Вы находитесь на странице: 1из 19

Social Class and Corporal Punishment in Childrearing: A Reassessment

Author(s): Howard S. Erlanger


Source: American Sociological Review, Vol. 39, No. 1 (Feb., 1974), pp. 68-85
Published by: American Sociological Association
Stable URL: http://www.jstor.org/stable/2094277
Accessed: 25/01/2010 19:38

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=asa.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to
American Sociological Review.

http://www.jstor.org
SOCIALCLASS AND CORPORALPUNISHMENTIN CHILDREARING:
A REASSESSMENT*

HOWARDS. ERLANGER
Universityof Wisconsin,Madison
American Sociological Review 1974, Vol. 39 (February): 68-85
In 1958, in his reviewof availableliteratureon socializationand social class, Bronfenbrenner
concludedthat workingclass parents more often use physicalpunishment,while the middle
classresortsto psychologicaltechniquesof punishment.Thepresentpaperupdatesthis analysis
drawingon morerecentpublishedstudiesand on secondaryanalysisof a nationalsurvey;it also
examinesthe magnitudeof class differencesratherthan only their statisticalsignificance.This
analysissuggeststhat, althoughvariousstudieshavefound a statisticallysignificantrelationship,
the relationshipis weak.Analysisby type of indicatorof punishment,qualityof sample,age of
children,or year of study does not alter this conclusion.However, none of the studies is
definitive,and suggestionsare offered about topics to be pursuedin future research.
Besidestheir directrelevanceto the study of socializationand socialclass, the datareviewed
here suggest that several more general inferencesdrawnfrom earlierstudies are at present
empiricallyunsubstantiated.These include the hypotheses that physicalpunishmentleads to
"workingclass authoritarianism;" that childhood punishmentexperiencesexplain the greater
probabilitythat workingclass adults, as opposed to middleclassadults, will commit homicide;
that generaluse of corporalpunishmentis a precursorto child abuse;and that use of corporal
punishmentis part of a subculturalpositiveevaluationof violence.

For over a decade, the most comprehensive techniques of discipline. A major conclusion
and influential work in the field of social in the paper was that in general, in the
class and socialization has been post-war period the middle class has been
Bronfenbrenner's (1958) reviewpaper,"Social- more permissive than the working or lower
ization and Social Class Through Time and classin its childrearingpractices.
Space," which has been widely reprinted in In his analysis of techniques of discipline,
anthologies on social stratification and often Bronfenbrenner had to rely on only six
quoted in texts. Bronfenbrenner'spaper re- studies: a large-scalenational sample taken in
viewsfifteen publishedandunpublishedstudies 1932, a sample of one hundred mothers in
of childrearingconducted between 1932 and Chicago in 1943, and samples taken in the
1957 and reaches important conclusions period 1950-53 in Detroit, Eugene (Oregon),
about variations in such aspects of and the Boston area, with N's rangingfrom
childrearing as demand feeding, weaning, 115 to 372. Review of these studies led him
toilet training,permissivenesstoward impulse to the conclusion that:
expression, freedom of movement, and
The most consistent finding documented
*Of the many people who read the manuscript [in the area of techniques of discipline] is
and offered valuablesuggestions,I would especially the more frequent use of physical punish-
like to thank MelvinL. Kohnand DavidMechanic.I
am also indebted to H. H. Winsboroughfor his ment by working class parents.The middle
methodologicaladvice and to IreneW. Rodgersfor class, in contrast, resortto reasoning,isola-
her very capableresearchassistance.Theresearchre- tion, and. . . "love oriented" discipline
ported here was supportedin part by funds granted
by the National Institute of Mental Health and by techniques( 1958:419).
funds grantedto the Institutefor Researchon Pover-
ty at the Universityof Wisconsinby the Office of This conclusion is based on the finding of a
Economic Opportunitypursuantto the provisionsof
the EconomicOpportunityAct of 1964. The author statistically significant difference by social
retainsresponsibilityfor the analysisand conclusion. class in at least some aspect of discipline(e.g.,
68
CLASSAND PUNISHMENT
IN CHILDREARING 69
use of physical punishment, reasoning,isola- the relationshipbetween social class and the
tion, or "love-orientedtechnique")in three of use of spankingis relativelyweak. Examina-
the six samples available; no analysis of tion of the percentage differences shows a
percentage point differences or measures of range of +6 ("middleclass"more likely to use
associationwas made. Bronfenbrennerargued corporal punishment) to -41 ("working"or
that the class difference has persisted over "lower" class more likely) with a mode of
time, although he found some evidence to around -16 and a mean of about -12 (mode
suggest that the working class might be and mean computedby sample,not row).
movingtoward the middle classpattern. Secondary analysis of retrospective data
Since the publication of Bronfenbrenner's from a national survey taken for the Na-
article, at least three more major studies of tional Commission on the Causes and Pre-
discipline techniques have been reported in vention of Violence gives results compatible
the literature: Washington, D.C. in 1956 with this conclusion. The survey instrument
(N=313), the State of California in 1956 was designed by several sociologists and
(N=809, with an additional812 from Contra political scientists under the general super-
Costa County), and the Boston area in 1959 vision of Sandra Ball-Rokeach; the actual
(N=260). The California study is especially data collection was done by Louis Harris
important because it is based on a large Associates. The sample included 941 whites
carefully drawn sample, and because, except and 195 blacks over age eighteen, in one
for the 1932 national sample and some items hundred clusters in all parts of the United
from a national survey in 1968 (to be re- States (40 other nonwhites were omitted from
ported below), it has the only data based on the analysis). The individualto be interviewed
more than a single metropolitan area. Re- was specified in advance, but no call-backs
grettably, it is also the most obscure of the were made. Preliminaryanalysis of some of
studies; insofar as I can determine it is the data reported on here may be found in
available only as a pamphlet from the Baker and Ball (1969), and Starkand McEvoy
CaliforniaDepartmentof Public Health. (1970). The latter authors do not control for
Updating Bronfenbrenner'sanalysis with age and thus underestimate the extent of
data from the more recent studies, and ex- involvementin and approvalof some formsof
amining the magnitudeof class differencesin physical aggression by persons with low
addition to statistical significance of dif- income or education.
ferences, I was led to a different conclusion Adult respondentswere asked, "As a child?
about the relationship between social class were you spanked frequently, sometimes, or
and the use of physical punishment. Chart I never?";and class of origin was estimatedby
summarizesthe findings from all previously asking, "What classici would you say your
reportedstudies which could be located,' and family was when you were growing
suggests that, although various studies have up-middle class or working class?" The
found a statistically significant relationship, difficulties with these items are obvious, and
need not be outlined. What is noteworthy,
'In addition, Eron, et al., in a study done in however, is that cross tabulation of these
ColumbiaCounty, New York, in 1960, found that items (Table 1) gives frequencies and
parentsdid not differby occupationalgroupin their
use of psychologicalorphysicalpunishmentfor direct percentage point differences which are quite
aggressionin their children(1971:128). A study of consistent with those reported in Chart I.
seventh grade children living mainly in the urban
areas of Central Ohio and CentralNorth Carolina Surprisingly, controlling for sex or age of
was undertakenby Elder and Bowerman(1963) in respondentdoes not affect the finding.
1960. The sample includes over 1,200 cases, fairly Table 2 is a multiple classificationanalysis
evenly divided between boys and girls and the
"middle" (white collar) and "lower" (blue collar) (MCA), which shows the net effect of
classes. The data on physical punishmentare not available variables that can probably be
included in the chart because the collapsingof the assumedto be prior to or contemporarywith
physicalpunishmentindicatorlimited its usefulness,
but dataon the use of symbolicrewardarepresented the spankingexperience. (MCAis one of the
below. The study, primarily concerned with the many kinds of analyses deriving from the
effects of family size, found a relationshipbetween
class and the use of corporalpunishmentcomparable general linear model [Cohen, 1968]. In this
to those reportedin the chart. method, membership in each category of a
70 REVIEW
AMERICANSOCIOLOGICAL

