You are on page 1of 25

Early Schooling:

The Handicap of Being Poor and Male


Doris R. Entwisle
Karl L. Alexander
Linda S. Olson
Johns Hopkins University

In trying to understand the origin of gender differences favoring girls in reading skills, analysts
have examined mainly the performance of students who are in the same grade, with samples
pooled across socioeconomic status (SES). Using a longitudinal sample in Baltimore, where all
students in a randomly selected panel are the same age and are followed from the beginning
of the first grade, the authors found that the early reading skills of boys who are receiving meal
subsidies—those who are disadvantaged—are lower than those of girls. Among children who
are not on meal subsidies, boys do about the same as girls. This gender gap that emerges over
the elementary school years is explained in terms of the higher retention rate of disadvantaged
boys, which traces back to teachers' low ratings of classroom behavior and reading skills for
boys on meal subsidies and to their parents' lower expectations for boys’ school performance.
The longitudinal design of this study, the early point from which children are followed (age 6),
and the attention given to SES differences in how parents and teachers treat boys are key dif-
ferences between this research and other studies of gender differences in reading compre-
hension. The discussion points up the critical nature of the first-grade transition in relation to
the gender gap and some of its long-term implications.

S
ocial institutions are organized along served, researchers have focused alternately
lines that reflect both common sense on the school problems of one sex and then
and society’s values. Boys and girls, the other (see Grant and Pong 2003).
therefore, generally attend school together Throughout most of the 20th century, the
because having children of the two sexes in relatively poorer performance of boys in lan-
guage arts and reading spurred research on
one school is more economical and this prac-
school practices that might have negative
tice matches social norms of gender egalitar-
effects on boys (Brophy 1985; Gates 1961;
ianism. Still, boys and girls perform different-
Powell 1967; Robinson 1955; Stanchfield 1973;
ly in school, and these differences persist Stroud and Lindquist 1942; Wilson, Burke, and
despite efforts, legislative and otherwise, to Fleming 1939). With the rise of feminism and,
make school truly equal for all. At some times, even more, with women’s wholesale entry into
it has seemed that boys are better served by the workforce, questions have come to the fore
schools than are girls, but at other times, girls about school practices that could have negative
seem better served. Reflecting society’s effects on girls, especially in math (see, e.g.,
changing opinions about which sex is better Schwartz and Hanson 1992).

