Cultural Capital and School Success: The Impact of Status Culture Participation on the

Grades of U.S. High School Students

Author(s): Paul DiMaggio
Source: American Sociological Review , Apr., 1982, Vol. 47, No. 2 (Apr., 1982), pp. 189-
201
Published by: American Sociological Association

Stable URL: https://www.jstor.org/stable/2094962

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.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

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

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 CULTURAL CAPITAL AND SCHOOL SUCCESS:
THE IMPACT OF STATUS CULTURE PARTICIPATION
ON THE GRADES OF U.S.HIGH SCHOOL STUDENTS*

PAUL DIMAGGIO
Yale University

Ethnographers and other students of interaction have documented the impact of status
factors on students' success in school. Yet survey research data consistently show the
absence of family background measure effects on high school grades. It is argued that
conventional measures of family background fail to capture those cultural elements of
status that make a difference in school interactions. Drawing on Weber's work on status
groups and status cultures, and on Bourdieu's work on cultural capital, this paper reports
the findings of an effort to assess the impact of one component of 'status culture
participation-cultural capital-on one aspect of life chances-students' high school
grades. A composite measure of cultural capital has a significant impact on grades,
controlling for family background and measured ability. The pattern of relationships,
however, differs strikingly by gender.

It takes more than measured ability to do measured ability is controlled (Crouse et al.,
well in school. From Warner et al. (1944) and 1979; Sewell and Hauser, 1975; Portes and
Hollingshead (1949) to Coleman (1961) and Wilson, 1976). If measured ability is not the
Cicourel and Kitsuse (1963), ethnographers sole predictor of high school grades and if mea-
have chronicled the impact of class on almost sured differences in family background are not
every aspect of the experience of American either, then to what do we attribute variation in
high school students. More recently, student grades? And how may we square our
ethnomethodologists and constituent ethnog- survey research findings with the observations
raphers have documented the impact of cul- of ethnographers that schools are places in
tural styles on students' relationships with which status and culture matter and "par-
counselors (Erickson, 1975), test scores ticularistic leakages" (Erickson, 1975) abound?
(Mehan, 1974), and classroom instruction The answer may be that aspects of cultural
(McDermott, 1977). Similarly, recent work in style only loosely associated with such mea-
the status attainment tradition finds that mea- sures of family background as father's educa-
sured intelligence explains no more than 15 to tion or head of households occupation make
30 percent of the variation in students' high an important difference.' Max Weber's notion
school grades (Crouse et al., 1979; Sewell and of status culture (1968) may be useful in this
Hauser, 1975). regard. Weber noted that elite status groups-
At the same time, however, measures of collectivities bound together by personal ties
family socioeconomic status have been found and a common sense of honor based upon and
to have a negligible impact on grades when reinforced by shared conventions-generate or
appropriate as their own specific distinctive
cultural traits, tastes, and styles. This shared
* Direct all correspondence to: Paul DiMaggio,
status culture aids group efforts to monopolize
Department of Sociology, Yale University, New
for the group as a whole scarce social, eco-
Haven CT 06520.
nomic, and cultural resources by providing co-
This paper has benefited from the advice and crit-
icism of Ann Swidler, Christopher Jencks, Harrison herence to existing social networks and
White, Susan Bartlett, Ronald Breiger, Randall Col- facilitating the development of comembership,
lins, Steven Brint, Kevin Dougherty, Jerome respect, and affection out of which new net-
Karabel, David Kamens, David Karen, Sally Kil- works are constructed. The content of a status
gore, Katherine McClelland, Susan Olzak, Bernice culture is arbitrary; status honor "may be con-
Pescosolido, Richard A. Peterson, Frank Romo,
David Swartz, Michael Useem, Blair Wheaton, Vera
Zolberg, and two anonymous reviewers. Those listed I After assessing the impact of demographic family
are emphatically not responsible for any remaining background characteristics, Jencks and his col-
inadequacies. Computer and institutional support leagues (1979) found that "unmeasured background
from the Harvard University and Yale University characteristics that vary among families with similar
Sociology Departments and from Yale University's demographic profiles seem to account for significant
Institution for Social and Policy Studies is gratefully amounts of variance in occupational status and
acknowledged. earnings" (p. 81).

American Sociological Review 1982, Vol. 47 (April: 189-201) 189

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 190 AMERICAN SOCIOLOGICAL REVIEW

