You are on page 1of 10

A CIRCUMFLEX MODEL FOR MATERNAL BEHAVIOR

EARL S. SCHAEFER
National Institute of Mental Health, Bethesda, Maryland

T
HE goal of this research is to develop dividing factor analysis into the earlier approach of
an ordered, parsimonious nomological common-iactors and his notion of order-lactors, although
he points out many connections between the two kinds
network for the domain of maternal be- of factor theories. The circumplex order is described as
havior. A theoretical basis for this organiza- follows: "It is an order which has no beginning and no
tion of maternal behavior is provided by end, namely, a circular order. A set of variables obeying
Cronbach and Meehl's (1955) analysis of such a law will be called a circumplex to designate a
construct validity and by Guttman's (1954) 'circular order of complexity'." In this article, as in
Guttman's presentation of the model, the law of circu-
circumplex model for the analysis of the inter- lar order or law of neighboring is applied to "the
relationships of a set of concepts. Both articles simplest case of the radex which can be completely
suggest that if a systematic set of relationships portrayed by a simple, two-dimensional diagram." The
among concepts can be demonstrated, the data that radex theory was designed to order are sets of
product-moment correlation coefficients. Guttman de-
scientific usefulness of a conceptual scheme is veloped the circumplex model to organize mental test
increased. The purpose of this paper is to data but has found "some data from the Minnesota
demonstrate the generality of a social interac- Multiphasic Inventory to form an approximate circum-
tion conceptualization of maternal behavior by plex."
ordering the intcrcorrelation matrices of three Guttman (1954) points out the need to define a
universe of content before attempting to order data.
sets of data on maternal behavior. When The universe of content of this paper is the social and
ordered both with factor analysis and with emotional behavior of a mother toward an individual
Guttman's circumplex model, similar two- child. Therefore, variables that only indirectly relate to
dimensional organizations of maternal behavior maternal behavior with the individual child were ex-
cluded from the sets of variables to be ordered. Ex-
concepts were found for the three sets of data. amples of variables that were excluded are behaviors
Additional confirmation of the two-dimen- toward persons other than the child, such as Coopera-
sional organization was found in Baldwin, tiveness with Examiners, Sociability, and Dependency,
Kalhorn, and Breese's (1945) data on the Fels and situational stresses such as Financial Stress.
Parent Behavior Rating Scales, although Although Guttman did not attempt to limit the
methods by which a circumplex order can be dis-
Roff's (1949) factor analysis and Lorr and covered, he explicitly included inspection of the matrix.
Jenkins' (1953) second-order factor analysis Guttman's (1954) statement of this point, although
demonstrate that additional dimensions can made in relation to the simplex, applies equally to the
also be found in the Fels scales. circumplex if "order" is substituted for "hierarchy."
"I submit that a hierarchy, if it really exists, may often
DATA AND METHOD reveal itself to inspection alone. Not always, to be sure;
we shall see reasons why a hierarchy can sometimes be
Descriptions of the data ordered in this paper can obscured to direct inspection. But surely the converse
be found in published studies by Schaefer, Bell, and holds, if inspection reveals a hierarchy, then the
Bayley (in press), Sanford, Adkins, Miller, and Gobi) hierarchy exists. . . ."
(1943), and Baldwin, Kalhorn, and Brcese (1945). Another method, not reported by Guttman (1954),
These studies have in common the use of concepts that which was developed in this study to search for a cir-
describe molar social and emotional interactions. This cumplex order was to choose two variables from the
level of conceptualization is supported by Symond's universe of content that had high correlations with
(1939) emphasis upon the quality of the parent-child other variables in the matrix and zero or low correla-
relationship and by Orlansky's (1949) conclusion that tions with one another. By plotting the correlations of