Sociology of Education 2007, Vol. 80 (April): 114–138 114


Early Schooling 115

In these early years of the 21st century, how- Entwisle, and Dauber, 2003; Entwisle,
ever, boys’ reading and language issues are Alexander, and Olson, 2005, 2006). For this
once more taking center stage, probably for at reason, even seemingly small differences in
least three reasons. One reason is the convic- the ways boys and girls are treated in the
tion that reading is a pivotal skill for everyone. early grades could have long-lasting repercus-
To use a telephone book or a map, to follow sions, and we know that such differences in
instructions for assembling a toy, to pay bills, or treatment exist. In preschool, for example,
to buy groceries all depend on reading. Poor boys are allowed to spend free time climbing
reading skills are a tremendous burden irre- on blocks or crawling on the floor, while girls,
spective of gender. The second reason is par- at the teacher’s suggestion, sit at tables cut-
ents’ opinions that boys will do better in school ting things out of paper (Sadker and Sadker
if they start school later than girls. Especially if 1994). Preschool teachers are also more likely
boys have birthdays late in the calendar year, to let boys, rather than girls, call out without
parents now tend to “redshirt” their sons and raising their hands, talk loudly, or engage in
enter them later than their daughters rough play (Martin 1998).
(Bellisimo, Sacks, & Mergendoller 1995). The In elementary school, the evidence is also
third reason is the increasing gender imbalance extensive that teachers interact differently
favoring women in postsecondary education, with boys and girls. Teachers consistently rate
coupled with the fact that the national econo- girls higher than boys in deportment (Brophy
my increasingly depends on intellectual assets 1985), and much of their contact with boys
(Powell and Snellman 2004). The labor market tends to be negative and disciplinary (Bossert
that young men face is deteriorating (Zemsky 1981; Huston 1983). Parents also treat ele-
et al. 1998). Between 1971 and 2002, the earn- mentary school children of the two sexes dif-
ings of young men aged 25–34 without post- ferently and expect them to behave different-
secondary education shrank by about 30 per- ly in school. Mothers talk more and use more
cent, and the number of jobs that were avail- supportive speech with daughters than with
able to them declined (U.S. Department of sons (Leaper, Anderson, and Sanders 1998),
Education 2004, Table 14-1). Put another way, and although mothers of boys and girls report
in 1975, full-time workers without a high similar reading practices, direct observation in
school diploma earned 90 percent as much as the home says otherwise. Mothers actually
high school graduates, but by 1999, they request more information from and talk more
earned only 70 percent as much (U.S. Census about books with their daughters, and their
Bureau 2002: 1-13). conversations about stories include more
praise and encouragement for their daugh-
ters than for their sons (Porche, Ross, and
Snow 2004). Moreover, when children start
GENDER DIFFERENCES IN school, parents expect that girls will find
SCHOOLING school a more comfortable place than boys
will and expect that their daughters will get
In seeking to understand gender differences better conduct marks than their sons over the
in schooling, analysts have focused much early grades (Entwisle and Baker 1983;
more intently on the school performance of Entwisle and Hayduk 1981).
adolescents than on that of young children. When inquiring into the possible sources of
There are reasons to suppose, however, that gender differences in early schooling, it is
children’s earliest school experiences matter a important to consider socioeconomic status
great deal for their long-term educational (SES) (see Mickelson 2003), and upon closer
attainment. In fact, the literature has suggest- inspection, the gender gap in reading seems to
ed that children are launched into achieve- be characteristic mainly of children from eco-
ment trajectories when they start formal nomically disadvantaged families. For instance,
schooling or even before and that the pat- using data from the Current Population Survey,
terns of these trajectories are highly stable Bianchi (1984) found that, in poverty-level fam-
over childhood and adolescence (Alexander, ilies, boys are more likely than girls to be above
116 Entwisle, Alexander, and Olson

the modal age for grade (49 percent versus 36 regularly, and these marks predict how much
percent), whereas, in better-off families, boys children learn (Alexander and Entwisle 1988).
and girls are equally likely to be above the If there is a bias in teachers’ marks favoring
modal age (18 percent versus 17 percent). one sex over the other, that bias could be one
Similarly, using data from the National Education source of a gender gap in learning. We know
Longitudinal Study of 1988, Burbridge (1991a) that “pure” academic performance is not the
found an association between the gender gap in only thing that influences teachers’ marks. In
reading and family SES and reported that boys in the early grades especially, children’s deport-
the lowest SES quartile are much more likely ment in school is an important predictor of
than are girls to have been held back. These reading achievement. For decades, halo
findings prompted us to explore achievement effects have dotted the reading literature, and
patterns in reading comprehension for children noncognitive factors clearly predict teachers’
who were receiving meal subsidies in a panel of marks in both reading and conduct. In fact,
Baltimore children and to compare them to classroom behavior can account for as much
patterns for children who were not receiving as 50 percent of the variance in academic
meal subsidies. marks (Farkas et al. 1990; see also Lavy 2004).
We found that boys and girls began the first Parents could play a role, too. They have
grade with roughly equivalent reading scores, expectations for children’s school perfor-
but that over the elementary years, a significant mance and for how far children will go in
gender gap developed. At the end of the fifth school. They expect girls to be better
year, girls scored 18 points higher than boys behaved in school than boys, and their
(495 versus 477). Furthermore, subsidy status expectations are another strong influence on
and gender interact (the rightmost two panels children’s learning in school (see Entwisle,
of Table 1) because after five years in school, Alexander, and Olson 1997). Furthermore,
boys and girls who were not receiving meal parents’ expectations predict the specific
subsidies had equal scores (529 for both), but actions that parents take to help children
among children who were receiving meal sub- learn, and many of these actions vary by SES.
sidies, the scores of boys were significantly Higher-SES parents have higher expectations
lower than those of girls (458 versus 477). for their children’s school performance than
Thus, the approximate parity in reading do lower-SES parents. They also read to their
achievement when the children began school children more, see their children’s school
faded over the first five years, but only for stu- records more often, ensure that their children
dents in the subsidy group. What are the factors borrow books from the library in the summer,
at home and at school that account for this and take their children on more summer trips
gap? How could boys and girls in the same than do parents with lower expectations (see
classroom end up with different levels of facility Entwisle et al. 1997).
in reading? The major purpose of this article is Although SES is often taken to represent
to report on our development of a multivariate mainly parents’ economic resources, it
model that addresses this question and then to reflects much more, including parents’ tastes
use this model and related information to gain (see, e.g., Mayer 1997). When children begin
a better understanding of how gender differ- formal schooling, parents of all SES levels tell
ences emerge. The next section lays out some them about school and how they should
considerations related to formulating such a behave, both implicitly and explicitly (Hess
model. and Shipman 1965). However, these mes-
sages vary in content and style, depending on
parents’ own backgrounds, especially par-
ents’ SES. For instance, middle-SES parents
SCHOOL, HOME, AND THE advocate for their children at school more
GENDER GAP often than do lower-SES parents (Baker and
Stevenson 1986; Hess 1970; Lareau 1987,
Teachers evaluate children by way of marks 1992, 2002; Useem 1991) and give their chil-
on report cards that students and parents see dren a sense of entitlement. Middle-SES par-
Early Schooling 117