nected with any quality shared by a plurality" greater.2 Individuals may have a repertoire of
(Weber, 1968: part 2, ch. 9). status cultures that they draw on selectively
The impact of a student's cultural resources (see, e.g., Gumperz and Hymes, 1972, on code
on his or her success in school has been treated switching). In such societies, status culture
explicitly by Bourdieu (1977; Bourdieu and participation may be deployed unconsciously
Passeron, 1977), Collins (1975, 1979), and at the level of daily interaction.
others. According to Bourdieu, schools reward For this reason, it may be more accurate to
students on the basis of their cultural capital, speak of status culture participation than of
defined as "instruments for the appropriation status group membership, and to think of
of symbolic wealth socially designated as status as a cultural process rather than as an
worthy of being sought and possessed" (Bour- attribute of individuals. A person who is "at
dieu, 1977). Teachers, it is argued, communi- home" in a prestigious status culture can dis-
cate more easily with students who participate play tastes, styles, or understandings that
in elite status cultures, give them more atten- serve as cultural resources, making communi-
tion and special assistance, and perceive them cation easier and indicating status group mem-
as more intelligent or gifted than students who bership (see Goffman, 1951; Collins, 1981). In
lack cultural capital. such a fluid world childhood experience and
If, indeed, participation in prestigious status family background may only partially and
cultures represents a kind of cultural capital, modestly determine a person's stock of cultural
we would expect to find the following: capital. Active participation in prestigious
status cultures may be a practical and useful
Hypothesis 1: Measures of cultural capital
strategy for low status students who aspire
are related to one another in a manner that
suggests the existence of a coherent status
towards upward mobility. By contrast, both
culture of which they are elements.
high status students (who, presumably, receive
Hypothesis 2: Cultural capital is positively
cultural resources in the home) and nonmobile
related to school success, in particular, to
low status students may prefer to participate in
high school grades.
adversarial youth subcultures while in high
school (Coleman, 1961).
In much of both the ethnographic and the If this is the case, we would expect the fol-
Weberian tradition, status cultures are seen as lowing:
resources used to promote intergenerational
Hypothesis 3b: Cultural capital's impact on
status persistence; cultural capital is passed
school success is largely net that offamily
down from upper- and upper-middle parents to
background.
their children. If this is the case, then
Hypothesis 4b: Returns to cultural capital
Hypothesis 3a: Cultural capital mediates the are highest for students who are least ad-
relationship between family background and vantaged.
school outcomes
I will refer to this as the cultural mobility
What is more, if, as Bourdieu has argued, cul-
model.
tural capital is inculcated in early childhood and In the sections that follow, I develop a mea-
the response of others to cultural capital is sure of cultural capital and then assess both
predicated in part on the social position of its
models using analyses of the relative impact of
possessor, then measured family background, measured
Hypothesis 4a: Returns to cultural capital ability, and cultural capital on the grades of a
are highest for students from high status national cross-section of white U.S. high
families and least to students from low school students.
status families.
Let us call this, following Bourdieu, the cul- DATA AND MEASURES
tural reproduction model.
The analyses reported below were undertaken
By contrast, consider the possibility that, as with data from a random sample of white re-
Weber predicted, the rise of the market has spondents to PROJECT TALENT. TALENT
severely corroded the status order. While has by far the richest variety of measures of
ideal-typical status groups are well defined and cultural attitudes, information, and activities of
strictly demarcated, in modern societies status approximately twenty data sets reviewed for
cultures are more diffuse and more loosely
bounded. As the potential membership of a 2 Thus Coleman (1971) found that residents of
status group becomes less known to any single Kansas City relied on much more indirect cues to
member, the importance of the shared status assess the status of individuals than did citizens of
culture-those cultural cues that define a per- the small towns studied by earlier community re-
son as a member to other members-becomes searchers.

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 CULTURAL CAPITAL AND SCHOOL SUCCESS 191

this investigation. Its disadvantages include its or composites generated by TALENT are in-
use of self-reported grades and the relatively cluded among the attitude measures. Three in-
low correlations among its demographic mea- ventories combine questions tapping, re-
sures (Porter, 1974), both of which should tend spectively, artistic, musical, and literary inter-
to depress the relationships reported here. The ests.3 A fourth composite, the cultivated
sample includes 1427 men and 1479 women self-image scale, is based on ten self-evaluation
who were in the eleventh grade in public, questions, such as "I enjoy beautiful things," or
parochial, and private high schools in 1960, "I am a cultured person." 2) Activities mea-
when they were surveyed. The sample is sures are based on questions about the extent
weighted to reflect a cross-section of white to which students have created visual arts,
American high school students. performed publically, attended arts events, or
Following Bourdieu, I measure high school read literature. Except for the arts-attendance
students' cultural capital using self-reports of questions, which could include school trips,
involvement in art, music, and literature. these questions explicitly exclude activities
While it would be preferable to ground these undertaken for school course work.4 3) Infor-
measures in observed cultures of dominant mation measures are based on TALENT-
status groups, in the absence of such a rigorous administered tests of information about lit-
data base, high cultural measures represent the erature, music, and art. All these tests tapped
best alternative for several reasons. First, art, familiarity, appreciation, and historical knowl-
classical music, and literature represent the edge, rather than technical skills of the sort
most popular of the prestigious art forms. Pat- developed in practice. In the music information
terns of art museum visitation, concert attend- test, for example, students were asked about
ance, and literature reading in the United famous composers rather than about the
States are similar to those found in France and structure of tonic or dominant chords (Marion
other western countries, with attendance and Shaycroft, personal communication, June 27,
reading concentrated in the upper middle and 1979).
upper classes (DiMaggio and Useem, 1978). Fuller descriptions, including means and
Second, to the extent that there is a common standard deviations, of the measures used in
cultural currency among American elites, it in- these analyses are provided in the Appendix.
volves at least a modest familiarity with the Boys' reported cultural interests and activities
arts and literature. Such preoccupations as were low; girls' were moderately higher. In
racquetball, wine, or ancient history are likely short, high culture was a minority taste in
to characterize smaller, more localized status American high schools in the 1960s.
groups. Minimal familiarity with high culture,
by contrast, transcends cleavages of age or
ANALYSIS
region. Third, art and music have received rel-
atively superficial attention in the curricula of Both the cultural reproduction model and the
American high schools (Rindskopf, 1979). If, cultural mobility model yield the prediction
as Bourdieu contends, cultural capital consists that separate measures of high cultural in-
of familiarity with precisely those subjects that volvements should be positively correlated
schools do not teach but that elites value, then with one another. This prediction inheres in
including art and music permits us to tap di- the definition of cultural capital as the mastery
mensions of cultural capital that are inculcated of elements of a prestigious status culture.
outside of the school. Finally, high culture is an There is no a priori reason that students who
element of elite culture that school teachers care about any one art form-art, music, or
appear to regard as legitimate. While American literature-should be concerned about any
teachers are recruited largely from the lower
middle class (Lortie, 1975), they are overrep-
resented in arts audiences (in proportion to their
I These questions cover a broader range (and are
share of the labor force) more strongly than less strictly high cultural) than the specific interest
any other group (DiMaggio and Useem, 1978). questions included in Table 1. The Artistic Interest
The first step in the analysis was to build a Inventory, for example, includes responses to ques-
scale of measures of cultural capital. Three tions about interest in becoming a decorator or de-
signer.
kinds of measures from PROJECT TALENT
4 It is possible that students from upper middle-
were employed. 1) Attitude measures asked
class families were tempted to exaggerate these re-
students to rate their interest in specified artis-
ports, although the low correlations between the
tic activities and occupations on a scale from self-reports and parental education make this seem
one to five. Unlike aspirations questions, the somewhat unlikely. False reports, if skewed in this
occupational-interest questions simply asked way, would attenuate the relationships between
the student to rate the attractiveness of a wide cultural capital measures and grades once back-
range of careers. In addition, four inventories ground was controlled.