the total pattern of a child's experience rather than all variables with the two relatively orthogonal variables
specific infant care practices is the important factor in one could determine, by inspection, if a circular order
personality development. The relatively molar and of the variables was revealed. Use of this method re-
abstract concepts used have been given behavioral sulted in the ordering of both positively and negatively
definitions or have unambiguous interpretations. correlated variables. Guttman had stated that radex
The primary statistical model used in the organiza- theory could allow for negative correlations although
tion of these studies was Guttman's radex theory he did not discuss them further in his original paper.
(1954). Radex theory, which Guttman (1954) calls "a All circumplex orders reported in this paper were estab-
new approach to factor analysis," includes two com- lished by plotting correlations of variables with two
ponents, the simplex, which is not pertinent to this orthogonal variables and then, by inspection, rearrang-
paper, and the circumplex. Guttman (1954) suggests ing the correlation matrix into circumplex order.
226
TABLE 1
CIRCUMFLEX ORDERED CORRELATION MATRIX, UNROTATED FACTOR LOADINGS, AND RESIDUALS FOR ZERO TO THREE-YEAR MATERNAL RATINGS JUDGED
TO BELONG TO UNIVERSE
(N = 56)
Unrotated
Factor
Miternal Behaviors Loadings CircuBplex Ordered Matrix and Residuals
I II 1 2 3 1* 5 6 7 8 9 10 11 12 13 Ik 15 16 17 18
1. Autonomy -1*7 67 00 -06 07 -oi* -01 -06 05 12 -01 11 03 -17 01 00 02 oi* -03 >
2. Ignoring -37 -08 36 -03 08 -05 -08 -09 06 01 -01 -1C -01 02 oi* -01 -02 -02 -01* n
3. Punltiveness -76 -1*9 •03 67 00 00 00 oi* 00 -02 -06 -06 01 01 -05 oi* 02 05 05 5
n
1*. Perceives Child as Burden -7k -1*9 09 76 3o -12 -03 -10 07 07 00 07 02 -07 01 -09 00 01+ -06 d
5. Strictness -66 -50 •07 56 75 62 oi* 07 -07 -05 01 09 00 05 -03 -02 -02 -06 02
6. Use of Fear to Control -67 -57 •08 55 79 75 76 13 -06 00 05 01 -07 -03 oi* 00 oi* 02 Oh £2
7. Punishment -61 -53 •13 1*8 76 61 73 81* 01 -12 07 -02 -06 03 -06 12 06 Ok 10 a
X
8. Irritability -57 -67 •13 61 76 82 61* 70 72 09 03 -06 01 -03 02 oi* oi* 01 01
9- Anxiety 21 -69 •1*1* -11 16
08
25 15 25 12 1*3 00 -oi* 09 -09 06 05 03 00 -03 §
O
10. Intrusiveness 29 -73 •61* -20 15 18 23 28 35 56 12 -05 -02 05 -03 02 02 02 o
11. Concern about Health 31* -53 •1*1 -36 -06 08 13 08 05 11 1*0 61 12 -13 -03 -13 -02 oi* -03 H
tr*
12. Achievement Demand 53 -50 •56 -1*3 -li* -12 -10 -15 -12 05 51* 1*6 56 -10 -09 03 -06 05 03
13. Excessive Contact 63 -51 •81 -1*9 -22 -29 -11 -16 -08 -10 39 53 35 1*8 01 02 00 -02 02 O
Ik. Fostering Dependency 72 -1*3 •62 -56 -39 -31 -35 -20 -27 -10 51 57 1*1* 51 68 -02 oi* -02 -01 W
15- Emotional Involvement 76 -23 •51 -65 -1*3 -51* -1*1 -38 -22 -2l* 37 31 25 55 62 63 01* 02 03 g
16. Expression of Affection 3° 15 •30 -So -73 -73 -68 -65 -56 -57 12 17 20 33 US 62 69 03 02 f.
17-
18.
Etiualitarianism
Positive Evaluation
82
76
43
1*1*
•06 -76
10 -71*
-78
-75
-78 -81 -78 -69 -75 -13 -05 09
-81* -70 -72 -59 -71 -17 -08 -05
26
21
23
28
39 51* 82
32
01
H
S
35 51 77
Correlations of Other Variables with Circumplex Order t-1
H
H
Rejection of Homemaking 21 67 57 71* 33 1*6 37 61 22 -03 -15 -12 -29 -28 -28 -1*6 -52 -50
Negative Emotional States •12 1*9 1*3 1*9 1*1 1*3 30 1*8 12 10 -06 -12 10 oi* -23 -1*5 -55 -5k >
Msod Swings •27 56 57 59 50 67 57 71* 50 1*2 20 23 05 09 -35 -50 -53
-01 Q
Financial Stress -02 31 23 31 32 20 30 36 08 09 00 -19 05 -20 -25 -21* -35 -26
Poor Physical Health -ll* 23 33 1*2 22 21* 31 36 oi* 20 -01 -11 -07 -13 -17 -21 -23 -23
DoDinance -i*l* -16 08 17 05 19 11 50 72 57 1*8 72 27 1*7 38 ll* -02 01
Self-Abasenent -18 -ll* -05 ll* 00 Ok 02 25 1*0 1*5 1*2 17 22 29 05 15 Ok -oi*
Dependency -08 18 -01 29 -02 03 02 35 27 23 30 21 00 17 01 06 -06 -12
Over-Conscientiousness -53 -1*8 -16 -12 -02 -13 -06 -02 1*0 1*7 6U 62 55 1*2 39 35 23 12
Sociability •17 -32 -29 -22 -21 -27 -19 -09 17 26 33 1*5 17 25 39 !*6 38 36
Cooperativeness 22 -63 -62 -61* -1*7 -55 -51* -69 -20 -23 09 05 11 13 31* 56 65 60
Estimated Intelligence 22 -55 -28 -1*1 -19 -36 -23 -38 -13 -26 oi* 23 -01 -18 31 2l* 37 37
Narcissism 00 07 -01 02 02 -02 01 ll* 27 20 .16. 35 -Ik 03 07 00 01 05 to
228 EARL S. SCHAEFER