ents also tend to have the kinds of interper- erally feel toward school. That is, if boys gen-
sonal skills that foster children’s development erally find adjustment to the student role
(Hess and Shipman 1965). For example, they more difficult than girls do, then middle-class
avoid negative feedback with their children boys benefit from having parents who are
(saying no all the time) and use relatively more likely to reinforce the values and habits
complex language to structure everyday that are conducive to success as a student.
experiences (Hart and Risley 1995). Pointing The rest of this article is organized as fol-
to a picture of a lion in a storybook, for lows. After we describe the data archive, we
instance, they do not say simply “That is a present a univariate analysis of children’s
lion,” but add “Lions are animals; they eat marks and parents’ attitudes that points up
other animals, not just grass,” thus identify- gender differences. This analysis also de-
ing lions as one of a class of animals and then scribes the behaviors of teachers and parents
differentiating lions from herbivorous ani- that could bear on gendered activities and
mals. Middle-SES parents are also more likely attitudes in school or at home. Next, we pre-
to read aloud to their children to have fun or sent a multivariate model that is designed to
to start conversations (Mikulecky, 1996). The explain children’s gains in achievement in
imprint of parents’ status and class standing relation to gender differences. Then, we dis-
runs much deeper than economic and lin- cuss the model’s implications, drawing on
guistic patterns, however, and parents’ differ- qualitative information about parents’ and
ent worldviews also constrain the horizons of teachers’ behaviors related to gender. Finally,
some children but expand the horizons of we summarize the longer-term consequences
others--a “sociology of the probable” (see, of a male reading deficit and consider various
e.g., Bourdieu and Passeron 1977; Lamont policy changes that may be beneficial.
and Lareau 1988).
Relatively little attention has been paid to
how teachers’ and parents’ treatment of boys THE DATA
and girls differ along SES lines when it comes
to issues related to school. No doubt boys’ The Beginning School Study (BSS) provided
low behavioral ratings reveal real differences the necessary panel data for estimating our
in how boys and girls behave in school. Yet model. The BSS is limited to Baltimore City
because most teachers are middle class, they but is probably representative of other large
may find the behavior of middle-SES boys to eastern cities with high poverty rates. Its sam-
be more compatible with their standards than pling and research design have some particu-
the behavior of low-SES boys, since less social lar strengths for addressing issues that are
distance separates teachers from middle-SES related to gender differences in schooling.
boys. In parallel fashion, fathers of middle-SES One is that children come from a fairly wide
boys, by virtue of their own interests and range of SES backgrounds and are followed
worldviews, may model behaviors that are closely from the time they begin the first
more relevant to school than may fathers of grade. Standardized test scores for reading
low-SES boys. In other words, middle-SES comprehension and marks are available over
parents probably socialize their sons in ways the elementary school years. The children in
that are more compatible with the student the panel were randomly selected and started
role than do low-SES parents by reading for school in the same system in September
pleasure or playing board games for fun, for 1982, so grading standards, teachers’ salaries,
example. curricula, and the like were the same across
Actually, the interaction of students’ home schools.
and school experiences may hold the key to
how SES and gender interact in the schooling Sample
process. The cultural enrichment and support
for education that middle-class parents tend BSS sampling proceeded in two stages. First,
to give both sons and daughters can serve as in 1982, a random sample of 20 Baltimore
a counterweight to alienation that boys gen- City public elementary schools, stratified by
118 Entwisle, Alexander, and Olson