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 192 AMERICAN SOCIOLOGICAL REVIEW

Table 1. Median Correlations and Ranges (in parentheses) of Correlations Between Sets of Cultural
Measures

Interests Activities Information

Median Range N* Median Range N Median Range N

Interests
W .71 (.47-.80) 9
MB .45 (.32-.60) 27
W .73 (.48-.82) 9
F B .47 (.35-.63) 27
A activities

M W .31 (.22-.40) 9 NA
B .20 (.11-.38) 27 .20 (.12-.29) 6
F W .33 (.21-.44) 9 NA
B .18 (.06-.34) 27 .17 (.13-.22) 6
Information
MW .14 (.06-.24) 9 .11 (.04-.14) 3 NA
B .09 (-.01-.20) 18 .08 (-.03-.26) 9 .67 (.61-.68) 3
W .29 (.18-.37) 9 .17 (.15-.28) 3 NA
FB .26 (.09-.37) 18 .18 (.01-.31) 9 .66 (.60-.68) 3
* N refers to the number of correlations in each category.
W = Correlations between variables in same discipline (art, music, or literature).
B = Correlations between variables in different disciplines.
NA = Not applicable (only one measure per discipline in this category).

other. Indeed, psychological research indicates areas.5 Finally, students who engage in one
that the practice of different art forms draws onkind of cultural activity are more likely than
substantially different cognitive skills (Wolf others to be interested in any other high cul-
and Gardiner, 1979). If we do find that mea- tural activity.
sures of involvement in different artistic disci- It may be objected, however, that the posi-
plines are related, we must look beyond psy- tive correlations simply indicate that all of
chological explanations for the answer. these measures tap some underlying personal-
The notion of status culture leads to just ity attribute like creativity. Fortunately, TAL-
such an explanation. To the extent that art, ENT also reports activity measures for several
music, and literature are part of a coherent middlebrow cultural passtimes-photography,
status culture, we would expect students in- crafts, woodworking, and needlework. If high
terested in music to be interested in literature cultural involvements really constitute part of a
and art, and vice versa. Milieus that inculcate coherent status culture, we would expect to
an interest in any single artistic discipline will
find these measures less strongly correlated
also be likely to inculcate an interest in any with the high culture measures than the latter
other high culture form. This expectation is are with one another. Again, this is the case.
particularly strong for the attitude and infor- Two-thirds and one-half of all correlations be-
mation measures. Participation takes time, so tween cultural attitudes and cultural activities,
students who value the arts may tend to spe- for boys and girls respectively, are greater than
cialize in practicing one form, while maintain- or equal to .20. None of the correlations be-
ing interest in and knowledge about others. Of tween cultural attributes and middlebrow
particular interest are correlations between activities reaches this level. Similar findings
measures of involvement in different forms. If emerge when we compare correlations be-
high cultural involvements constitute elements tween pairs of cultural activities with correla-
in a coherent status culture, these between- tions between cultural and middlebrow activi-
discipline correlations should be consistently ties. (Tables available upon request.)
and significantly positive. These findings are consistent, then, with the
The findings are summarized in Table 1. As first proposition of each model, that different
expected, relationships among high culture at-
titude measures are strongly positive. The me-
dian between-discipine correlations are .45 and 5 Ability is measured by the student's composite
vocabulary score. The commonly employed measure
.47 for boys and girls, respectively. Relation-
of ability-the composite of thirty-seven information
ships among cultural activities are less strong,
tests administered by TALENT-is contaminated by
but all significant at p S .001. Cultural infor-
the inclusion therein of the cultural information test
mation test scores in different cultural disci- scores. Thus the vocabulary composite, which cor-
plines are also strongly associated, even when relates at .94 with the information composite, is used
one controls for ability test scores in other as a proxy.

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 CULTURAL CAPITAL AND SCHOOL SUCCESS 193

Table 2. Factor Loadings for Factors Derived from Factor Analysis of Data on Cultural Interests, Activities,
and Test Scores, for Male and Female Eleventh Graders, Project TALENT*

Factor Eigen- Percentage

Factor Variable Loadings value of Variance
MALES
1) Cultural Interests Musical composer .529 4.217 23.4
Poet .649
Artist .857
Visiting art galleries .656
Reading literature .454

2) Cultural Information English literature .832 2.624 14.6

Music .725
Art .804

3) Cultural Capital Symphony concerts (int.) .609 1.460 8.1

Performing (act.) .442
Arts attendance (act.) .560
Literature reading (act.) .350
Cultivated self-image .361

4) Middlebrow Activity Drawing .406 1.229 6.8

Photography .389
Crafts .540
Woodworking .482

FEMALES
I) Cultural Interests Musical composer .569 4.566 25.4
Poet .691
Artist .768
Visiting art galleries .629
Reading literature .506

2) Cultural Information English literature .831 1.947 10.8

Music .730
Art .785

3) Cultural Capital Symphony concerts (int.) .499 1.607 8.9

Performing (act.) .418
Arts attendance .447
Cultivated self-image .377
4) Middlebrow Activity Drawing .544 1.259 7.0
Crafts .571
Sewing .365

* Principal component factor anal

of a variable in a factor.

dimensions of involvement with different high with eigenvalues over 1.0. Factor 1, cultural
culture disciplines are part of a relatively co- interests, consists of all the attitude measures
herent status culture. Note, however, the weak except interest in attending symphony concerts
relationships between attitudes or activities in and cultivated self-image, which loaded onto
any single discipline and scores on tests of factor 3. Factor 2, cultural information, con-
information about any other. This finding sug- sists of the three cultural information test
gests the importance of distinguishing among scores. Factor 3 is the factor of greatest inter-
the three dimensions of cultural involvement in est because it combines both attitude and ac-
assessing and explaining their effects. It also tivity measures that are particularly high cul-
suggests that, at least among teenagers in 1960, tural in nature. For this reason, it is interpreted
artistic attitudes and activities were more im- as representing cultural capital in its purest
portant elements of status culture participation form.6 Factor 4, middlebrow activities, con-
than was cultural information.
In order to exploit further the recognition
6 While the designation of Factor 3 as "cultural
that different cultural dimensions may have
capital" seemed intuitively obvious, there is no doubt
different relationships to one another and to
an element of subjectivity in any such choice. Some
school success, and to simplify the subsequent of the conscious elements entering into the decision
analyses, the cultural measures were factor an- include the following: the cultivated self-image scale
alyzed. Separate analyses for male and female includes some self-assessment items that are very
respondents each yielded four similar factors close to the idea of participation in a prestigious