? UZ 'HO

,
H UZ;~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C
4oe0 - 4

LuHLL C-4 + U)

ZULU 0 0H o
(HLLu -4 v CO 0 . -C
)U-V; o q v; oV o. 0

Z0; E- \ - ~ - o .H . . o H

u _ H (H ) 0) LfC4)

HC '4 s
a) 1-4 H a
CO bObo
H

U) ~~~~~~~~~H0 ~~~~~~~~
'-~~~~~~~~~~~~~3
'- ~ 'CTS la)-
LU
> ~ ~ '-'CO
~~C U4
n
0C ~ A FJ~
CI)0U
H
,--
'-40 4- U)Hv o?\'
0)
00
0)
A
0
'-'40)
F-
CO t
OH
H z
X ~g ~ v I--- H
X
0-
rL)
4-~~)
CO,-O 0
-CO
0 U CO
(I)
LO )l)P CO
0
F-
O) ~ 0)m )4
g
4J)
4i
LI)
F U
-, CO
Z; cdul
ooi H u) H V) d) O ) H 0)I)
Z 1 "-.0)
Y
F-II)qH) U U1
4J 0) H

c) :g: ? S L)O1
Q H 4-j0 a 4 0)4

,-D 4C-) C/rC CO


V 0 HO
Z 1uw:4 1- 4-)h- h P. > H>
X LU 000
FY v_ O &O W 0 a 'CO' C0 u
00 00.CO 4~C h 00-H
00Y
:= H
0Z tnH tn :CA4-)
:=> U) Ln 04-H0 0)0)
YP
- P,
0n t'O>. V0)
C
I'd
0tLH 00 00 00
h 0 '-0)n V)h O )
H 0 U '4 "U '0 O
'-' -' -c P4 '- 0
~~~
U)
0 >
H ;~~~~~~~~-4
I~~~~~-l
z
')d0
P4 t
C/C
LI)
c)
CO-H 1 U
4=' Id 4='CO0) 1)
C
O
004- r-4 -4 )
(1)1O
0)
C/CH on
^ 4='
P HO
0)
CO

U) Z; F- F- 4
CO
*H
H
'0
D -F-
P)0
'0
0)
~ 'FO
)H
'0F-I0
0
H OLIII F" -4 -4 ** ,*
A t h-4 >P. O O
dX 00:) oP4 0 ot gC O0 1000
.0 0 0 d bO

0H3
H v 11
Pi4-1
h .
Oc
4-10
I:$.. ..
cdO
~ U ~~~4-)E
~~~ U 1
I
.. ..
cdO-H
C
>
H4-1
c* O
H4-)1I
C 0)
-H4-q10T O .
cdfC S q
)-4 F IH0Hu )4H FIH UH F- 0 -1
?0 ZI) 0) M P wW
M
*H O CO -HOwHW
1.HI--IO
0 : a)
0 01) - 0) '0o
CO o ' $ o i 041'rS O O O0 OEb
cd
0 0 CO~~~~~~~~~~~~~~~.
Cd=r44COOO Vo
C 0 40C

z
Z z

0. 00 00
LUP- h CO ,-Oh -4 0
H cd 0

? U) *H0 U1 0)0 00)0 O

4 O PO 0 LU~~ Z
00 ~~~~

H
UO
~~ 4-~~~~-4
~N ~ *H'-4~~C1
nC B CO
B
00- r4 -4
CO

CO C 0 0)OC 0)
u c/C'-O
Z-C CO I- '-CO- Z F-z0n , I

LU?CO
U t
V) t 4 LI ns~~~~~CC
FH* 0)COt
0)
fcd
0n-
i l
~ ~ ~ ~0~'-
~,D ~H ~U0) ~ ~ ~ wI)-4COr
> X . 0h

Z LU_
0 0 Cd C 0) q OC
wH COO- 'Hw 0)- CO- *H
CLASSAND PUNISHMENTIN CHILDREARING 71

? Z

;g4
U~~~ . .

> a) un tn tnu t t
w
C)* * 0) ) 0 )

0) 0
(n 00

e'
:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cOSt
mI
t 0 0)0)
) | 00 tr- =)) 0) CZ
CD 00 ct g~+)
Z0) N 410) H) -1 t ) 0 4)0 0)
b0)0 to W H e 4--n W W
L) b)41 0 00
H= 0)) L a)Io LLo)
)Ia 0 )-4a)
O c

S U?) S00a;; 0) 0)(H 0 0) cd/ o O


1) ) 0 0) 4-
ZC/) 50)0 >
0).H1U)la 0) 0)0)t-0)H4)o4 0)
00
: H-I
uz~~~~~~~~~~~~~~~~~~~~~~~~~u
4Z 0
rY
.HH
tQ )m
) -4ri * 0i 0O l)0
~--10
4-) -4
0)0t)
uu
1)
0)
-4
U-
-
) - 40)0 )0
t0;
I- Ma)
oO
o0? hH0). h t0)0d . h 4 h4 4 -4 0 H 40)
0 U OH H
0) 0 0)
n h
C) U 0)4)0
:O O
a.,H) HI t 0))0 )-
t Z ;: tt_
)-
ff
h 0) 0>4-.)
4 IH
)
0) h 0)uCd 4)
)) 0)
0)0)
0))
0 S0)H0
. H0))

C0
-0) -4 a
H t-.A0 - 0 ) - 4I
H) s 0: \-\L( ) a - N 4- 1 e' ) h, 0 )r4a FUcd 1
,.
H0) U '- 0H 4 h -' - . - o1 o 00 o ) 00 0,1 a)0)O O h H
Z0) *H-Z H~0) '-< /-00 h 1f- Qgg40)0I 0)

::) o= -4
H D H04
< t )-4 ;1h o-Hh 0~) o-H hOo- = 0)C0)0)LO C
Cu O O :2: -1 4 la) 4 u)
(n a CZ ,,o
0 00 0)t 7ol)
"-i 00 ;-H ) um g
U > 0) U t 4I-
4-J w 0) Lf) 4-I ) n = )04-I.
=I w eH 0)-
U > = O
0~~~~~~~00) P <
0)~~~~~~~~~~~~~~~~~~~~~
m p t
0)
w
0)~~~~~~~~~~~~~~~~4-I)
O~~~~~~~~~~~~~~~~~~
M 0) 01 )
U <~~~~~~~~~~~~~~~~~~~~~~~~~~
H- 0 0) 0
..
11
0)n
) 0C ..cs4 0 c0.
1)Z4-) (
0Z
4 ) l h h4
..ot0) nz$
O
1,c
, (:S 4 2wY
)
0)0)11 0) 0) 00)0 0)-H 0) 0 0)
-H
m 0J 1() 01 ) i() (10) 0
u) u 0 OO4- U 'IZ "
h
0. 4O rl
(L) 0- t- bO 4 H- +- O- bO 4- U 7
0) 0 1)h O0) 9) b4 h) -0)4- V S -0) 0)4 Oq )0)0 0

0) 0)
H0 as o 4-) (L0)I a4 ) 1- 4O0 ) 4 o 4) -4 a) a) o*-
4- )
o H 0) o
*Hi t 4-4 0 )- H - 0)) h 0
0 )0)0 0 0) O
C C -0)0 0 0) -. O :

00
)v :HOago
~Eu C) g O a) Ez 0

0) 0~~~~~~~~~~~~)
C )