racial mix (6 mainly African American, 6 pre- The analysis sample (N = 403) is listwise
dominantly white, and 8 integrated) and by present. To be included, cases must have
SES (14 inner-city or working-class and 6 mid- scores on the California Achievement Test
dle- class schools), was selected. The propor- (CAT) at the beginning of Year 1 and end of
tion of African American students averaged Year 5, plus a complete set of the predictor
99.5 percent in the 6 African American schools, variables used in the regression model. In the
6 percent in the 6 white schools, and 48 per- analysis sample, 68 percent of children were
cent in the 8 integrated schools (range 17 per- on meal subsidies and 32 percent were not.
cent to 87 percent). Second, within each It is important to note that the sample is
school, about a dozen students who were start- longitudinal. Every sample member had been
ing the first grade were randomly sampled in the Baltimore school system for five years,
from each first-grade classroom (at least 2, usu- but sample members who were held back or
ally 3, classrooms per school) using kinder- who were in special education could be in the
garten lists from the previous spring supple- fourth or lower grades. The analysis sample
mented by class rosters after school began in here is thus of a different composition from
fall 1982. Ninety-seven percent of the parents cross-sectional samples based on grade level.
of the randomly chosen children consented to See Table A1 for further information.
participate in the project. The final BSS sample
consisted of 790 students who began the first Procedures
grade (nonrepeaters) in fall 1982. The city pub-
lic school system at the time enrolled about 77 When the children started the first grade,
percent African Americans (U.S. Census Bureau their race, gender, and eligibility for subsi-
1983), so to sustain black–white comparisons, dized meals (indicative of a low family
the BSS oversampled whites. As a result, the income) were determined from school
final sample was 55 percent African American. records. Parents’ education, occupations, and
The timing of BSS observations reduces expectations for their children’s marks in
some problems of endogeneity. Parents’ reading and math, as well as parents’ expec-
expectations for their children’s success in tations of how far children would go in
reading are ascertained before they see their school, were ascertained in interviews either
children’s first report cards, for example, and prior to or concurrent with the children’s
so reveal whether the parents’ expectations enrollment in the first grade. Additional inter-
favor one sex over the other before there is views with students and parents were con-
feedback from the school. Also, panel mem- ducted on a regular schedule (see Entwisle et
bers are followed up to age 22, so we could al. 1997).
see whether deficits in reading comprehen- Teacher questionnaires covered the teach-
sion early in school matter for students’ ers’ social and professional backgrounds; atti-
employment or educational attainment in tudes toward the classroom and school envi-
early adulthood. ronment; and, at the end of each year, their
Children who are eligible for meal subsidies expectations for individual children and ratings
can be compared with those who are not eligi- of the children’s personal qualities. School
ble. Using meal subsidy as an indicator of fam- records provided data on the children’s meal-
ily SES makes our findings comparable to those subsidy status, marks, retention, test scores,
of many other studies, but we stress that meal and much more. The schools administered
subsidy is an indicator that implies much more. CATs in verbal comprehension in October and
Parents of BSS children who are on meal subsi- again in May near the end of each school year.
dies differ from parents of children who are not Data on the educational attainment and
in many other ways that are relevant for chil- employment of the panel members at age 22
dren’s schooling. Subsidy level is a proxy for a came from a Young Adult Survey in 1998–99,
wide range of characteristics that are often sub- 16 years after the students began the first
sumed under family SES, including parents’ grade and 4–5 years after on-time students
educational level, job prestige level, resident graduated from high school. This survey was
father, and much else (see Appendix Table A1). conducted by telephone, when possible, or in
Early Schooling 119