of the three judges were retained. In order to


Autonomy
explore for a circumplex order, the correlations
60-}- of each variable with Equalitarianism and
• Ignoring
40 .| Intrusiveness were plotted. The variables ap-
Perceives Child as Puiitlveness <
peared to fall into a circular order. The corre-
Burden -60 -40 -20 20 40 60 80 100 lation matrix was rearranged by inspection
• Positive Evaluation
Equfllrtorianism-*
o( Chlld into the ordered matrix reported in Table 1.
Since inspection revealed, within error limits,

\
• Expression of Affection
Concern About Health, a well-ordered matrix, it was concluded that
Anxiety—*-.
• Emotional Involvement
11
the variables included in this universe form a
Intrusive ness *"*- -™isiue
U\ ^Excess*
•Fostering Dependency
Contact
quasi-circumplex as defined by Guttman
Achievement Demand (1954). To verify this ordering with a well-
FIG. 1. A CIRCUMFLEX OT MATERNAL known method, the matrix was factor-analyzed
BEHAVIOR RATINGS with Thurstone's (1947) complete centroid
Rotated factor loadings based on data from observations method. The factor loadings and the second
of maternal behavior between the child's age factor residuals are reported in Table 1. The
of zero to three years
second factor residuals are small and do not
To verify the orderings so obtained with a more justify extraction of a third factor. Before
familiar method, the correlation matrices were factor plotting the factor loadings in Fig. 1, the fac-
analyzed using Thurstone's (1947) complete ccntroid tors were orthogonally rotated clockwise 35° 3'.
method. In the judgment of the author, the residuals
after extraction of two factors do not justify extraction Plots of the factor loadings give a circular
of a third factor since the error variance, due to the order that is identical to that derived by inspec-
small size of the samples, is relatively large. The com- tion from the correlation matrix.
munalities of practically all variables are relatively The findings could be interpreted in terms
high, and the circular order obtained by plotting the
loadings of the variables on the two factors is practically
of two bipolar dimensions which could be
identical to the circular order of variables obtained by labeled Autonomy vs. Control and Love vs.
inspection of the correlation matrix. Although plotting Hostility. The first dimension would be repre-
the unrotatecl factor loadings revealed the circumplex sented by Autonomy at one pole and Anxiety
order, orthogonal rotations of two sets of data were of the Mother, Intrusiveness, Concern about
made to give the plotted figures a common orientation.
These rotations made it possible to interpret the results Health, Achievement Demand, Excessive Con-
in terms of both cowwzow-factors and order-factors. tact, Fostering Dependency, and Emotional
Involvement at the other pole. One pole of this
RESULTS AND DISCUSSION dimension resembles the concept of maternal
The first circumplex organization of mater- overprotcction (Levy, 1943). The positive pole
nal behavior was developed from ratings by of the second dimension would be Positive
Schaefer et al. (in press) of data gathered at Evaluation of the Child, Equalitarianism, and
the University of California Institute of Child Expression of Affection, and the negative pole
Welfare (Jones & Bayley, 1941). Fifty-six would be Ignoring, Punitiveness, Perceives the
mothers were rated on 32 behavior variables Child as a Burden, Strictness, Use of Fear to
from sets of notes that had been made by an Control, Punishment, and Irritability. The
observer during 10 to 20 testing sessions, with dimension of acceptance vs. rejection is similar
each mother and child between the child's age to this dimension (Symonds, 1939) although
of one month to three years. Reliabilities of Love vs. Hostility is suggested to represent it
the combined scores of three judges ranged here. However, consideration only of the
from .67 to .89, with a median of .82. Pearson comwow-f actors would not reveal the sequential
product-moment correlations were computed ordering within dimensions that is shown by
among the 32 variables. Three judges were Fig. 1.