Z
U C H~~~~~~~~~~~~~~~~~~~~~~~~~~~~i CH

a., I40)0
OHO. 4-Ii 0 OH 0 0
-41-1?0: O-M) 0)r LO (1) 41~L) h) 4-0 LO
? C ) hC-4) 0 )U I*-)C7)
44
j r * CJ)

O~L. 0) 0). * i0 0 )
ooo m
0 Cd uu u
? O Q B n U)U

-0)
0)
0) co
~~~~0.
0
0)
*H
0)
H a
(A
Z
h O
Z z0l oo
0) _z +o) C 0 7)
z)
\ I'

UZ uz
-403)1 i 0)~0) )
O 0 ) 0))
S : v :I
72 REVIEW
AMERICANSOCIOLOGICAL
CHARTII
z PERCENTOF CALIFORNIA MOTHERSUSING
z PHYSICAL PUNISHMENTAS USUAL METHODOF
$~~~~~~~~~~~~~~~~4.
z ux PUNISHMENTBY SES AND SEX OF CHILD
[5
E-.
" . O (1956 DATA)
04

SEX OF CHILD
EDUCATIONOF MOTHER Male Female
C14 C1. .4.)
U 0
8 Years or Less 42 49
u 1.
9-11 Years 51 50
v$ 12 Years 56 52
44
1-3 Years of College 59 50
4 or More Years of
0 College 56 32
z $4f; * Lr 0 No College (N) 52(352) 51(304)
College (N) 58 (80) 44 (72)
dP dP
N 0^ I OCCUPATIONOF SEX OF CHILD
E-40~~~~~ HEAD OF HOUSEHOLD Male Female
Farm Laborers,
0 4.
cj~~~~~~
,?
~U) ~~~> b
$4
Laborers 49 47
- ^ h h -1 0 Operatives and
0 0 0 Service Workers 50 45
04 Craftsmen and Foremen 55 59
_1 _1 w
Clerical, Sales 47 51
CJ C Cd_ Cd Professional,
dP 0 0
Managerial 59 45
_t
0Utt 0 t
Blue Collar (N) 52(264) 51(216)
CDn 0z O
dP
tn A
dP OU
u
-_4
~ White Collar (N) 55(168) 47(160)
0i t) U
UO
00~~~~~~~~~ 4-
SEX OF CHILD
4444 ANNUALFAMILY INCOME Male Female
~.U)
00*~~~~-1 Under $3,000 54 57
Cd
0h
0P h
e
h h)
Q
4i) 4.) 3,000-3,999 52 42
4,000-4,999 46 57
4 5,000-5,999 58 53
,=-0 < 2~0<fg 6,000-7,999 58 48
cd 8,000 and Over 51 44
0 0
4-4
z z u i Under $5,000 (N) 50(219) S1(170)
4.i S,000 and Over (N) 57(211) 48(202)

z
Source: Heinstein (1964:60-62)
o n 00 0

. 00 nominal variable is treated in the linear


equations as a binary ["dummy"] variable.
0 Thus, the technique is often referred to as
U). "dummy variable regression." Regression
coefficients of the dummy variablesare most
0r easily understood when expressedin terms of
0JN 0
U)0'- Lfl 44W
44
o~~Cd
deviations from the grand mean of the
? uz w cd a)
dependent variable [Andrews, et al., 1967;
0 Melichar, 1965]. Thus, Table 2 [as well as
O U)
4.)
Table 5] shows the deviation from the grand
0
mean for each category of an independent
4-4 variable, net of the effects of all other
u U)
Z 0 )w 0 independent variables in the equation. The
Pm *44'0Q technique does not yield standardized
)z
W
:
UI-
mH
LL U regression coefficients that are easily
interpreted.) The analysis shows religion to
CLASSAND PUNISHMENTIN CHILDREARING 73
Table 1. Percent Spanked as a Child

By Race, Sex, and Parents' Social Class


WHITE BLACK
Middle Working Working
Class Class Total Class Total
MEN
Never 2 4 4 3 3
Sometimes 77 61 65 51 52
Frequently 20 35 31 46 45
(N) (124) (321) (455) (87) (101)

WOMEN
Never 8 7 7 5 5
Sometimes 67 59 62 52 53
Frequently 25 34 30 43 42
(N) (165) (265) (443) (75) (87)

have the most marked effect, with Baptists around the home, and to which response is
being most likely to report that they were likely to be immediate. The other two are
spanked often as a child. Tables I and 2 also unusual situations which involve outside
present data for blacks; limited though they agents and in which discipline is likely to
are, they are just about the only data avail- come after a longer lapse of time and after a
able. Of the studies reviewedabove, almost all more explicit decision as to the appropriate
use samplesof whites only, and none presents type of punishment. Because of the concep-
separatecross tabulationsby race. tual differences,two indiceswere constructed.
In addition to these data on experiences, An index of approvalof spankingin everyday
secondary analysis of data on parents' atti- situationswas constructedby scoringitem (a)
tudes toward the use of spanking under and items (b-l) and (b-2) 0 for a reply of
various circumstancessuggests that approval "No," I for "Not sure,"and 2 for "Yes."The
does not vary much by social class. In the resultingindex has a rangeof 0-6. An index of
Violence Commission survey, respondents approvalof spankingin unusualsituationswas
were asked first whether they approved of constructed in an analogous manner using
spankingin general,and if they did, were then items (a), (b-3), and (b-4). Extended analysis
asked about four specific instancesin which a of the separate indices and of a combined
parent might strike a child. The items were index using all items revealedno major varia-
these: (a) Are there any situations that you tions among them; therefore, only the first
can imaginein which you would approveof a ("index of approvalof spankingin every day
parent spankinghis or her child, assumingthe situations")will be presentedin detail.
child is healthy and over a year old? If yes: Contrary to popular belief, among whites
(b) Would you approve if the child: (1) was poorly educated parentshave the highest rate
noisy and getting on the parents'nerves;(2) of outright rejection of spanking.Sixteen per-
had been disobedient all day; (3) had been cent of parents in this group, as comparedto
expelled from school; (4) had broken a law? no more than five percent of parents at any
In each case, the respondent could reply other level of education score "0" on the in-
"Yes,"1"No," or "Not sure." dex (see row 1 in Table 3). The distributionof
Conceptually,two quite different types of index scores for adults without children (not
examples were presented. Two instances deal shown) is similar.The highest rate of rejection
with situations which may commonly arise of spankingfor any college educatedsubgroup
74 AMERICANSOCIOLOGICAL
REVIEW
Table 2. Differentials in Childhood
Spanking Experiencea by gelect-
but then disapproveof spankingin each of the
ed Demographic Variables four situations. However, this combination
does not affect the relationship. Even if one
Table shows deviation from grand mean were to make the hostile assumptionthat all
(2.26) controlling for other variables
in the table. N=1136. people who answered "No" to item (a) did
not know what they were talking about and
N Deviation
should be dropped from the analysis, the re-
SEX sulting percentages do not change the thrust
Male (578) +,02
Female (550) -.02 of the analysispresentedin this paper.
At the high end of the approval of
RACE spankingscale, for whites with children there
White (941) - .01
Black (195) +.08 is wide variationby educationbut no clearre-
lationship (Table 3). The difference between
CHILDHOODRESIDENCE
"Farm" (351) +.03 the grade-school educated and the higher
"Town"t ( 3 14) - .01 educated groups in approval of spanking is
"Small City" (164) 4.02
"Big City" (291) - .04 narrowed,but grade-schooleducated parents
still score relatively low. Paradoxically,those
PARENTS' SOCIAL CLASS least likely to score high on spanking are
"Middle Class" (315) -.05
"Working Class" (768) +.02 college drop-outs, while those most likely to
score high are those who graduated from
CURRENTRELIGION
Baptist (283) +.13 college. For whites without children(data not
Methodist (135) - .07 shown), the pattern is similar,although there
Lutheran ( 69) - . 12 is less variation among the college educated.
Episcopalian ( 38) +.01
Other Protestants (213) - .00 Cross tabular analysis of income, age, sex,
Catholic (279) -.03 regionof residence,and religionrevealedsome
AGE patternsof interest,2but had no effect on the
18-25 (154) +.00 generalfinding of no fundamentaldifferences
26-35 (232) 4.03
36-45 (230) -.03 among white parentsof differentsocial classes
46-60 (282) +.03 in their approvalof spanking.Unfortunately,
Over 60 (238) -.03 occupationaldata were not coded by the con-
aMultiple Classification (Dummy Vari- tractor, and this indicator of class could not
able Regression) Analysis. Spanking be analyzed.
experience is scored: 1=Never, 2=Some-
times, 3=Ftequently.
The data for blacks are considerablymore
consistent with conventional notions than are
bBlacks weighted .53, but unweighted those for whites; they are also consistent with
N's are shown. Missing data were
assigned the modal value if a variable the data on actual experiences reported
had a clear mode. If not, missing earlier. Table 4 shows a clear downwardpro-
data were included in the regression