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 194 AMERICAN SOCIOLOGICAL REVIEW

sists of nonhigh culture creative pursuits, ex- this correlation suggests that students' cultural
cluding, for each gender, those in which the information test scores were largely deter-
fewest students reported participation. It also mined by some underlying set of aptitudes,
includes drawing, which, for both genders, skills, and motivations that lead students to do
clustered with the crafts rather than with the well or poorly on tests. (Partial correlations,
arts activities. not reported here, indicated that the relation-
Each of these four factors represents a kind ship between grades and cultural information
of cultural resource, and each represents a co- test scores largely evaporated when measured
herent set of interrelated traits. Factor 4 should ability was controlled.)
have little, if any, positive impact on students' Dependent variables in the analysis were
grades, unless, perhaps, it represents a mea- students' self-reported grades in English, in
sure of creativity. Factor 1 should have less of History and Social Studies, and in Mathemat-
an impact on grades than factors 2 or 3, be- ics, and a TALENT composite of self-reported
cause it measures attitudes rather than actual grades in all subjects. The use of self-reported
behavior or information. If status culture par- grades, with restricted distributions, can be
ticipation influences grades because students expected to depress R2s in these analyses; but it
display their knowledge in a manner that im- does not affect the utility of the data for com-
presses teachers or boosts their performance parisons of the relative effects of independent
on tests, we would expect factor 2 to have a variables (Picou and Carter, 1976). English,
major impact. If we believe that cultural capital History, and Social Studies are subjects in
consists of a set of interests, dispositions, be- which cultural capital can be expected to make
haviors, and styles that are learned and a difference; standards are diffuse and evalua-
enacted socially, then we would predict that tion is likely to be relatively subjective. By
factor 3, cultural capital, would have the contrast, Mathematics requires the acquisition
greatest impact. This is the case not just be- of specific skills in the classroom setting, and
cause factor 3 includes measures of high cul- students are evaluated primarily on the basis of
tural activities, but because the factor is the their success in generating correct answers to
only one that crosscuts question types as well. sets of problems. Thus Welch et al. (1980) re-
While all of these factors are, of course, only port that Mathematics achievement test scores
indirect measures of cultural resources that are much more strongly influenced by years of
students bring to interactions with significant school subject matter instruction than are
others, it is predicted that factor 3, cultural achievement test scores in English and Civics.
capital, will have the greatest impact on The regression results are displayed in Table
grades. 3. They provide striking confirmation of the
To test the hypothesis that cultural capital hypothesis that cultural capital is positively
significantly influences grades, separate re- related to high school grades. Standardized re-
gressions were executed for male and female gression coefficients for cultural capital (factor
eleventh graders. Independent variables in- 3) are significant at p - .001 for both males and
clude cultural factors 1, 3, and 4, the student's females for grades in all subjects but Mathe-
report of his or her father's educational attain- matics, where effects are smaller, but still sig-
ment, and the student's composite score in the nificant. For English, History/Social Studies,
TALENT vocabulary tests.7 Factor 2, Cultural and All Grades, the impact of cultural capital is
Information, was excluded from the analysis of the same order of magnitude as the effect of
because of its high collinearity (over .800) with measured ability. Cultural interests (factor 1)
the composite ability measure. The strength of and middlebrow activity (factor 4) have no sig-
nificant impact on grades. As expected, the
impact of father's education is minimal.
status culture; arts attendance is the single measure These results support the expectation of
most commonly used as a proxy for cultural capital both the cultural reproduction and the cultural
in other work; literature reading is also a proxy that mobility models that participation in presti-
has been used in the past; symphony audiences are gious status cultures has a significantly positive
the most elite of all arts audiences; performance is impact on grades. (Factors such as self-
probably related to formal training; and each of these
reported grades that depress the R2s should not
variables is relatively strongly related to ability and
affect the relative weights of ability and cul-
to family status.
tural capital. The latter, rather than the total
7 Father's education was included as the sole
background measure because analysis of similar
variance explained, is the focus of this
TALENT samples (e.g. Crouse et al., 1979) showed analysis.) Indeed, the magnitude of the effects
little benefit from inclusion of multiple background relative to those of ability was unexpectedly
measures as predictors of grades, and because the great. The findings also tend to disconfirm two
cultural measures were more strongly related to fa- possible alternative explanations of the associ-
ther's than to mother's education for both genders. ation between grades and cultural measures. If

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 CULTURAL CAPITAL AND SCHOOL SUCCESS 195

Table 3. Results of Regression of Grades on Ability (1), Father's Education (2), Cultural Capital (3),
Middlebrow Activity (4), and Cultural Interests (5) for Male and Female Eleventh Graders

Reduction in
beta of
Increase in father's
Dependent R2 with education with
Variable 1 2 3 4 5 R2 Vars. 3-5 Vars. 3-5

