated.
In fact, it can be shown that all the factors with an eigenvalue less
or equal to K1 simply code these additional dimensions (K = 10 in
our example).
Two corrections formulas are often used, the first one is due
to Benzécri (1979), the second one to Greenacre (1993). These
formulas take into account that the eigenvalues smaller than K1
are coding for the extra dimensions and that MCA is equivalent to
the analysis of the Burt matrix whose eigenvalues are equal to the
squared eigenvalues of the analysis of X. Specifically, if we denote
by λ` the eigenvalues obtained from the analysis of the indicator
matrix, then the corrected eigenvalues, denoted c λ are obtained as

1 2
·µ
K
¶µ ¶¸
1
λ` − if λ` >



 K −1

K K
c λ` = . (7)

 1
if λ` ≤

0

K
Using this formula gives a better estimate of the inertia, extracted
by each eigenvalue.
Traditionally, the percentages of inertia are computed by di-
viding each eigenvalue by the sum of the eigenvalues, and this ap-
proach could be used here also. However, it will give an optimistic
estimation of the percentage of inertia. A better estimation of the
inertia has been proposed by Greenacre (1993) who suggested in-
stead to evaluate the percentage of inertia relative to the average
inertia of the off-diagonal blocks of the Burt matrix. This average

6
 Table 2: Eigenvalues, corrected eigenvalues, proportion of explained inertia and corrected proportion of explained
inertia. The eigenvalues of the Burt matrix are equal to the squared eigenvalues of the indicator matrix; The
corrected eigenvalues for Benzécri and Greenacre are the same, but the proportion of explained variance differ.
Eigenvalues are denoted by λ, proportions of explained inertia by τ (note that the average inertia used to
compute Greenacre’s correction is equal to I = .7358).

Indicator Burt Benzécri Greenacre

Matrix Matrix Correction Correction

7
Factor Iλ τI Bλ τB Zλ τZ cλ τc
1 .8532 .7110 .7280 .9306 .7004 .9823 .7004 .9519
2 .2000 .1667 .0400 .0511 .0123 .0173 .0123 .0168
3 .1151 .0959 .0133 .0169 .0003 .0004 .0003 .0004
4 .0317 .0264 .0010 .0013 0 0 0 0
P
1.2000 1 .7822 1 .7130 1 .7130 .9691
H. Abdi & D. Valentin: Multiple Correspondence Analysis
 H. Abdi & D. Valentin: Multiple Correspondence Analysis

inertia, denoted I can be computed as

à !
K X 2 J −K
I= × λ` − (8)
K −1 ` K2

According to this approach, the percentage of inertia would be ob-

tained by the ratio

cλ cλ
τc = instead of P . (9)
I c λ`

6 Interpreting MCA
As with CA, the interpretation in MCA is often based upon proxim-
ities between points in a low-dimensional map (i.e., two or three
dimensions). As well as for CA, proximities are meaningful only
between points from the same set (i.e., rows with rows, columns
with columns). Specifically, when two row points are close to each
other they tend to select the same levels of the nominal variables.
For the proximity between variables we need to distinguish two
cases. First, the proximity between levels of different nominal vari-
ables means that these levels tend to appear together in the obser-
vations. Second, because the levels of the same nominal variable
cannot occur together, we need a different type of interpretation
for this case. Here the proximity between levels means that the
groups of observations associated with these two levels are them-
selves similar.

6.1 The example

Table 2 lists the corrected eigenvalues and proportion of explained
inertia obtained with the Benzécri/Greenacre correction formula.
Tables 3 and 4 give the corrected factor scores, cosines, and con-
tributions for the rows and columns of Table 1. Figure 1 displays
the projections of the rows and the columns. We have separated
these two sets, but, because the projections have the same vari-
ance, these two graphs could be displayed together (as long as one

8
 Table 3: Factor scores, squared cosines, and contributions for the observations ( I -set). The eigenvalues and
proportions of explained inertia are corrected using Benzécri/Greenacre formula. Contributions corresponding
to negative scores are in italic. The mystery wine (Wine ?) is a supplementary observation. Only the first two
factors are reported.

Wine 1 Wine 2 Wine 3 Wine 4 Wine 5 Wine 6 Wine ?

F cλ %c Factor Scores

1 .7004 95 0.86 −0.71 −0.92 −0.86 0.92 0.71 0.03

9
2 .0123 2 0.08 −0.16 0.08 0.08 0.08 −0.16 −0.16

F Squared Cosines

1 .62 .42 .71 .62 .71 .42 .04

2 .01 .02 .01 .01 .01 .02 .96

F Contributions ×1000

1 177 121 202 177 202 121 −

2 83 333 83 83 83 333 −
H. Abdi & D. Valentin: Multiple Correspondence Analysis
 Expert 1 Expert 2 Expert 3 Oak
λ 2 = .0123
4 τ2 = 2%
1 5 Vanilla−a Vanilla−c
3 Wood−a Wood−c
Wood−y Fruit−n Wood−n
Fruit−y Roast−n
Wood−yRoast−y Wood−n