as a separate category but are not gressionin approvalof spankingby education
shown in the table. In such cases N's for black parents, and for respondentswho
do not add to 1136 and all categories have not at least graduatedfrom high school,
shown could deviate from the grand
mean in the same direction. the percentage scoring high on the index is
substantially higher than those for the com-
is for non-parentswith post-graduateeduca- parable white groups. However, the 79 per-
tion, but even for this small group only nine cent figure for the grade-schooleducated is
percentreject spanking.One may question the
validity of this finding by askingwhether per-
2Males are generally more approvingthan fe-
sons with highereducation were not just being males,and youngerparentsare moreapprovingthan
more "objective,"realizingthat there must be older ones, especially at higher levels of education.
some instances in which they would approve little Analysis by region and area of residencerevealed
variation, except that the South (especially
of spanking.To examine this issue further, we urbanareas) had a higher rate of approval(52 per-
can combine the respondentswho said they cent as comparedto an averageof 36 percent for
other regions).The effects of religionare not easily
would never approve of spanking with those summarized,but in generalBaptistsare highandde-
who gave approval to the general question, vout Catholicslow.
CLASSAND PUNISHMENTIN CHILDREARING 75
Table 3. Percent Who Would Approve of Spanking in Everyday Situations

By Education
For White Parents With Children Under 18

Some
Grade High High Some College Post
Score on Index School School School College Graduate Graduate

0 16 5 3 4 0 4
1-2 LOW 9 6 7 3 11 6
3-4 MED 35 43 49 60 39 50
5-6 HIGH 40 46 41 33 50 40

(N) (43) (89) (193) (92) (36) (28)

based on a quite small N and is unstable;the indices. All blacks and whites in the sample
figure for grade-schooleducated blacks with- are included in this analysis. The results are
out children is lower (65 percent, N = 45). not fully consistentwith the precedingdiscus-
For blackswith college experience,the rate of sion because the MCA program used, like
approval is lower than that of whites with most standardregressionprograms,suppresses
similarexperience.(The percentagefor college interaction effects. In particularthe MCAin
graduateswith childrenis unstable, but since Table 5 shows a tendency for respondents
that for persons without children [not with post graduate education to be low in
shown] is similar, it lends confidence to the approvalof spanking, and for education and
31 percent figure.)Again,the distributionsby income to have opposite effects on approval.
income and for adults without children are
similar, and extensive additional analysis did
DIFFICULTIES IN INTERPRETING THE STUDIES
not change the generalfindingsreportedhere.
Table 5 is a multiple classificationanalysis Although there have been about ten sur-
(MCA), showing the net effect of various veys on punishmentpractices,each is limited
demographic variables on the two spanking either by the narrowrangeof its items or by

Table 4. Percent Who Would Approve of Spanking in Everyday Situations

By Education
For Black Parents With Children Under 18

Some
Grade High High Some
Score on Index School School School College+

0 0 0 3 23

1-2 LOW 0 5 7 0

3-4 MED 21 32 43 46

5-6 HIGH 79 63 47 31

(N) (14) (41) (30) (13)a

aPercentages based on such a small number of cases are unreliable.


76 REVIEW
AMERICANSOCIOLOGICAL
Table 5. Differentials in Indies of Approval of Spankinga by Selected
Demographic Variables

Table shows deviation from grand mean, controlling


for other variables in the table. (Grand means:
index A=4.34, index B=4.27; N=1136)

Deviation
A. Index B. Index
of Everyday of Unusual
N Situations Situations
SEXbb
Male (578) .01 02b
Female (550) .01 .01

RACE
White (941) -.04 -.05
Black (195) .39 .43

CHILDHOODRESIDENCE
"Farm" (351) .08 .10
"Town" (314) .00 -.11
"Small City" (164) -.16 -.17
"Big City" (291) -.02 .07

PARENTS' SOCIAL CLASS


"Middle Class" (315) .03b -.01
"Working Class" (768) .04 .05

EDUCATION
Grade School (243) . 07 .17
Some High School (233) .04 -.05
High School (352) .06 .08
Some College (182) -.14 -.23
College ( 72) - .02 .05
Post Graduate ( 54) -.33 -.33

CURRENTRELIGION
Baptist (283) .12 .29
Methodist (135) .10 -.16
Lutheran ( 69) .07 .26
Episcopalian ( 38) -.23 -.07
Other Protestants (213) -.04 -.15
Catholic (279) -.03 -.06

CURRENTREGION OF RESIDENCE
Northeast (312) -.03 -.03
South (319) .22 .21
Midwest (322) -.23 -.15
West (183) .11 -.03

CITY SIZE
SMSA (316) -.00 .06
Suburb (291) -.10 -.10
Town: 10-50,000 (128) .04 -.20
Town: Less than 10,000 (118) -.39 -.13
Rural (283) .24 .19

CURRENTINCOMEOF HEAD OF HOUSEHOLDC


$ 1-2 999 (178) -.33 - .19
3-4,999 (169) -.23 -.14
5-6,999 (197) .06 .01
7-9,999 (298) .08 .07
10-14,999 (205) .24 .09
15-19,999 ( 57) .40 .34
20,000 and over ( 32) .48 .44
CLASSAND PUNISHMENTIN CHILDREARING 77

Deviation
A. Index B. Indlex
of Everyday of Unusual
N Situations Situations
AGE
18-25 (154) .03 .02
26-35 (232) .42 .31
36-45 (230) .19 -.03
46-60 (282) - .22 -.03
Over 60 (238) -.34 -.25
CHILDRENUNDER 6
Have Children Under 6 (322) -.11
All Children Between 7 and 18 (252) -.12
No Children Under 18 (562) .12
TEENAGERS
Have Teenage Children (212) -.09
All Children Under 13 (362) .01
No Children Under 18 (562) .03

aMultiple Classification (Dummy Variable Regression) Analysis. Range on index


A and B is 0 - 6. Content of indices discussed in text.
bBlacks weighted .53, but unweighted N's are shown. Missing data were assigned
the modal value if a variable had a clear mode. If not, missing data were
included in the regression as a separate category but are not shown in the
table. In such cases N's do not add to 1136 and all categories shown could
deviate from the grand mean in the same direction.
cIt was assumed that respondents for whom income of head of household was not
recorded had only one wage earner in the family and family income was substi-
tuted. If family income was also missing, the mean income for the respondent's
educational group was assigned.

some aspect of its sample. In addition, the 1972; Parke, 1970; Solomon, 1964). Even if
fact that the time periods of the surveysare the "usual" method of punishment is the
not comparable introduces an additional same for all strata, the term is vague enough
source of variability, although it also opens for a wide variation in the extent to which
the possibility for a trend analysis. This psychological techniques are used. These
section will considereach of these aspectsand problemsabout which the presentanalysiscan
attempt to assess their implications for an do nothing; however, they do point to the
estimate of the relationship between social need for furtherresearchbefore a firm conclu-
classand the use of corporalpunishment. sion on social class variationscan be reached.
For the studies available, regroupingby
Natureof the Indicators different indicators does not affect the con-
All of the studies are limited by their clusions above. Although the indicators do
reliance on self-reportedsurvey data, almost vary, most of them are concerned with the
always involving a report of customaryprac- frequent use of spanking.Whenthe few that
tices or of likely response to a hypothetical are different are separated out, the relation-
situation. Few studies ask about the actualuse ship looks no stronger. In addition, many of
of spankingin the previous week or month. the studies contain some items which deal
More generally, previous studies tell us little specifically with psychologicalmethods;these
about what might be called the "punishment can be used to supplement the presentation
atmosphere,"i.e., the particularway in which above. Theory and conventional wisdom
different methods are combined by different would predict that a much largerpercentage
parents,the harshnessof discipline,the condi- of middle and upper class parents would
tions underwhich the child is punished,or the report using such methods as "reasoning"