face-to-face interviews. Sample attrition for girls). By contrast, among the nonsubsidy
between the first grade and age 22 was rea- parents, there was no difference in expecta-
sonably low. At age 22, 80 percent of the tions for marks by gender, either when the
original random sample provided data on children began the first grade or four years
their developmental histories and current later.
employment and/or school enrollment. Also, as in other studies, the BSS teachers
The sample that was used to estimate the consistently rated the girls’ classroom behav-
model consists of listwise present data for 276 ior higher than the boys’. At the beginning of
students who were eligible for meal subsidies the first grade, the gender gap in conduct rat-
and for 127 students who were classified as ings for children on subsidy was greater than
“nonsubsidy” (see the Appendix and that for children who were not on subsidy
Appendix Table A1), all of whom attended (.21 versus .01; see Table 1). After four years
Baltimore City public schools from 1982 to in school, a standardized measure that com-
1987. The mothers of children in the subsidy bined six classroom behaviors showed gender
group had about 10 years of schooling, and gaps for both the students on subsidy and
the mothers of children in the nonsubsidy those not on subsidy (.25 and .12).
group had about 13 years of schooling.
Retention
Poor behavioral ratings lead to retention in
DESCRIPTIVE ANALYSIS
grade (Alexander et al. 2003), and in this
Descriptive data include the children’s marks sample, by the end of Grade 1, 25 percent of
in reading and deportment and retention the boys on subsidy were retained compared
rates, as well as the parents’ expectations, all with 17% of the girls. Among students who
summarized by gender and meal-subsidy sta- were not on subsidy, the retention rate was
tus in Table 1. Girls generally get better marks much lower for both sexes and virtually the
in reading than boys do (Entwisle and Hayduk same (6 percent and 5 percent; see Table 1,
1982) even if their standardized test scores next to last row). By the end of Year 5, 59 per-
are equalized (Brophy and Good 1970; cent of the boys on subsidy versus 43 percent
Prawat and Jarvis 1980; Rehberg and of the girls on subsidy had been retained. By
Rosenthal 1978). Likewise, BSS girls’ first comparison, nonsubsidy boys and girls had
reading marks are a little higher than BSS equivalent and much lower rates of retention
boys’ (1.87 versus 1.84, corresponding to just (18 percent and 16 percent). For children on
under a C (B = 3, C =2) (see Table 1), but not subsidy, the much higher retention rate of
significantly. By the end of Year 4, however, boys than girls could, in itself, produce a gen-
girls’ marks are significantly higher than der gap in achievement scores because after
boys’. This gap (.19) is significant for the sub- five years in school, students who have been
sidy group (.24) but smaller (.11) and not sig- held back will not have gone as far in the ele-
nificant (and higher for boys) for the nonsub- mentary school curriculum as their promoted
sidy group (Table 1). counterparts. If they are in Grade 4, rather
These gender differences in the marks that than Grade 5, their progress will be assessed
children receive are anticipated in parents’ by CAT tests that are appropriate for Grade 4,
reading expectations for their children. When not for Grade 5.
the BSS children started school, the parents Boys’ marks in classroom behavior are
were asked to guess their first mark in reading lower than girls’, and boys’ marks count more
before there was feedback from the school. At than girls’ for making retention decisions.
the beginning of the first grade, the parents Given an unsatisfactory conduct rating, the
of boys on subsidy expected to see lower probability of being retained in the first grade
reading marks than did the parents of girls on is 27 percent for boys and 18 percent for girls.
subsidy, and by the end of Year 4, these dif- Teachers’ reactions to students matter in
ferences in expectations had become more another way, too, because despite the parity
pronounced (2.54 for boys compared to 2.78 by gender in standardized test scores when
Table 1. Measures of Students’ CAT Reading Scores, Marks, Classroom Behavior, and Retention in Elementary School, by Gender and Meal-Subsidy
120
Status
Total Sample (N = 403) Meal Subsidy (n = 276) No Meal Subsidy (n = 127)
Gender Gender Gender
Measure Male Female Gap Male Female Gap Male Female Gap