asked to select the rating scales that were The correlations with the circumplex or-
directly related to the social and emotional dered variables of the variables that were
behavior of the mother with an individual excluded as not clearly relating to the universe
child. High agreement between judges was of content are also reported in Table 1. Many
found. The 18 scales selected by at least two of these variables are not highly related to the
TABLE 2
CIRCUMFLEX ORDERED CORRELATION MATRIX, UNROTATED FACTOR LOADINGS, AND RESIDUALS FOR NINE- TO FOURTEEN-YEAR
INTERVIEW RATINGS
(N = 34)
Unrotated
Scale Factor
Ho. Maternal Behaviors Loadings Circtmrplex Ordered Correlation Matrix and Residuals
Y~~
I II 1 15 7 12 18 21 28 8 9 20 13 24 17 5 23 10 27 19 n
1 Ignoring •80 44 04 01 05 01 -01 -02 02 06 -03 00 06 03 -02 02 04 00 04 S
o
15 Punitiveness •90 11 81 10 03 -02 -06 01 -03 06 06 04 00 n 01 03 04 -02 03 cj
7 Irritability •87 07 74 89 10 04 -04 00 -15 05 -05 19 03 17-09 04 06 02 09
12 Perceives Child as Burden •42 -03 38 41 46 07 04 03 01 00 -17 04 12 -02 04 -04 05 05 23 sw
13 Use of Fear to Control •85 -05 67 74 78 43 11 09 -11 -02 -n -02 04 -01 02 -05 -02 03 09 X
21 Strictness •62 -17 42 48 49 30 65 13 -08 00 -24 -17 20 00 05 -14 -ll 02 07
28 Wish to Control Child •82 -46 44 70 68 38 81 72 -08 -03 -05 -08 02 -03 00 -04 -03 -01 01 g
8 Social Isolation of Child •67 -38 39 53 4o 30 48 40 64 -06 14 -15 00 -13 o4 -08 01 -04 -16 o
9 Intrusiveness •66 -56 59 58 30 57 51 77 59 01 04 05 -05 03 05 05 01 -02 H
20 Excessive Contact •4l -64 02 36 27 02 27 12 58 65 64 11 -20 07 01 10 03 -02 -18 t-<
13 Fostering Dependency •16 -60 -13 11 29 13 15 03 33 19 48 55 -08 28 -20 15 01 03 10 NJ
24 Achievement Demand 06 -09 11 29 13 27 O
19 -64 -37 -24 -17 19 15 13 -03 -07 -02 02 05
17 Emotional Involvement 19 -75 -45 -14 -04 -08 -13 -08 15 02 22 47 70 65 -05 09 03 -02 -02
5 Positive Evaluation of Child 65 -55 -78 -63 -69 -21 -50 -26 -28 -19 -09 09 03 44 48 -01 00 -05 -02 §
23 Expression of Affection 80 -46 -82 -74 -69 -37 -71 -56 -49 -45 -22 06 30 37 58 76 02 00 04 >
10 Equalitarianism 88 -29 -79 -78 -72 -31 -76 -60 -62 -47 -37 -15 04 33 42 73 85 02 02 H
27 Positive Mather-Child Relationship 92 -22 -84 -87 -80 -33 -74 -51 -66 -58 -48 -26 01 33 31 67 84 89 13 §
19 Autonomy 45 62 -05 -30 -26 02 -32 -31 -64 -70 -67 -76 -34 27 -40 -07 12 24 40 >
f
Correlations of Other Variables with Circumplex Order w
S
77 73 78 28 59 24 4o 25 31 07 06 -32 -19
<j>
6 Rejection of Hbmemaking Role -67 -66 -57 -72 -11
lit Anxiety of Mother 44 70 67 33 43 37 65 43 64 42 34 -09 21 -26 -39 -52 -65 -51
26 Emotional Withdrawal of Mather 38 63 56 18 55 43 64 49 51 52 20 -11 20 -18 -43 -48 -58 -52 o
3 Dependency of Mother 19 36 42 43 14 13 22 16 30 11 42 n 27 -13 -11 -19 -31 -08
it Sociability of Mother -50 -66 -51 -18 -52 -25 -66 -58 -46 -58 -02 25 10 34 50 59 67 49
2 Cooperativeness -76 -88 -So -43 -71 -47 -75 -57 -67 -46 -14 19 08 48 65 70 83 46
11 Marital Happiness -55 -73 -73 -17 -52 -29 -50 -38 -38 -44 -23 13 -12 47 54 53 62 30
22 Estimated Intelligence of Mother -33 -31 -09 -29 -08 -47 -40 -29 -47 -12 43 20 31 33 46 56 44
16 Positive Emotional State of Mother -31 -36 -38 -30 -36 -14 -37 -22 -35 -13 -01 31 20 22 36 36 45 22
25 Communicativeness of Mother -40 -58 -37 -36 -42 -34 -58 -56 -50 -44 05 17 08 09 39 39 53 51
230 EARL S. SCHAEFER

circumplex set, while the correlations of others 100-

with the circumplex set are relatively high. It 60 i


Autonomy
seems probable that if variables that do not 60-

belong to a universe are included in a matrix 40


• Positive Mother- Child
they may obscure the circumplex order that 20-
Equa!«tarioni3m» Relationship
Expression of Affection
exists. -100 -SO -60 -40 -20 20 40 60 80 * 100

The circumplex organization of maternal be- Ignoring *


Positive Evaluation of Child
"20 '
havior was confirmed from a second set of data Perceives Child as Burde^

reported by Schaefcr et al. (in press). The data Puniliveness^ 1rri,ob||i,y -10 • • Achievement Demand
• Emotional Involvement
U» of Ftor to Control • S,r[^fl9s _6Q
were ratings made from written notes on either Social Isolation*
* Fostering Dependency

one or two home interviews with 34 mothers Wlih to Control .