MALES
Grades in B .5078 .1158 2.2330 .1488 -.1256 .1228 .0286 .0202
All Subjects s.e. .0623 .0971 .4330 .4830 .3830
beta .2791*** .0409 .1706*** .0106 -.0113
Grades in B .0493 .0223 .2730 -.0025 .0412 .1034 .0310 .0211
English s.e. .0077 .0120 .0540 .0603 .0476
beta .2255*** .0660 .1716*** -.0015 .0309
Grades in B .0715 .0012 .2646 -.0059 .0259 .1279 .0254 .0193
History s.e. .0080 .0125 .0561 .0627 .0495
beta .3096*** .0033 .1577*** -.0033 .0184
Grades in B .0510 .0189 .1685 .0133 -.0387 .0723 .0102 .0124
Mathematics s.e. .0082 .0128 .0575 .0625 .0507
beta .2223*** .0531 .1011** .0075 -.0277
N = 809
FEMALES
Grades in B .5988 .0939 2.4314 .4223 -.1250 .1897 .0338 .0297
All Subjects s.e. .0575 .0857 .3887 .3925 .3361
beta .3374*** .0345 .1901*** .0346 -.0124
Grades in B .0602 .0150 .3412 -.0062 .0127 .1683 .0463 .0300
English s.e. .0072 .0106 .0482 .0485 .0412
beta .2797*** .0457 .2211*** -.0042 .0106
Grades in B .0710 .0111 .3354 .0352 .0443 .1713 .0382 .0294
History s.e. .0079 .0116 .0531 .0534 .0453
beta .2991*** .0305 .1970*** .0218 .0333
Grades in B .0653 -.0066 .1302 .0750 -.0671 .0857 .0079 .0091
Mathematics s.e. .0082 .0121 .0551 .0555 .0471
beta .2781*** -.0185 .0774* .0469 -.0511
N = 917

* p S .05, two-tailed.
** p S .01, two-tailed.
*** p S .001, two-tailed.

these measures tapped some general dimension status homes than from low status back-
of academic achievement motivation, we grounds. By contrast, the cultural mobility
would expect the impact on grades in Mathe- model posits that the impact of cultural capital
matics to equal those on other subjects. In fact, will be greater on the grades of less advantaged
it does not. If the scores reflected some un- youth, for whom the acquisition and display of
derlying dimension of creativity, factor 4 prestigious cultural resources may be a vital
would have a significant impact on grades; part of upward mobility.
again, it does not. The male and female samples were each di-
The findings provide limited support for the vided into three groups on the basis of father's
expectations of either model about the extent education: sons and daughters (respectively) of
to which cultural capital mediates the relation- college graduates, sons and daughters of high
ship between family background and school school graduates who did not graduate from
success. While the inclusion of the cultural college, and sons and daughters of men who
capital measures does reduce the betas for fa- did not hold high school diplomas. Separate
ther's education by 20 to 80 percent, the origi- regressions were run on each of these six sub-
nal betas are so low that these figures are samples.
somewhat trivial. The extent to which these Tables 4 and 5 indicate divergent results for
measures affect grades independent of the im- men and women. Among women, the impact of
pact of father's education squares with the pre- cultural capital on all four grade measures rises
dictions of the cultural mobility model. monotonically with father's education. As the
The third proposition of the cultural repro- cultural reproduction model predicts, returns
duction model holds that returns to cultural to cultural capital are greatest to women from
capital will be greater for students from high high status families and least to women from

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 196 AMERICAN SOCIOLOGICAL REVIEW

Table 4. Results of Regressions of Grades on Ability (1), Cultural Attitudes (2), Cultural Capital (3), and
Middlebrow Activity (4) for Male Eleventh Graders with Non-High School Graduate, High School
Graduate, and College Graduate Fathers

Increase in
R2 with
Dependent Variable 1 2 3 4 R2 Vars. 2-4

Males with Non-High School Graduate Fathers N=494

Grades in All Subjects B .4384 -.6358 2.0638 .7770 .0865 .0297
s.e. .0724 .4530 .5302 .5828
beta .2515*** -.0614 .1616*** .0583
Grades in English B .0559 -.0743 .3026 .1267 .0998 .0404
s.e. .0093 .0586 .0693 .0764
beta .2604*** -.0577 .1885*** .0752
Grades in History B .0632 -.0195 .2549 .0061 .0957 .0223
s.e. .0098 .0618 .0732 .0806
beta .2796*** -.0144 .1508*** .0034
Grades in Mathematics B .0345 -.0560 .0490 .0492 .0257 .0025
s.e. .0098 .0622 .0736 .0811
beta .1572*** -.0426 .0299 .0286

Males with High School Graduate Fathers N=298

Grades in All Subjects B .5397 .6665 2.0630 -.3403 .1216 .0289
s.e. .0986 .6205 .7008 .7307
beta .2903*** .0596 .1568** -.0258
Grades in English B .0504 .0172 .2396 -.0888 .0951 .0387
s.e. .0129 .0804 .0910 .0954
beta .2184*** .1264* .1482** -.0549
Grades in History B .0857 -.0374 .3433 -.0570 .1760 .0422
s.e. .0128 .0799 .0904 .0948
beta .3568*** - .0264 .2039*** - .0338
Grades in Mathematics B .0608 .0230 .2385 - .0731 .0809 .0189
s.e. .0142 .0881 .0997 .1046
beta .2424*** .0156 .1357* -.0415

Males with College Graduate Fathers N= 130

Grades in All Subjects B .4274 .2767 1.5914 .1906 .0702 .0178
s.e. .1735 1.0292 1.0230 1.2428
beta .2088* .0237 .1315 .0135
Grades in English B .0242 .1253 .1774 -.0555 .0337 .0175
s.e. .0214 .0493 .1243 .1553
beta .1010 .0371 .1276 -.0337
Grades in History B .0651 .2500 .0723 -.0339 .0959 .0285
s.e. .0233 .1361 .1350 .1687
beta .2421** .1676 .0463 -.0183
Grades in Mathematics B .0625 -.1276 .0963 .1436 .0670 .0123
s.e. .0233 .1362 .1351 .1688
beta .2362** - .0867 .0625 .0787

* p<.05, two-tailed.
** p<.Ol, two-tailed.
*** p<.001, two-tailed.

low status families. Among the former group, GENDER DIFFERENCES

the impact of cultural capital exceeds that of
ability on grades in History and approaches it The male and female samples differed mark-
even for grades in Mathematics. edly in the relationships between family back-
By contrast, among males the positive im- ground and returns to cultural capital. As the
pact of cultural capital on grades is restricted to cultural reproduction model would predict,
students from lower and middle status house- cultural capital had its largest impact on the
holds. Sons of college graduates were no more daughters of women whose fathers were col-
likely to receive good grades if they scored lege graduates. Effects on grades of daughters
high on factor 3 than if they did not. These of high school graduates without college de-
results for males are consistent with the ex- grees were smaller, and effects on grades of
pectations of the cultural mobility model. daughters of men without high school diplomas