Reading Comprehension, Fall of Grade 1 277 282 4 272 276 4 292 293 1
Reading Comprehension, End of Year 5 477 495 18* 458 477 19* 529 529 0
Reading Mark, Beginning of Grade 1 1.84 1.87 .03 1.68 1.68 .00 2.25 2.21 -.04
Reading Mark, End of Year 4 2.17 2.36 .19* 1.90 2.14 .24* 2.88 2.77 -.11
Parent’s Reading Mark Expectation, Beginning of Grade 1 2.63 2.75 .12 2.50 2.66 .16 2.96 2.91 -.05
Parent’s Reading Mark Expectation, Beginning of Year 4 2.74 2.91 .17 2.54 2.78 .24* 3.15 3.13 -.02

Classroom Behavior, Beginning of First Gradea


Conduct 1.70 1.86 .16* 1.62 1.83 .21* 1.90 1.91 .01
Effort 1.64 1.73 .09 1.56 1.64 .07 1.84 1.91 .07
Attention 1.57 1.73 .16* 1.48 1.64 .15* 1.78 1.89 .11
Class participation 1.75 1.73 -.02 1.70 1.66 -.04 1.88 1.84 -.04
Completes assignments 1.65 1.72 .07 1.58 1.64 .06 1.84 1.88 .04
Works independently 1.51 1.63 .12* 1.42 1.53 .10 1.74 1.81 .07
Classroom Behavior Compositeb 1.63 1.73 .10* 1.56 1.65 .09* 1.82 1.86 .04
Classroom Behavior, End of Year 4
Conduct 1.60 1.81 .21* 1.52 1.75 .24* 1.83 1.92 .09
Effort 1.57 1.83 .26* 1.49 1.79 .30* 1.79 1.91 .12
Attention 1.49 1.81 .32* 1.38 1.75 .37* 1.77 1.91 .14
Class participation 1.71 1.84 .13* 1.65 1.80 .15* 1.85 1.91 .05
Completes assignments 1.59 1.85 .26* 1.51 1.80 .29* 1.79 1.95 .16*
Works independently 1.62 1.80 .18* 1.58 1.76 .18* 1.72 1.88 .16*
Classroom Behavior Compositeb 1.60 1.82 .22* 1.53 1.77 .25* 1.79 1.91 .12*
Retention in Grade 1 .20 .13 .07 .25 .17 .08 .06 .05 .01
Cumulative Retention Through Year 5 .48 .33 .15* .59 .43 .17* .18 .16 .02
aMeasures of classroom behavior are coded 1 = needs improvement, 2 = satisfactory.
Entwisle, Alexander, and Olson

bClassroom Behavior Composite is the mean of at least 4 of the 6 possible work-habit ratings.
*p < .05.
Early Schooling 121

BSS children begin school, teachers place children on subsidy and for those not on sub-
more boys than girls in low reading groups, sidy to pinpoint the factors that help explain
especially among subsidy students: 34 percent the gap.
of boys on subsidy are in low reading groups Students arrive at school with some read-
compared to 24 percent of girls. By the end ing skills, so the beginning CAT score serves
of Grade 1, 36 percent of boys and 31 per- as a baseline against which to estimate how
cent of girls on subsidy have not advanced much other variables contribute to five-year
from the primer to the Grade 1 reader, com- gains in reading achievement. Gender is the
pared with just 11 percent of boys and 13 first independent variable in the model, and
percent of girls who are not on subsidy. These by entering the other variables in the model
patterns suggest another reason why stu- sequentially after it, we can see how the gen-
dents on subsidy fail to progress as fast as der coefficient changes as other variables are
those who are not on subsidy, especially boys. entered. Race and family SES, which are
In sum, lower reading marks, lower conduct entered next, control for differences in race or
marks, tracking differences beginning in the SES within the meal-subsidy groups. Parents’
first grade, and gender differences in weights psychological support, entered next, is a
assigned to conduct in making retention composite of parents expectations for chil-
decisions all contribute to higher retention dren’s marks in reading and math, plus their
rates for the boys on subsidy. estimates of how far the children will go in
school (see Table A1). Inserting parents’ sup-
port in the model tests whether it can explain
MULTIVARIATE ANALYSIS part of the gender gap. Parents’ support does
predict students’ levels of school achievement
The multivariate model (see Figure 1), which (Entwisle and Hayduk 1982; Entwisle et al.
is proposed to explain achievement gains, 1997), and the descriptive information
was adapted from an earlier one (see Entwisle reviewed earlier suggests that parents’ influ-
and Alexander 1996). By estimating test ence could shape classroom behavior in ways
scores at the end of Year 5 and controlling for that lead to gender differences.
the initial test score, the model evaluates chil- Marks are entered next. Girls’ better marks
dren’s gains in reading comprehension from could lead girls to learn more than boys, and
the beginning of Year 1 to the end of Year 5. students’ marks in reading by the end of
This model can be estimated separately for Grade 1 already favor girls.