way punishment is perceivedand interpreted with the child who has acted improperlyand
by the parent and the child (cf. Johnston, extensive use of reward or praise for proper
78 REVIEW
AMERICANSOCIOLOGICAL
behavior.Here again, although the data often exists in the quality and breadth of the
show the expected relationship,the findings samplesused in the studies. Variousproblems
are not uniform and the relationships are of comparability of the earlier samples are
often very weak. Anderson's 1932 data show describedin Bronfenbrenner(1958) and need
the relationshipclearly enough: 73 percent of not be dealt with here. The importantpoint is
Class I, versus 43 percent of Class VII used that the quality of the sample does not seem
reasoning;and 44 percent versus 19 percent to be systematically related to findings. Per-
used deprivation of pleasure. However, in haps the most serious drawbackof the sam-
1943, Davis and Havighurstfound that 57 ples is the lack of attention to the social class
percent of their working class respondents, gradient;for example, severalsimply compare
versus 53 percent of the upper-middleclass two available groups of parents thought to
respondents, thought that reasoning was a represent different social classes. The best
successful way to get childrento obey. (The sample since the 1932 national sample is
relationship in that study was significantly Heinstein's study, based on data collected in
reversed,however, on the question of use of 1956 through a statewide Californiasample
reward or praise, in which the upper-middle stratified to accurately reflect all income
class gave 78 percent approval, while the groups and geographicareas. The findingsby
working class gave only 53 percent.) In education are summarizedin ChartI; but as
Eugene, Oregon, in 1950, Littman, et al., noted there, grade-school educated mothers
found very little difference between the were actually less likely to emphasize
middle and lower classparentsin their sample spankingthan were other mothers without a
on the use of reasoning,scolding, deprivation college education. Heinsteinfinds a curvilinear
of privileges or possessions, isolation, or relationship between mother's education or
ignoringor distractingthe child. Searset al. in the occupation of head of household and the
1951, found virtuallyno differencesbetween use of physical punishment, and no relation-
the upper-middleand upper-lowerclasses on ship between family income and the use of
the extent of use of reward,praise, isolation, physical punishment. Because the study is
reason, scolding statements involving with- ratherinaccessible,the majortables are repro-
drawal of love, or deprivation of privileges duced in ChartII.
(Maccobyand Gibbs, 1954). Heinstein's1956 Another problem with the samplesis wide
Californiasurvey, which asked mothers open- variationin the age of the child referredto in
endedly about their "usualmethod of punish- the questions. One may argue that with very
ment" showed "apparently little use of so- young children, reasoningis impractical,and
called constructive methods" (1964:64); but the use of spankingwould be relativelyhigh in
this findingwas partly attributedto the young all social classes;then, as the child gets older,
age of the children studied. The 1960 study differences in technique would become
by Elder and Bowerman(see note 1 above) greater.3The findings for some of the studies
found only about a five percentage point are consistent with this hypothesis; but
differencebetween the "middleclass" and the others, particularly those of Kohn and
"lower class" in the use of symbolic rewards Littman, are at variancewith it. The Violence
by mothersor fathersin raisingboys. For girls Commission items on approval of spanking
the gap was somewhat larger. The major lend some additional suggestive evidence on
recent evidence for a relationship between this point. As noted earlier,two indices were
class and the explicit use of psychological developed, one referring to everyday misbe-
punishmentcomes from Millerand Swanson's havior, such as a child's "getting on the
1953 work in Detroit. The proportion of parents' nerves" and one referring to more
parents using at least some form of psycho- unusual extreme occurrences, such as the
logical manipulation is the reciprocal of the child's havingbroken the law. The hypothesis
rates of corporal punishment reported for would seem to predict that for parents of
Detroit in ChartI.
Nature of the Samples 3This argument, while tenable, overlooks alterna-
tives to spanking such as the ignoring or tolerance of
There is no question that wide variation the undesired behavior, or the isolation of the child.
CLASSAND PUNISHMENTIN CHILDREARING 79
young children, there may be only small psychologicallyorientedmiddle or upperclass
differences by social class on the index of parent would consider spanking especially
approval of spanking for everyday misbe- futile in such instances, and reasoningwith
havior,while for parents of older childrenthe the child and understandinghis motivations
differences would be substantial. Analysis of especially important. Analysis shows, how-
the index by parent's education or income, ever, that if anything, college graduateswith
controlling for age of children does not teenage children tend to have a higher mean
support these hypotheses (see, e.g., Table 6a). score on the index than other parents of
For example, white college educated parents teenagers (Table 6b), with a similar pattern
of childrenundersix have a highermean score holding for income. (For blacks there is a
than other white parents on the index of tendency for less-educated [or poorer] par-
approval of spanking for everyday misbe- ents to score higher on the index, irrespective
havior; and, while there is a slight reversalin of the control for age of child.)
this trend for parentswith no childrenunder
TrendOver Time
six, the difference is not statistically signifi-
cant. (For blacks, parentswith young children At the time Bronfenbrennerwrote, there
are more likely than those with older children seemed to him to be a relationshipbetween
to approveof spanking;but the differenceby social class and punishment technique which
education noted earlier is unchangedby the suggested that the difference was narrowing.
control variable.)For the index of approvalof However, now that additional data are avail-
spankingfor unusual situations, the hypothe- able, it seems instead that over the years find-
sis would seem to predict substantial differ- ings have fluctuated around zero, with some
ences by social class for parents of teenagers. tendency to be positive. Looking over the
Conventional wisdom would hold that the data in Chart I, one sees that the most consis-

Table 6. Mean Score Indices of Approval of Spanking

For Parents with Children Aged 0-18


By Race, Education, and Age of Child

WHITE BLACK
Less
Some Than High
Eighth High High Some College High School
Grade School School College or More Total School or More Total
a. Mean score, index of approval of spanking for everyday
misbehavior

Have child 3.41 4.46 4.61 4.45 5.00 4.53 5.45 4.30 4.89
aged 0-6
(17) (41) (113) (60) (35) (266) (29) (27) (56)

All 4.13 4.64 4.36 4.13 4.10 4.33 4.92 4.25 4.67
children
aged 7-18 (23) (47) (80) (31) (29) (210) (26) (16) (42)

b. Mean score, index of approval of spanking in unusual situations

Have child 3.65 4.38 4.21 3.69 4.26 4.11 5.38 4.53 5.03
aged 13-18
(20) (37) (70) (26) (23) (176) (21) (15) (36)

All 4.75 4.53 4.40 4.15 4.78 4.44 5.18 4.54 4.89
children
under 13 (20) (51) (123) (65) (41) (300) (34) (28) (62)

N.B. Range of each index is 0-6.


80 AMERICANSOCIOLOGICAL
REVIEW
Table 7. Percent Spanked as a Child

By Race, Parents' Class and Age


WHITE BLACK
18-30 31-50 51-64 65+ Total 18-50 51+ Total
MIDDLE CLASS
Never 4 5 6 9 6 0 0
Sometimes 70 78 63 64 71 73 69
Frequently 26 17 31 27 23 27 (1) 31

(N) (90) (94) (51) (44) (290) (15) (1) (16)

WORKING CLASS
Never 2 3 6 14 5 5 3 4
Sometimes 65 67 51 50 60 51 53 52
Frequently 34 30 43 36 34 44 44 44

(N) (113) (226) (132) (101) (591) (102) (57) (163)