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 CULTURAL CAPITAL AND SCHOOL SUCCESS 197

Table 5. Results of Regressions of Grades on Ability (1), Cultural Attitudes (2), Cultural Capital (3), and
Middlebrow Activity (4) for Female Eleventh Graders with Non-High School Graduate, High
School Graduate, and College Graduate Fathers

Increase in
R2 with
Dependent Variable 1 2 3 4 R2 Vars. 2-4
Females with Non-High School Graduate Fathers N=582
Grades in All Subjects B .5006 .3136 2.2030 .7400 .1453 .0310
s.e. .0660 .3966 .5010 .4966
beta .2930*** .0331 .1636*** .0614
Grades in English B .0440 .0906 .3312 -.0195 .1148 .0452
s.e. .0086 .0504 .0651 .0633
beta .2093*** .0788 .2036*** - .0133
Grades in History B .0654 .0958 .2638 .0213 .1358 .0272
s.e. .0093 .0546 .0705 .0685
beta .2840*** .0761 .1479*** .0145
Grades in Mathematics B .0450 -.0693 -.0005 .1535 .0442 .0075
s.e. .0098 .0577 .0745 .0724
beta .1943*** -.0548 -.0003 .0951*

Females with High School Graduate Fathers N=342

Grades in All Subjects B .6216 -.8526 2.7058 .3466 .1776 .0494
s.e. .0980 .5542 .6034 .6064
beta .3266*** - .0844 .2238*** .0301
Grades in English B .0629 -.0831 .3641 -.0114 .1755 .0659
s.e. .0115 .0657 .0715 .0722
beta .2906*** - .0720 .2608*** .0086
Grades in History B .0634 -.0008 .4029 .0278 .1640 .0611
s.e. .0132 .0749 .0816 .0824
beta .2585*** - .0006 .2546*** .0185
Grades in Mathematics B .0870 -.1431 .2245 .0647 .1384 .0230
s.e. .0141 .0804 .0875 .0884
beta .3356*** -.1034 .1343* .0406

Females with College Graduate Fathers N=113

Grades in All Subjects B .7317 -.6325 4.1952 -.2707 .2034 .0910
s.e. .1738 1.0617 1.1567 1.2902
beta .3714*** -.0530 .2968*** -.0174
Grades in English B .0913 -.0148 .4244 -.0566 .2093 .0729
s.e. .0214 .1264 .1386 .1561
beta .3894*** -.0108 .2649** -.0317
Grades in History B .0673 -.0764 .6174 .0944 .1644 .1160
s.e. .0247 .1459 .1600 .1801
beta .2557** -.0496 .3431*** .0470
Grades in Mathematics B .0702 -.0924 .3971 .0661 .1216 .0578
s.e. .0235 .1387 .1521 .1713
beta .2877** - .0647 .2380* .0355

* p<.05, two-tailed.
** p<.O1, two-tailed.
p<.001, two-tailed.

were smaller still. By contrast the impact of tion in high culture activities than did the men.
cultural capital on grades was substantial, rel- Second, the individual cultural measures were
ative to that of ability, for sons of men in the more strongly related to ability scores for
two less educated groups, but negligible for males. Third, the specific attitude, activity,
sons of college graduates. This finding is con- and information measures were, in every case,
sistent with the cultural mobility model. more strongly correlated with family back-
The divergent findings for male and female ground (both father's and mother's education)
samples were part of an overall pattern of gen- for girls than for boys (table available upon
der differences that together suggest that cul- request). And the intercorrelations among the
tural capital plays a different role in the mobil- cultural measures were stronger for high status
ity strategies of men and women. First, the girls than for lower status girls, suggesting that
women in the sample expressed substantially a more coherent status culture participation
more interest and reported greater participa- pattern existed within the high status group.

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 198 AMERICAN SOCIOLOGICAL REVIEW

No such differences appeared for boys. (Three granted. By the same token, teachers may have
way cross-tabulations, controlling for father's rewarded students from lower status back-
education, were executed for each pair of the grounds who exhibited interests and behavior
cultural interest and cultural activity questions. expected from higher status students.
The bivariate relationships were strongest Conversely, cultural capital seems to have
among college educated men's daughters, but been part of an identity shared by academically
not among their sons.) successful high school girls. In a society in
These findings suggest that cultural interests which men monopolize careers and control the
and activities were culturally prescribed for material rewards they carry, it is all the more
teenage girls, while for adolescent boys they important for women to distinguish themselves
were less strongly prescribed, perhaps even through fundamentally cultural markets. Elder
negatively sanctioned by peers. High cultural (1974: ch. 8) has observed that women lead
involvements may have been part of an identity "contingent careers"; to a greater extent than
kit that academically successful, high status men they are evaluated and rewarded on the
girls, but not similar boys, possessed. basis of ascriptive and diffuse criteria. Women
This interpretation is consistent, as well, who wish to be recognized as eligible partners
with an additional gender difference revealed for men from high status backgrounds may
in the correlation matrix. Reading literature need cultural capital to a greater extent than
was more strongly related to cultural attitudes men who wish to achieve in the world of work.
for girls than other activities, while arts attend- For boys from high status families, it may be
ance and performing were more strongly re- more important, in high school, to develop a
lated to attitudes for boys. If having cultural taste for women who appreciate culture than to
interests were part of the good student role for develop a taste for high culture itself.
girls, then we would expect literature reading
to be most strongly related to interest in other
CONCLUSIONS
cultural fields; for in literature reading the cul-
ture inculcated by the upper middle-class home The relatively low correlations between pa-
and that inculcated by the school reinforce rental education and cultural capital are nota-
each other. As may have been the case for sons ble. An analysis of the responses of a cross-
of college graduates, to the extent that cultur- sectional sample of American adults to ques-
ally oriented students were not necessarily tions that included a broader range of cultural
successful or conformist students, cultural attendance activities, but required a greater
interests would most likely be expressed pre- specificity as to the extent of the activity,
cisely through those activities that the school found correlations of both occupational status
does not teach-for example, performing or and educational attainment with culture con-
arts attendance. sumption of approximately .40 (Gruenberg,
Coleman found in The Adolescent Society 1975:200). To the extent that the TALENT
(1961:118-27) that leading-crowd girls were data on arts attendance are comparable to the
more likely to exhibit conformist charac- ISR data employed by Gruenberg, they indi-
teristics and less likely to be as interested in cate that well-educated parents passed down
popular music as their peers than leading- 30 percent of their cultural advantage to their
crowd boys. The high school years- sons and 60 percent to their daughters.8 If, as
particularly in an era in which high school en- Bourdieu argues, early socialization is critical
rollments were rising rapidly-may have been to an adult's inclination and ability to consume
a kind of cultural latency period during which high culture, the inheritance may lessen with
high status boys rebelled against parental age. It is also possible, however, that opportu-
values. Boys with educated fathers who were nities afforded by the school and peer group
culturally oriented may have been less in- only temporarily attenuate the relationship
volved in their peer groups and less likely to between family background and cultural in-
possess other traits that lead to school success volvement during high school; or that the ado-
than were other boys. Those high status boys lescent rebellion described by Coleman (1961)
who were not academically oriented may have and Stinchcombe (1964) during the early 1960s
used high culture as an alternative arena for temporarily depressed the correlation among
achievement. By contrast, boys from humble the boys in the TALENT sample. While firmer
backgrounds who were upwardly mobile may conclusions await analysis of the impact of an
have begun to enter the status culture of the array of background measures on students'
upper middle class during high school. Up-
wardly mobile boys may have been more in- 8 These figures are the ratios of the correlations
clined to express cultural interests and to par- between arts attendance and father's education to the
ticipate in cultural activities than were upper .40 correlation Gruenberg reports, for boys and girls
middle-class boys, who took such interests for respectively.