• Excessive Contact
.-80 •

between the child's age of 9 to 14 years. The Intrusivenesff

subjects were again from the Berkeley Growth FIG. 2. A CIRCUMPLEX OF MATERNAL
Study (Jones & Bayley, 1941). The concepts BEHAVIOR RATINGS
were revised from the first set of ratings to be Rotated factor loadings based on data from interviews
with mothers between the child's age
appropriate for the age level of the children of nine to fourteen years
and for the type of data available. Median
reliability after combining the ratings of three Positive Mother-Child Relationship, Equali-
judges was .85. The 18 variables selected by at tarianism, Expression of Affection, Positive
least two of the three judges as belonging to Evaluation of the Child, and Emotional In-
the universe arc reported in circumplex order volvement as opposed to Ignoring, Punitive-
in Table 2. The order was established by ness, Irritability, and Use of Fear to Control.
plotting the correlations of all variables with However, a summary of these data in terms of
two orthogonal variables and then rearrang- cowMMow-factors seems to the author a less
ing the correlation matrix by inspection with adequate presentation than an organization of
the plotted order as a preliminary guide. The the variables in circumplex order, as reported
fact that the ordering of this matrix is not as in Fig. 2.
clear as the first may be due to the fact that The organization of the interview data on
the interview data may not be as valid as maternal behavior is similar to the organiza-
observation data. Again two factors were ex- tion of the behavior observations. Despite some
tracted from this matrix, and the second factor changes in the intercorrelation of variables, the
residuals were computed. Since the small size ordering is quite similar with a sequence of
of the sample suggests that much of the re- high autonomy, an equalitarian positive rela-
maining variance may be error variance, a tionship, high positive involvement, possessive-
third factor was not extracted. The communali- ness, hostile control, hostile irritability, and
ties of most variables are high. The factors hostile ignoring. Although some sectors of the
were rotated counterclockwise orthogonally circular space are less well represented than
35° 58' to an orientation similar to that of the others, by projecting polar opposites of the
first set of data and are plotted in Fig. 2. included variables the entire two dimensional
The orthogonal rotation did not result in space would be adequately represented.
simple structure nor is it obvious that simple The ratings by Sanford et al. (1943) of
structure in terms of common-factors exists for parental press variables were also organized
this matrix. There is, however, a relatively with the circumplex model. The intcrcorrcla-
clear ordering of variables within a two- tions of the variables that were judged to be
dimensional space. The rotated reference axes directly related to the universe are ordered in
could again be interpreted as Autonomy vs. Table 3. The unrotated factor loadings and
Control as defined by Autonomy and the op- second-factor residuals using Thurstone's
posed variables of Strictness, Social Isolation (1947) centroid method are also reported in
of the Child, Wish to Control the Child, In- Fig. 3 and Table 3. Although the sampling of
trusiveness, Excessive Contact, and Fostering the universe is less extensive, the ordering ob-
Dependency. The horizontal axis might be tained can be interpreted as similar to that of
labeled Love vs. Hostility as represented by the other studies. The area of low Dominance
A CIRCUMFLEX MODEL FOR MATERNAL BEHAVIOR 231