2 ? 6 λ 1= .7004 O−1 Butter−y Butter−n O−2

τ 1 = 95% Fruit−n Coffee−y Coffee−nFruit−y
Fruit−n Fruit−y
Vanilla−b

10
Wood−b

a b
Figure 1: Multiple Correspondence Analysis. Projections on the first 2 dimensions. The eigenvalues (λ) and
proportion of explained inertia (τ) have been corrected with Benzécri/Greenacre formula. (a) The I set: rows
(i.e., wines), wine ? is a supplementary element. (b) The J set: columns (i.e., adjectives). Oak 1 and Oak 2 are
supplementary elements. (the projection points have been slightly moved to increase readability). (Projections
from Tables 3 and 4).
H. Abdi & D. Valentin: Multiple Correspondence Analysis
 Table 4: Factor scores, squared cosines, and contributions for the for the variables ( J -set). The eigenvalues and
percentages of inertia have been corrected using Benzécri/Greenacre formula. Contributions corresponding to
negative scores are in italic. Oak 1 and 2 are supplementary variables.

Expert 1 Expert 2 Expert 3

red
fruity woody coffee fruit roasted vanillin woody fruity butter woody Oak

y n 1 2 3 y n y n y n 1 2 3 y n y n y n y n 1 2

F c λ %c Factor Scores

11
1 .7004 95 .90 −.90 −.97 .00 .97 −.90 .90 .90 −.90 −.90 .90 −.97 .00 .97 −.90 .90 .28 −.28 −.90 .90 −.90 .90 .90 −.90
2 .0103 2 .00 .00 .18 −.35 .18 .00 .00 .00 .00 .00 .00 .18 −.35 .18 .00 .00 .00 .00 .00 .00 .00 .00 .00 .00

F Squared Cosines

1 .81 .81 .47 .00 .47 .81 .81 .81 .81 .81 .81 .47 .00 .47 .81 .81 .08 .08 .81 .81 .81 .81 1.00 1.00
2 .00 .00 .02 .06 .02 .00 .00 .00 .00 .00 .00 .02 .06 .02 .00 .00 .00 .00 .00 .00 .00 .00 .00 .00

F Contributions ×1000

1 58 58 44 0 44 58 58 58 58 58 58 44 0 44 58 58 6 6 58 58 58 58 − −
2 0 0 83 333 83 0 0 0 0 0 0 83 333 83 0 0 0 0 0 0 0 0 − −
H. Abdi & D. Valentin: Multiple Correspondence Analysis
 H. Abdi & D. Valentin: Multiple Correspondence Analysis

keeps in mind that distances between point are meaningful only

within the same set). The analysis is essentially uni-dimensional,
with Wines 2, 3, and 4 being clustered on the negative side of the
factors and Wines 1,5, and 6 on the positive side. The supplemen-
tary (mystery) wine does not seem to belong to either clusters. The
analysis of the columns shows that the negative side of the factor
is characterized as being non fruity, non-woody and coffee by Ex-
pert 1, roasted, non fruity, low in vanilla and woody for Expert 2,
and buttery and woody for Expert 3. The positive side, here gives
the reverse pattern. The supplementary elements indicate that the
negative side is correlated with the second type of oak whereas the
positive side is correlated with the first type of oak.

7 Alternatives to MCA
Because the interpretation of MCA is more delicate than simple
CA , several approaches have been suggested to offer the simplic-
ity of interpretation of CA for indicator matrices. One approach
is to use a different metric than χ2 , the most attractive alterna-
tive being the Hellinger distance (see entry on distances and Es-
cofier, 1978; Rao, 1994). Another approach, called joint correspon-
dence analysis, fits only the off-diagonal tables of the Burt matrix
(see Greenacre, 1993), and can be interpreted as a factor analytic
model.

References
[1] Benzécri, J.P. (1979). Sur le calcul des taux d’inertie dans
l’analyse d’un questionnaire. Cahiers de l’Analyse des Données,
4, 377–378.
[2] Clausen, S.E. (1998). Applied correspondence analysis. Thousand
Oaks (CA): Sage.
[3] Escofier, B. (1978). Analyse factorielle et distances répondant au
principe d’équivalence distributionnelle. Revue de Statistiques
Appliquées, 26, 29–37.

12
 H. Abdi & D. Valentin: Multiple Correspondence Analysis

[4] Greenacre, M.J. (1984). Theory and applications of correspon-

dence analysis. London: Academic Press.
[5] Greenacre, M.J. (1993). Correspondence analysis in practice. Lon-
don: Academic Press.
[6] Rao, C. (1995). Use of Hellinger distance in graphical displays.
In E.-M. Tiit, T. Kollo, & H. Niemi (Ed.): Multivariate statistics
and matrices in statistics. Leiden (Netherland): Brill Academic
Publisher. pp. 143–161.
[7] Weller S.C., & Romney, A.K. (1990). Metric scaling: Correspon-
dence analysis. Thousand Oaks (CA): Sage.

Acknowledgments
Many thanks to Szymon Czarnik and Michael Greenacre for point-
ing out a mistake in a previous version of this paper.

13