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 CULTURAL CAPITAL AND SCHOOL SUCCESS 199

cultural capital, these findings lend tentative suited for them. The use of self-reported
support to Jencks and Riesman's assertion grades, the elimination of high school dropouts
(1968) that the level of cultural mobility in the and minority children from the data, and the
United States has been relatively high. The self-reports of cultural participation all can be
findings also suggest that cultural capital is less expected to reduce the impact of cultural cap-
strongly tied to parental background traits than ital on grades. What is more, the use of na-
Bourdieu's theory or similar discussions of tional data limits this assessment to the impact
class and culture in the United States would of participation in a national elite status cul-
predict. Whether more direct measures than ture, and does not permit estimation of the
education or parents' cultural capital would re- effects of local status culture variations.
veal a stronger inheritance remains to be seen. Even with these limitations, the data show
It follows from these findings that educa- that cultural capital has an impact on high
tional attainment is a very imperfect proxy for school grades that is highly significant and that,
cultural capital. A second, and related, lesson in nontechnical subjects, approaches the con-
is that single measures of "cultural capital" or tribution of measured ability. This finding con-
participation in status cultures are inadequate. firms rather dramatically the utility of the per-
Abandoning the use of such variables as edu- spective advanced here. It remains, however,
cational attainment or self-reported arts at- to assess the impact of cultural capital on such
tendance as single proxies for status culture outcomes as educational attainment, college
participation raises formidable methodological quality, marital selection, and occupational
puzzles, but these must be confronted. attainment; to develop better measures of cul-
An ideal data set for our purposes would tural capital; to assess the differing role cul-
contain measures of cultural capital grounded tural capital may play in the mobility strategies
in research on adult elites in a single commu- of different class segments; and to compare the
nity; objective measures of grades, stan- influence of cultural capital in different kinds of
dardized by school; data on teachers' evalua- educational and occupational settings. In all
tions of students' characters and aptitudes; and these arenas, conceiving of status as a cultural
observationally grounded measures of stu- process which influences success by affecting
dents' interaction style, both linguistic and the outcomes of interactions may yield impor-
nonverbal. The TALENT data were not col- tant gains in our ability to understand the status
lected for our purposes and are not ideally attainment process as a whole.

APPENDIX

Table A-1. Means and Standard Deviations of Talent Measures

Question Mean Standard Deviation N

INTEREST (5 = high; 1 = low)

In being a musical composer 2.29 (boys) 1.33 1355
2.84 (girls) 1.46 1401
In being a poet 1.97 1.19 1354
2.49 1.47 1380
In being an artist 2.48 1.43 1364
3.11 1.53 1406
In visiting art galleries 2.10 1.29 1310
2.73 1.48 1385
In reading literature 2.51 1.30 1258
3.14 1.47 1353
In attending symphony concerts 2.33 1.38 1251
2.86 1.52 1356
Artistic Interest Inventory (TALENT scale) 16.64 9.16 1384
21.43 9.87 1426
Musical Interest Inventory (TALENT scale) 13.87 10.85 1384
18.80 11.41 1425
Literary Linguistic Interest Inventory (TALENT scale) 15.58 8.29 1385
21.20 8.91 1426
Cultivated Self-Image (TALENT scale) 4.64 2.30 1400
5.95 2.21 1464
ACTIVITIES
Drawing, etc., in the past three years 2.24 (boys) 1.27 1398
(5 = very often; 1 = never) 2.69 (girls) 1.34 1416
Acting, etc., in the past three years 1.97 1.23 1403
(5 = very often; 1 = never) 2.62 1.40 1411

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 200 AMERICAN SOCIOLOGICAL REVIEW

Appendix (Continued)
Table A-l. Means and Standard Deviations of Talent Measures

Question Mean Standard Deviation N

Attending concerts, etc., in the past three years 3.85 1.46 1372
(1 = very often; 6 = never) 3.37 1.42 1420
Books, plays, poetry, etc. read the past three months 1.15 1.75 1340
(0= none; 5 = 5 or more) 1.80 1.99 1396
INFORMATION
Literature information test (low = 0; high = 24) 13.28 4.60 1402
13.12 4.47 1463
Music information test (low = 0; high = 12) 6.29 3.04 1401
6.88 3.00 1461
Art information test (low = 0; high = 12) 6.38 2.51 1386
6.62 2.58 1423