tent finding is not the difference between the youngest. It is very difficult to speculate
social classes, but the relatively small differ- on the meaning of the finding for the group
ence that appearsin almost all the samples, over sixty-five4 who were reared in the late
independent of time. The Violence Commis- 1800's and in the early part of this century.
sion retrospective reports of childhood On the one hand, it tends to contradict the
experiencessupportthis conclusion. Whenthe commonly held view that "in the olden days"
data presented in Table 1 are brokendown by parentswere "tough" with their children.On
age group (Table 7), class differences are re- the other hand, no matter what the finding,
markablystable. In addition,regressionanaly- recollections of events which occurredover a
sis shows the correlation between parents' half-centuryago probably have little validity.
social class and spanking experience to be Comparison of the first two columns for
almost unaffectedby age. (The same is true of whites also puts in question a commonly held
the relationship between race and spanking view. Respondents aged eighteen to thirty
experience.) were reared primarilyin the post-warperiod,
Moreover, contrary to the assessment of in which the "progressive"views of Dr. Spock
many commentators, both scholarly (e.g., have dominated the childrearingliterature.
Miller and Swanson, 1958; Bronfenbrenner, Spock's views are widely believed to have
1958, 1961) and popular, there is evidence greatly influenced spankinghabits, and many
over time of at best a moderate ratherthan a people believe that his advicehas bred a gener-
strong shift away from the use of spanking. ation of "brats." The Violence Commission
Table 7 suggeststhat the relationshipbetween data seem to indicate, however, that although
age and spankingexperience is somewhatcur- there probably was a shift in the use of spank-
vilinear for whites; however, the zero order ing from the 1920's onward, there may have
multiple correlation coefficient between been no dramaticshift away from spankingin
spankingexperience and age as a dummy vari-
able is only .08 (p=.32). For whites of both 4 In Table 7 age groups were collapsedin a way
social classes, persons aged thirty-one to fifty which preservedthe pattern of the data. For the
(raised roughly in the second quarterof this middle class, the cell for age sixty-one to sixty-five
(N=8) shows a 25 percentfrequency,so the observa-
century) report the lowest percentage of tion in the text for the group over sixty-five may
"spanked frequently" while the groups on pertain to this groupaswell. But the N is really too
smallto give a reliablepercentage,and at any rate a
either side are higher, and the oldest group variationof five years is of no consequenceto the
(over sixty-five) is at about the same level as discussion.
CLASSAND PUNISHMENTIN CHILDREARING 81
the 1940's. Note, too, that overall there is no be of great theoretical or practical signifi-
relationshipbetween parents'social class and cance.' For all whites and blacks in the Vio-
the percent reportingthat they were "never" lence Commissionsurvey, the zero order cor-
spanked.Moreover,this responseis positively relation between parents' social class and
related to age, and is given by only four per- spankingexperienceis -.10 (significantat .001
cent of respondentsunderfifty. level), controlling for race and age (as a set of
Unfortunately, neither the collection of dummy variables)in a regressionanalysis the
studies in the literature nor the survey data beta is -.09 (significantat .002 level).6 Separa-
here give definitive data on the question of ting respondents by race, the correlation is
changesin the use of spankingover time. Be- slightly higher for whites than for blacks.
sides the inferences of Bronfenbrennerand of Combiningthis finding with post-hoc correla-
Miller and Swanson, a few studies, based on tional analysis on data reported in the litera-
limited samples,have found a relationshipbe- ture, an estimate of somewherein the .10 to
tween exposure to professional childrearing .15 range emerges,indicatingthat social class
advice in the post-war period and changesin may explain between one and two percent of
childrearing practices. (See references in the variance.Moreover,in almost all samples
Bronfenbrenner [1961], which, drawing on the modal response for all social classes stud-
Children'sBureau pamphlets and articles in ied is the same; it is thus inappropriateto
popular magazines,document the changes in characterizethe classesas being different. (On
advice. White [1957] has data on the relation- the other hand, the findingsare not irrelevant,
ship.) On the other hand, note that as early.as given that social class may still be one of the
1936 Andersoncould write: "One of the most best predictorsavailable.)
universally condemned methods of punish- For race, there are much less data;but they
ment for children is spanking..."(1936:213). do indicate that race has an effect indepen-
At any rate, the availabledata are sufficiently dent of social class. The tables presented
ambiguous to cast doubt on the prevailing above show this more clearly than correlation
view. or regression analysis do, because the latter
Because of the relatively small sample of are sensitive to the 9:1 disparityin N's. (The
blacks and the fact that few blackshave been zero-order correlation between being black
allowed to attain middle class status, the data and reporting physical punishmentas a child
include only sixteen blacks raised in middle is .08 [significantat .01 level] ; controllingfor
class homes. Fifteen of these respondentsare social class the beta is .06 [significantat .03
under age fifty, and as a group their experi- level]). However, data presented below sug-
ences are comparable to those of younger gest that insofar as approvalis concerned,the
middle class whites. Although the N's are too greaterlikelihood of black parentsto support
small to give stable percentages,the complete the use of physical punishmentis more likely
age breakdowndoes seem to indicate a trend the product of differences in belief about the
toward less spankingin the black middle class. efficacy of spankingratherthan of differences
For blacks raised in working class homes, the in attitude about violence.
percentageof respondentsreportingfrequent The conclusionon the effects of social class
spankingis essentially the same for all age has important implicationsfor severalareasof
groups, at 44 percent, averagingabout 10 concern in sociology. Many statementson the
percentagepoints higherthan the white work- relationshipbetween social class and the use
ing class and 21 points higher than the white of spanking have implied not just statistical
middleclass.
I The lack of clearrelationshipsbetween classand
DISCUSSION childrearingtechniques is also evident with other
aspects, such as feeding and toilet training.See, for
Giventhe data currentlyavailable,the best example, Bronfenbrenner (1958, passing) or
conclusion about the relationship between Heinstein (1964, passim).
6
Physical punishment is scored as a three
social class and the use of physical punish- category continuous variable,with "never" = 0,
ment is that there is indeed some correlation, "sometimes"= 1, and "frequent"= 2. Blacks are
but that it probably is not strong enough to weighted.53.
82 AMERICANSOCIOLOGICAL
REVIEW
significance,but the existence of a very strong gument has been advanced by several theo-
relationshipbetween those variables.This im- rists, most notably Miller and Swanson
plication has then become the grounds for (1960), who argue that in corporal punish-
using spanking to explain various behaviors ment, the punisheris clearly identified and a
also thought to be much more common in the physically aggressiverole model is established;
lower or working class than in the middle nius the child grows up oriented to express
class. A numberof these implicationswill now aggressionoutwardly.By contrast,psychologi-
be treatedin turn. al manipulation,which is based on appealsto
guilt and is said to result in the development