can be interpreted as autonomy, Deference and Roff's (1949) factor analysis of the Fels data
Affiliation as a loving positive relationship, resulted in seven factors, thus confirming that
Nurturance as a more involved but positive the total set of data is factorially complex.
relationship, Dominance and Retention as con- Lorr and Jenkins (1953) performed a second-
trol, Aggression as hostility, and Rejection order factor analysis using Roff's matrix of
and Capricious Discipline as hostile rejection. correlations of the primary factors and re-
Presses not included in this universe tend to ported three second-order factors that were
have lower correlations except for Lack labeled dependency-encouraging, democracy
(Family Support) that is highly related to Ag- of child training, and organization and effec-
gression and Rejection. Some variables of tiveness of control in the home. Roff's (1949)
parental behavior are not highly related to the first two unrotated factors included most of
universe as defined. In denning the universe of the common factor variance of the scales in-
content as the social and emotional behavior cluded in Baldwin et al.'s (1945) three major
of the mother with the individual child, many clusters and also gave a sequential ordering of
variables that are also thought to be relevant variables that appeared similar to the circum-
to the development of the child have been plex orders of maternal behavior found in
omitted from these analyses. Although these previous studies. An attempt was therefore
other dimensions should also be studied in- made to organize the set of variables included
tensively, they should not be permitted to in Baldwin et al.'s (1945) three major clusters
obscure the circumplex order that exists within in circumplex order. The quasi-circumplex that
the specified universe. emerged is reported in Table 4. The unrotated
Published research on the Fels Parent Be- first-order factor loadings reported by Roff
havior Rating Scales allows a comparison of (1949) and the clusters to which the scales
different ways of organizing the same set of were assigned by Baldwin et al. (1945) are
data. Baldwin et al. (1945) reported the reported in circumplex order in Table 5. We
intercorrelations of the scales and also re- see a sequence from the cluster of Indulgence
ported an organization of 19 of the variables to Acceptance to Democracy. There is no
into three major clusters: Democracy in the sharp demarcation between these clusters.
home, Acceptance of the child, and Indulgence. Most of the variance of the scales is included
They also reported "a number of other syn- in the first two factors, and the loadings of the
dromes which seem of minor importance." scales on subsequent factors are low. Effective-

TABLE 3
CIRCUMFLEX ORDERED CORRELATION MATRIX, UNROTATED FACTOR LOADINGS, AND RESIDUALS OF PARENTAL
PRESS VARIABLES FROM SANFORD ET AL. (1943)

Unrotated Factor Circumplex Ordered Correlation Matrix and Residuals


Loadings
Maternal Behavior
I II 1 2 3 4 5 6 7 8

1. Deference 79 10 01 02 OS 05 03 13 -12
2. Affiliation 96 -IS 75 -04 -13 -03 07 -05 02
3. Nurturance 75 -SO 56 76 07 -03 -12 02 00
4. Retention -26 -65 -22 -28 19 02 -02 10 -IS
5. Dominance -46 -63 -37 -37 -05 55 06 08 -07
6. Aggression -96 -IS -75 -83 -76 33 59 -08 16
7. Rejection -78 38 -45 -86 -75 OS 20 61 -11
8. Capricious Discipline -66 54 -59 -69 -77 -33 -11 71 61
Correlations of Other Variables with Circumplex Order

Understanding 00 30 30 05 -12 -41 -26 -33


Exposition 12 14 46 48 56 -22 -12 -24
Excitance 06 37 10 -30 -24 -22 -05 37
Sentience 00 23 11 -10 10 -30 -30 -22
Lack (Family Support) -45 -59 -75 00 22 71 59 82
232 EARL S. SCHAEFER

100- AUTONOMY

> FREEDOM
80- •

Copriciousa 60-- •
INDIFFERENT

Discipline
» Rejection

NEGLECTING
Deference
-100 -80 -60 -40 -20 20 so * 100
I 1—I—I—1_ —t— -H 1 1
-100 -BO -60 -to -20 20 40 40 80
« Aggression i Affiliation
-20

-40-. -40 OVER INDULGENT •


i Nurluronce
-63
Domlnooce *

Retention7 OVER PROTECTIVE

-100
FIG. 3. A CIRCUMFLEX OK PARENTAL PRESS RATINGS
Unrotatcd factor loadings based on data from Sanford, FIG. 4. A HYPOTHETICAL CIRCUMFLEX OF MATERNAL
Adkins, Miller, and Cobb (1943) BEHAVIOR CONCEPTS