MIDDLEBROW ACTIVITY (3-year frequency,

5 = very often; 1 = never)
Photography (excluding occasional snapshots) 1.74 (boys) 1.15 1446
1.41 (girls) 0.87 1474
Making jewelry, pottery, etc. 1.62 0.95 1442
1.72 0.96 1472
Cabinet making or woodworking 2.36 1.24 1435
1.15 0.46 1469
Sewing, etc. 1.27 0.73 1432
3.65 1.19 1469
Mark the one answer indicating the highest level of education
your father reached: 11.27 (boys) 3.53 1166
4 = none, or some grade school 11.14 (girls) 3.60 1237
8 = completed grade school
10 = some high school, but did not graduate
12 = graduated from high school
13 = vocational or business school after high school
14 = some junior or regular college, but did not graduate
16 = graduated from regular 4-year college
17 = master's degree
18 = some work toward doctorate or professional degree
20 = completed doctorate or professional degree

Mark the one answer indicating the highest level of education

your mother reached: 11.52 (boys) 2.75 1204
11.28 (girls) 2.99 1277
The following questions ask you to report your grades in courses
you have taken in the ninth grade or later. Please consider only
semester grades. If you have not taken any courses in the
topic, skip the item. In these questions, choose the one answer
that best describes your grades.
6 = all A's or equivalent 3 = mostly B's and Cs or equivalent
5 = mostly A's or equivalent 2 = mostly Cs and D's or equivalent
4 = mostly A's and B's or equivalent 1 = mostly D's and below or equivalent
My grades in history and social studies have been 3.52 1.27 1296
3.61 1.25 1361
My grades in English courses have been 3.36 1.21 1299
3.59 1.02 1351
TALENT Composite of self-reported grades 23.34 9.71 1370
26.19 9.52 1424

REFERENCES Cicourel, Aaron and John Kitsuse

1963 The Education Decision-Makers. In-
Bourdieu, Pierre dianapolis: Bobbs-Merrill.
1977 "Cultural reproduction and social repro- Coleman, James C.
duction." Pp. 487-511 in Jerome Karabel 1961 The Adolescent Society. New York: Free
and A. H. Halsey (eds.), Power and Ideol- Press.
ogy in Education. New York: Oxford. Coleman, Richard
1977 Reproduction in Education, Society, Cul- 1971 Social Status in the City. San Francisco:
ture. Beverly Hills, Calif.: Sage. Jossey-Bass.

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms
 CULTURAL CAPITAL AND SCHOOL SUCCESS 201

Collins, Randall McDermott, R. P.

1975 Conflict Sociology. New York: Academic. 1977 "Social relations as contexts for learning in
1979 The Credential Society. New York: Aca- school." Harvard Educational Review
demic. 47:198-213.
1981 "The micro-foundations of macro- Mehan, Hugh
sociology." American Journal of Sociology 1974 "Accomplishing classroom lessons." Pp.
86:984-1014. 76-142 in Aaron Cicourel et al., Language
Crouse, James, Peter Mueser and Christopher Use and School Performance. New York:
Jencks Academic.
1979 "Latent variable models of status attain- Picou, J. Steven and Michael Carter
ment." Social Science Research 8:348-68.
1976 "Sipnificant-other influences and aspira-
DiMaggio, Paul and Michael Useem tions." Sociology of Education 49:12-22.
1978 "Cultural democracy in a period of cultural Porter, James N.
expansion: the social composition of arts 1974 "Race, socialization, and mobility in edu-
audiences in the United States." Social cational and early occupational attain-
Problems 26:180-97. ment." American Sociological Review
Elder, Glen 39:303-16.
1974 Children of the Great Depression. Chicago: Portes, Alejandro and Kenneth L. Wilson
University of Chicago Press. 1976 "Black-White differences in educational
Erickson, Fred attainment." American Sociological Review
1975 "Gatekeeping and the melting pot." Har- 41:414-31.
vard Educational Review 45:44-70. Rindskopf, David
Flanagan, John C. et al. 1979 Arts Education in Public Secondary
1964 The American High School Student. Schools: Offerings, Enrollments, and their
Pittsburgh: University of Pittsburgh, Determinants. Technical Memorandum
PROJECT TALENT. #AH-46. St. Louis: CEMREL, Inc.
Sewell, William and Robert M. Hauser
Goffman, Erving
1975 Education, Occupation, and Earnings:
1951 "Symbols of class status." British Journal of
Achievement in the Early Career. New
Sociology 2:298-312.
York: Academic.
Gruenberg, Barry
Stinchcombe, Arthur
1975 How Free is Free Time? Analysis of Some
1964 Rebellion in a High School. Chicago: Quad-
Determinants of Leisure Activity Patterns.
rangle.
Manuscript. Middletown, CT.
Warner, W. Lloyd, Robert J. Havighurst and Martin
Gumperz, John J. and Dell Hymes B. Loeb
1972 Directions in Sociolinguistics. New York: 1944 Who Shall Be Educated? New York:
Holt, Rinehart, and Winston. Harper and Brothers.
Hollingshead, August B. Weber, Max
1949 Elmtown's Youth. New York: Wiley. 1968 Economy and Society. New York: Bed-
minster Press.
Jencks, Christopher et al.
Welch, Wayne W., Ronald E. Anderson and Linda J.
1979 Who Gets Ahead? The Determinants of
Harris
Economic Success in America. New York:
1980 "The effects of schooling on math achieve-
Basic.
ment." Mimeo, University of Minnesota.
Jencks, Christopher and David Riesman Wolf, Dennis and Howard Gardiner
1968 The Academic Revolution. New York: 1979 "Style and sequence." Pp. 117-138 in
Doubleday. Margery Franklin and Nancy R. Smith
Lortie, Dan (eds.), Symbolic Functioning in Childhood.
1975 Schoolteacher. Chicago: University of Englewood Cliffs, NJ: Lawrence Erlbaum
Chicago Press Associates.

This content downloaded from

129.234.7.129 on Sun, 02 Oct 2022 15:46:21 UTC
All use subject to https://about.jstor.org/terms