of a strongsuper-ego,is postulated as leading
WorkingClassAuthoritarianism to turning aggression upon the self (cf.
Feshbach, 1970; Yarrow et al., 1968). Gold
In Lipset's classic discussion of "working (1958) and Coser (1963) have used this rea-
class authoritarianism,")"authoritarianfamily soningto explain the wide disparityby social
patterns" are hypothesized as one of the six class for the officially recorded rate of honi-d
most important elements contributingto the cide, by arguingthat members of the lower
authoritarianpredisposition. The salience of class are more likely to be physically violent
family patterns to the theory can be seen in because they are more likely to have been
one of the article's concluding statements: punishedcorporally.Coser,for example,relies
"To sum up, the lower classindividualis like- on the same oft-quoted statement by
ly to have been exposed to punishment,lack Bronfenbrennercited above.7 Although the
of love, and a generalatmosphereof tension model proposed by these authors has some
and aggressionsince early childhood-all ex- validity, it is important to note that the cor-
periences which tend to produce deep-rooted relation between social class and punishment
hostilities expressedby ethnic prejudice,polit- technique is not strongenough to supportthe
ical authoritarianismand chiliastic transvalua- inference that lower class personswho arevio-
tional religion"(1960:114). Yet the basic evi- lent were spanked as children. Individual
dence presented on this point is Bronfen- rather than aggregatedata are needed to test
brenner's conclusion on punishment tech- the model.
niques quoted at the beginningof the present
paper. The re-evaluationof the evidence on Relationshipto ChildAbuse
spanking has obvious implications, then, for
the theory of working class authoritarianism. The use of corporal punishmentis widely
Lipset also argues that "a further link be- believed to be associated with certain unde-
tween such [physically oriented] childrearing sirable behaviors and attitudes in the adults
practicesand adult hostility and authoritarian- who practice it. Gil, in an importantstudy of
ism is suggestedby the findingof two investi- child abuse, finds that officially recordedacts
gators in Boston and Detroit that physical of physical abuse against children are much
punishments for aggression,characteristicof more likely to be committed by low status
the working class, tend to increaseratherthan parents.He arguesthat for all social classes,a
decrease aggressive behavior" (1960:107). major cause of abuse is a generallypermissive
Actually, the work cited by Lipset does not attitude toward the use of physical force in
claim that physical punishmentfor aggression childrearing (apparently including spanking)
is "characteristic"of the working class; and, in American society. Since "studies of child-
in addition, further research has cast some rearingpatterns have found a strong associa-
doubt on the strength of the relationshipbe-
7 Both Coser and Lipset make the common error
tween physical punishment and subsequent
of quoting Bronfenbrenner somewhat erroneously.
aggression(Yarrowet al., 1968). The oft-uoted passage actually begins "The most
consistent finding documented in Table 10 is the
more frequent use of physical punishment by work-
ing class parents." Many writers leave out the refer-
TendencyTowardPhysical Violence ence to "Table 10," and the cursory reader could be
misled into thinking that Bronfenbrenner viewed the
Spankinghas also been thought to generate data on physical punishment as the most consistent
a propensity to be physically violent. This ar- of all the aspects of childrearing he examined.
CLASSAND PUNISHMENTIN CHILDREARING 83
tion between low socioeconomic status and slapping his wife, a man fighting with a
the use of physical means in discipliningchil- stranger, and teenage boys getting in fights
dren" (Gil, 1970:127), Gil finds it consistent (Baker and Ball, 1969; Stark and McEvoy,
that abuse would also occur disproportionate- 1970). The items on teenage fighting can be
ly in these families. In support of the state- combined to form an "index of support for
ment quoted above, only Miller and manliness" since they ask about such situa-
Swanson's second (1953) sample is cited. tions as fighting back when challenged or
Chart I shows that this particularsample had picked on. The index was constructed by
by far the most extreme findingin forty years scoringthe following items 0 for "No," I for
of research. Although little is known about "Not sure," and 2 for "Yes": (a) Are there
the general childrearingmethods of persons any situations you can imagine in which you
who on some occasion abusea child, data pre- would approve of one teenage boy punching
sented by Gil show that in less titanone-third another? If yes, or not sure: (b-I) If he didn't
of the cases studied "self-definitionof the per- like the other boy? (b-2) If he had been ridi-
petratoras stern, authoritativedisciplinarian" culed and picked on by the other boy? (b-3)
was a factor (1970:128). Although the clear If he had been challengedby the otherboy to
majorityof acts of abusewere in responseto a a fist fight'?(b-4) If he had been hit by the
specific act committed by the child, Gil does other boy? The rangeis 0-10.
not show that the reaction was relatedto dis- The index of support for manliness is
cipline rather than to stress. (cf. Erlanger, somewhat correlatedwith the other indices of
1971). approvalof aggression;for example, for white
parentsof childrenundereighteen, its correla-
Subcultureof Violence tion with approval of a man punching a
strangeris .29 and with approvalof marital
Finally, the use of corporalpunishmenthas fighting is .19. For black parents, the inter-
been believed to be part of a general pattern correlations are .53 and .16. However, for
of the use of physical aggressionin interper- both races, the correlationbetween the manli-
sonal relationships. Wolfgang (1958, 1967) ness index and approval of spankingin un-
has suggested that blacks and low income usual situations is markedlylower, at .08 for
whites are part of a "subcultureof violence" whites and virtually zero (-.01) for blacks.
which emphasizes the use of physical aggres- Moreover, for black parents of teenagers,
sion in both parent-childand peer encounters. approval of "manliness" and approval of
In general, this thesis is undemonstratedem- spanking actually are negatively correlated
pirically (Erlanger, 1973), and the relatively (-022). Compared to whites, blacks are also
small differences by class in the behavioral disproportionatelylow in their approval of
data discussed above seem too small to sup- fighting to demonstrate "manliness."For the
port the thesis of a class subculture.Moreover, index of approval of spanking for everyday
the concept of subcultureultimately refersto misbehavior, the correlations are similar.
values, not behavior;and if the data on ap- Thus, the disproportionately high rate of
proval of spankingby class are taken to indi- black approval of spanking reported earlier
cate such values, there is againno supportfor seems best explained by a rejection of norms
a class subculture. of violence, and not by adherence to a
Some readersmay see support for the sub- subculturalvalue system supportingviolence.
culture of violence thesis in the differencesin
childhood experiences and the differencesin QualifyingFactors
approval of spanking by race. The approval Althoughcurrentlyavailabledatado not show
items may be examinedin more detail by re- a strong relationshipbetween social class and
lating them to approval of other forms of choice of punishmenttechnique, this need not
aggression,using other items from the Vio- mean that parents of different social classes
lence Commissionsurvey.The surveyincluded are alike in their use of a disciplinetechnique.
questions on the conditions under which the Kohn, for example, found no significantdif-
respondent might approve of various acts of ferences by class in the use of spanking,but
interpersonal aggression, such as a husband found that