ness of Policy and Low Disciplinary Friction gence, protective indulgence, and overprotec-
have appreciable loadings on the third factor tiveness are placed in a quadrant that indicates
which may describe the success of parental both love for the child and an inability to treat
policy with the child. the child as a differentiated individual who
The three methods of organizing the inter- has his own activities and interests apart from
correlations of this set of data—the circumplex the parent. The variables of authoritarian,
ordering, the cluster method of Baldwin et al. dictatorial treatment and demanding, antag-
(1945), and Kofi's (1949) unrelated first- onistic behavior are placed in the quadrant
order factor analysis—give related organiza- that indicates both a hostile relationship to the
tions. These organizations are also related to child and control; while neglect, indifference,
Roff's (1949) rotated factor loadings and to and detachment are placed in the quadrant
Lorr and Jenkins' (1953) second-order factor that indicates varying combinations of hostil-
loadings that will not be discussed here. These ity and autonomy for the child. This organiza-
different ways of organizing the data illustrate tion seems to satisfy both a law of neighboring
the well-known fact that there are many differ- and a law of polar opposites.
ent ways to organize the same set of data. The This paper has demonstrated a two-dimen-
advantage of the circumplex order over the sional organization of maternal behavior con-
previous organizations is that it directs atten- cepts which describe molar social and emo-
tion to the sequential order of variables and tional interactions between mother and child.
docs not divide the matrix into discrete clusters A contrasting approach to the study of parental
or factors. behavior is that of Sears, Whiting, Nowlis, and
A generalization of this circumplex model Sears (1953), Sears, Maccoby, and Levin
can be used to order other concepts that have (1957), and Whiting and Child (1953), among
been used to describe maternal behavior. A others, who analyze more molecular variables
number of these concepts are hypothetically that relate to the socialization of specific sys-
ordered in Fig. 4. Freedom is placed on the
tems of behavior such as oral, anal, sexual, etc.
Autonomy vs. Control dimension with its polar
The results of these studies tend to show less
opposite, possessiveness. Democratic behavior
and cooperative behavior arc placed in the organization among variables. Whiting and
quadrant which indicates both loving accept- Child (1953) report practically no relationship
ance and autonomy. Acceptance and its polar between severity of socialization for different
opposite, rejection, are placed on the Love vs. systems of behavior in their cross-cultural
Hostility dimension. Excessive love, overindul- study. Sears et al. (1953, 1957) also report low
A CIRCUMFLEX MODEL FOR MATERNAL BEHAVIOR 233

TABLE 4
CIRCUMFLEX ORDERED CORRELATION MATRIX or TI
ET i
Scale
Kumber 2.11 Jf.l 1.91 U.2 7.1 2.12 7.2
2.11
l*.l 63
l.SO. 61 67
U.2 55 7U 71
7-1 50 53 6k 61*
2.12 1*7 Jj.y 63 62
1*7
7.2 1*8 5^t- 69 67 55 51
8.'* 1*2 UY 60 52 1*6 1*6 88
8.3 ia 53 60 50 1*3 53 79 86
5-2 35 ll^ 51 53 1& 1*9 61* 66 70
20 oli 23 28 33 31* 52 50 58 76
3.15
3.11* 18 22 27 32 39 39 52 76 92
6.1 20 18 29 31* 1*1 37 51 1*1* 67 79 87
3.22 15 18 23 31 37 38 51 53 71 88 88 78

8.1 11 18 20 30 16 27 51 61 60 69 71 75 70
3.16 08 08 26 36 31 29 1*3 35 36 62 70 7^ 60 61* 60

3.11 -05 ol* 08 03 00 10 33 U9 58 65 61* 56 63 56 1*2


Variables Reportj

3.17 2U 30 1*5 1*6 30 33 69 70 60 61 1*9 1*1* 37 1*3 W* 5^


1
3.18 21 23 39 1*5 33 23 66 66 50 5* 52 1*9 1*1 1*9 36 57 37

correlations between variables in their studies. and specific features of socialization as joint
Sears et al. (1957) reported a factor analysis of antecedent variables will be one of the impor-
parent behaviors derived from interview data tant directions taken by future research." The
which gave seven factors of parent behaviors, tentative integration of molar concepts of
yet the communality of the variables was low. parental behavior into an ordered conceptual
It is interesting to note that these investigators scheme, presented here, should facilitate such
(Sears et al, 1957) report that the quality of research.
maternal warmth is the most pervasive quality The conceptual model that has been pre-
of maternal behavior and that it has the most sented may also assist in the interpretation of
pervasive effects upon the child. Despite the existing maternal behavior research by sug-
different conceptual schemes, their quality of gesting the essential equivalence of diverse
maternal warmth appears similar to the dimen- conceptual schemes or may help in the selec-
sion of Love vs. Hostility of this study. Child tion of variables to be included in future re-
(1954) states, "It is probable that the com- search on maternal behavior. It may challenge
bined study of general parental characteristics investigators to explore new dimensions of
234 EARL S. SCHAEFER