84 AMERICANSOCIOLOGICAL
REVIEW
The principal difference between the sitive to absolute percentagesand the strength
classes is in the specific conditions under of correlations,ratherthan primarilyto estab-
which parents-particularly mothers-pun- lishing statisticalsignificance.In addition,null
ish children's misbehavior.Working class findingsbased on good quality data cannot be
parentsare more likely to punishor refrain overlooked. Although future research could
from punishingon the basis of the direct still confirm the class or ethnic generaliza-
and immediate consequences of children's tions, it is much more likely that class and
actions, middle class parentson the basis of ethnic differences in technique are relative,
their interpretationof children'sintent in not absolute, and that the practicesof parents
acting as they do. (1969:104) in different groups greatly overlap.Certainly
the data presentlyavailableimply as much.
Kohn finds these differences to be part of a
more general tendency for working class par- REFERENCES
ents to value conformity over self-direction,
Anderson, John E.
while middle class parentstend to have the re- 1936 The Young Child in the Home. White House
verse preference.He concludes that these dif- Conference on Child Health and Protection,
ferences are the product of differencesin edu- Committee on the Infant and Pre-school
cational attainment and in opportunities for Child. New York: Appleton-Century.
self-direction in one's occupation and are Andrews, Frank M., James N. Morgan, and Joghn
A. Sonquist
probably not strongly related to basic differ- 1967 Multiple Classification Analysis. Ann Arbor,
ences in personality. Michigan: Survey Research Center, The
Moreover,spankingsmay be administered University of Michigan.
differently. By "spanking," one parent may Baker, Robert K. and Sandra J. Ball
mean a firm but quick slap on the child's 1969 Mass Media and Violence. Report to the
National Commission on Causes and Preven-
bottom, while another may mean a repeated tion of Violence. Washington, D.C.: U. S.
slapping, perhaps administeredby the father Government Printing Office.
when he returns home from work. Although Bronfenbrenner, Urie
there is reason to believe that these differ- 1958 "Socialization and social class through time
and space." Pp. 400-25 in E. C. Maccoby, T.
ences are relatedto social class, they are prob- M. Newcomb, and E. L. Hartley (eds.),
ably not as great as many people believe. For Readings in Social Psychology (3rd ed.).
example, the large-scalemidtown Manhattan New York: Holt, Rinehart, and Winston.
study found that althoughthe use of strapsin 1961 "The changing American child, a specula-
tive analysis." Journal of Social Issues 17
spankingis much more common among par- (1):6-1 8.
ents with low income, over 80 percent of low 1968 "Multiple regression as a general
income parentsin all ethnic groupsdo not use data-analytic system." Psychological Bulle-
this method (Langner, 1969). Similarly,there tin 70 (December):426-43.
may be class or ethnic group differencesin the Coser, Lewis A.
1963 "Violence and the social structure." Science
general atmosphere in the home, in the ex- and Psychoanalysis 6:30-42.
pression of displeasureby grouchiness, the Davis, Allison and Robert J. Havighurst
assumption of an air of martyrdom,or other 1943 "Social class and color differences in child-
changes in behavior, or in the interpretation rearing." American Sociological Review II
(December):698-710.
of various parentaldiscipliningtactics by the Elder, Glen H., Jr. and Charles E. Bowerman
child (cf. Kohn, 1969:103). Clearly,future re- 1960 "Family structure and child-rearing pat-
searchmust be more sensitiveto subtle differ- terns: the effect of family size and sex com-
ences in punishmenttechniques.This research position." American Sociological Review 28
should also explore the salience of various (December): 891-905.
Erlanger, Howard S.
dimensions of class, and of explanatoryvari- 1971 Review of David Gil, Violence Against Chil-
ables that operate across class lines. Some dren. Social Service Review 45
previouscommentaryhas made the mistakeof (December): 514-15.
assuming or implying that class or ethnic 1973 "The empirical status of the subculture of
violence thesis." Department of Sociology,
groups can be characterizedby their methods University of Wisconsin, Madison.
of punishment. To avoid this tendency to Eron, Leonard D., Leopold 0. Walder, and Monroe
overgeneralize,future researchshould be sen- M. Lefkowitz
CLASSAND PUNISHMENTIN CHILDREARING 85
1971 Learning of Aggression in Children. Boston: ment. New York: Harcourt, Brace, and
Little Brown and Co. World.
Feshbach, Seymour McKinley, Donald G.
1970 "Aggression." Pp. 159-259 in Paul H. 1964 Social Class and Family Life. New York:
Mussen (ed.), Carmichael's Manual of Child The Free Press.
Psychology, Vol. 2. New York: John Wiley Melichar, Emanuel
and Sons. 1965 "Least squares analysis of economic survey
Gil, David G. data." Proceedings of the Business and Eco-
1970 Violence Against Children. Cambridge: nomics Section of the American Statistical
Harvard University Press. Association.
Gold, Martin Miller, Daniel R. and Guy E. Swanson
1958 "Suicide, homicide, and socialization of ag- 1958 The Changing American Parent. New York:
gression." American Journal of Sociology John Wiley and Sons.
63 (May):651-61. 1960 Inner Conflict and Defense. New York:
Havighurst, Robert J., and Allison Davis Henry Holt and Company.
1955 "Comparison of the Chicago and Harvard Miller, S. M. and Frank Riessman
studies of social class differences in child 1961 "Working class authoritarianism: a critique
rearing." American Sociological Review 20 of Lipset." British Journal of Sociology 12
(August):438-42. (September):26 3-76.
Heinstein, Martin Parke, Ross D.
1964 Childrearing in California. California State 1970 "The role of punishment in the socialization
Department of Public Health, Bureau of process." Pp. 81-143 in Ronald A. Hoppe,
Maternal and Child Health. Berkeley. G. Alexander Milton and Edward C. Simmel
Johnston, James M. (eds.), Early Experiences and the Processes
1972 "Punishment of human behavior." Ameri- of Socialization. New York: Academic
can Psychologist 27 (November):1033-53. Press.
Kohn, Melvin Sears, Robert R., Eleanor E. Maccoby and Harry
"Social class and parent-child relation- Levin
ships." American Journal of Sociology 68 1957 Patterns of Child Rearing. Evanston, Ill.:
(January). 471-80. Row, Peterson and Company.
Solomon, Richard L.
1969 Class and Conformity, A Study in Values.
1964 "Punishment." American Psychologist 19
Homewood, Ill.:Dorsey Press.
(April): 239-53.
Langner, Thomas, Joseph H. Herson, Edward L.
Greene, Jean D. Jamison, Jeanne A. Goff. Stark, Rodney and James McEvoy III
1969 "Children of the city: affluence, poverty 1970 "Middle class violence." Psychology Today
and mental health." Pp. 185-209 in Vernon 4 (June):52-54.
Allen (ed.), Psychological Factors in Pover- White, Martha 9turm
ty. Chicago: Markham. 1957 "Social class, child rearing practices, and
Lipset, Seymour M. child behavior." American Sociological Re-
1960 "Working class authoritarianism." Chapter 3 view 22 (December):704-12.
in Political Man. New York: Doubleday. Wolfgang, Marvin E.
Littman, Richard A., Robert C. A. Moore and John 1958 Patterns of Criminal Homicide. Phila-
Pierce-Jones delphia: University of Pennsylvania Press.
1957 "Social class differences in childrearing: A Wolfgang, Marvin E. and Franco Ferracuti
third community for comparison with 1967 The Subculture of Violence, Toward an
Chicago and Newton." American Sociologi- Integrated Theory of Criminology. London:
cal Review 22 (December):694-704. Tavistock-Social Science Paperbacks.
Maccoby, Eleanor E. and Patricia K. Gibbs Yarrow, Marian R., John D. Campbell and Roger V.
1954 "Methods of childrearing in two social class- Burton
es." Pp. 380-96 in W. E. Martin and C. B. 1968 Child Rearing: An Inquiry into Research
Standler (eds.), Readings in Child Develop- and Methods. San Francisco: Jossey-Bass.

Вам также может понравиться