TABLE S
's (1949) UNROIATED FACTOR LOADINGS POR THE FELS PARENT BEHAVIOR RATING SCALES WITH CLUSTER
MEMBERSHIP OF BALDWIN ET AL. (1945)
Scale Unrotated Factor Loadings
Number Scale Name I II III IV V VI VII Cluster Membership
2.11 Duration of Contact 68 -20 -06 -07 -06 -15 25 Indulgence
k.l General Babying 77 -20 -16 -10 -10 02 06 Indulgence
1.91 Child-Centeredne ss 8k -16 -01 15 10 00 02 Indulgence, Acceptance
k.2 General Protect!veness 80 -06 05 16 -18 03 27 Indulgence
7-1 Solicitouaness 72 -01 -17 12 -05 -38 -09 Indulgence
2.12 Intensity of Contact 69 Ok -2k 22 10 -09 -39 Indulgence
7.2 Acceptance 86 17 23 -05 -03 13 01 Indulgence, Acceptance
8.k Rapport 80 18 33 -19 07 19 -02 Acceptance
8.3 Affectionatenesa 79 25 Ok -2k 15 25 -12 Acceptance
5.2 Approval 69 53 03 00 15 -06 15 Acceptance, Democracy
3.15 Democracy k6 81 00 -12 05 -11 -10 Democracy
3.14 Justification of Policy k7 83 -07 02 -01 -10 -1k Democracy
6.1 Readiness of Explanation k5 78 -Ik 11 -11 -02 03 Democracy
3.22 Non-Coerciveness k5 79 -05 -Ok 09 -17 01 Democracy
8.1 Understanding kl 68 -05 -03 -18 35 25 Democracy
3.16 Clarity of Policy 36 68 2k 35 -16 -Ik 09 Democracy
3.H IJon-Restrictiveness 21 67 08 -29 31 19 -16 Democracy
Variable Reporting Success of Policy

3.17 Effectiveness of Policy 59 27 53 Ik 20 15 -10 Acceptance


3.18 Low Disciplinary Friction 53 33 63 -02 -03 -10 -10 Acceptance

maternal behavior rather than to reinvestigate 1. Two major dimensions of maternal be-
known dimensions with new names. havior can be isolated in all studies; these can
be labeled Love vs. Hostility and Autonomy vs.
SUMMARY AND CONCLUSIONS Control,
By organizing several empirical studies of 2. A circumplex organization of this universe
maternal behavior using Guttman's (1954) cir- shows a clearer nomological network than an
cumplex model, it was demonstrated that organization in terms of discrete dimensions.
many of the existing concepts of maternal
3. A theoretical generalization of the circum-
behavior can be ordered within a two-dimen-
sional space. Concepts that do not describe plex of maternal behavior is presented. It pro-
molar social and emotional interactions of vides an ordered, parsimonious nomological
mother and child often do not show a clear network or model for much of the variance of
circumplex ordering. The following interpreta- a mother's social and emotional behavior
tions of the findings were made: toward an individual child.
A CIRCUMFLEX MODEL FOR MATERNAL BEHAVIOR 235

REFERENCES SANI'ORD,R.N., ADKTNS, MARGARET M., MILLER, R. B.,


BALDWIN, A. L., KALHORN, JOAN, & BREESE, FAY & COBB, ELIZABETH. Physique, personality, and
HUFFMAN. Patterns of parent behavior. Psychol. scholarship. Monogr. Soc. Res. Child Develpm.,
Monogr., 1945, 58, No. 3 (Whole No. 268). 1943, 8, No. 1.
CHILD, I. L. Socialization. In G. Lindzey (Ed.), Hand- SCIIAEKER, E. S., BELL, R. Q., & BAYLEY, NANCY.
book of social psychology. Vol. II. Cambridge, Mass.: Development of a maternal behavior research
Addison-Wesley, 1954. instrument. /. genet. Psychol., in press.
CRONBACH, L. J., & MEEHL, P. E. Construct validity in SEARS, R. R., MACCOBY, E., & LEVIN, H. Patterns of
psychological tests. Psychol. Butt., 1955, 52, 281- child rearing. Evanston, 111.: Row, Peterson,
302. 1957.
GUTTMAN, L. A new approach to factor analysis: The SEARS, R. R., WHITING, J. W. M., NOWLIS, V., &
Radex. In P. F. Lazarfeld (Ed.), Mathematical SEARS, PAULINE S. Some child rearing antecedents
thinking in the social sciences. Glencoe, 111.: Free of aggression and dependency in young children,
Press, 1954. Genet, psychol. Monogr., 1953, 47, 134-234.
JONES, H. E., & BAYLEY, NANCY. The Berkeley growth SYMONDS, P. M. The psychology of parent-child relation-
study. Child Develpm., 1941, 12, 167-173. ships. New York: Appleton-Century, 1939.
LEVY, D. M. Maternal overprotection. New York:
Columbia Univer. Press, 1943. THURSTONE, L. L. Multiple-factor analysis. Chicago:
LORR, M., & JENKINS, R. L. Three factors in parent be- Univer. Chicago Press, 1947.
havior. /. consult. Psychol., 1953, 17, 306-308. WHITING, J. W. M., & CHILD, I. L. Child training and
ORLANSKY, H. Infant care and personality. Psychol. personality: A cross-cultural study. New Haven:
Bull., 1949, 46, 1-48. Yale Univer. Press, 1953.
ROFF, M. A factorial study of the Pels Parent Behavior
Scales. Child Develpm., 1949, 20, 29-45. Received July 14,1958.

You might also like