Figure 1. Conceptual Model Predicting Gains in Scores on Reading Comprehension Tests over
Elementary School
122 Entwisle, Alexander, and Olson

Students’ reading achievement responds model can be read as test score points.
to conduct and classroom behavior (Entwisle
and Alexander 1996; Farkas et al. 1990), so Children on Subsidy Girls on subsidy gain
children with low ratings in classroom behav- 17.3 points more than do boys in reading
iors (here a composite variable including con- achievement over the first five years of school
duct, level of effort, class participation, and (column 1, middle panel, Table 2). Adjusting
the like) would not be expected to gain as for any imbalance between boys and girls by
much on achievement tests as would those race or SES increases the gender coefficient
with more favorable ratings. To allow for pos- by only 1.5 points (17.3 to 18.8). Parents’
sible changing levels of behavior over the support reduces the gender coefficient by 2.2
early years of school, one behavior composite points (18.8 – 15.6). This reduction is consis-
was constructed for Year 2 and another was tent with the descriptive analyses that sug-
constructed for Year 4, and both are inserted gest that (1) parents of children on subsidy
in the model. think that girls will do better than boys, and
Finally, a dichotomous variable indicates (2) parents of children on subsidy have ideas
whether the child has been retained over the about gender-role differentiation that could
first five years of school. Retained children do help their daughters adapt to the student role
not gain as much on standardized tests as do more easily than their sons. Entering the coef-
their promoted peers (Alexander et al. 2003), ficient for reading mark reduces the gender
and in the BSS panel, as elsewhere (see coefficient by 4 points (to 11.6). Teachers’
Hauser, Pager, and Simmons 2000), more marks do favor girls, as was noted earlier, and
boys than girls are retained. Even more this imbalance accounts for some of the gap.
important here is that retained children are Students’ behavioral ratings in Year 2 reduce
taking CAT tests below the Grade 5 level. the gender coefficient by over 5 points, from
While there is little doubt that test scores 11.6 to 6.1, by more than any previous vari-
at the end of elementary school are related to able, and the gender difference is no longer
the predictors (beginning test scores, marks, significant. The Year 4 behavioral ratings
parental expectations, students’ behavior, reduce it further, by 3.2 points. Entering the
and grade retention), this model focuses retention variable finally reduces the gender
attention not on the strength of these indi- coefficient to less than 1 point.1 The full
vidual coefficients, but on how these predic- model thus explains 44 percent of the vari-
tors relate to the relative size of the gender- ance in gains in test scores among children
gap coefficient—that is, how the gender gap on subsidy and accounts for the gender gap
diminishes (or grows) as we control for each in test scores, implicating mainly parents’
of these predictors. Furthermore, by estimat- support, reading marks, behavioral ratings,
ing the model separately by meal-subsidy sta- and retention rates as factors that favor girls.
tus, we can determine whether these predic-
tors produce a gender gap in gains in test Children Not on Subsidy Gains in CAT
scores among subsidy students, nonsubsidy scores of the nonsubsidy group differ by gen-
students, or both. der only a little in the unadjusted model (b =
-0.8), favoring boys. With race and SES
Estimating the Model added, a small gender difference favors girls,
and the addition of the other independent
Gender coefficients are highlighted in bold in variables does little to change the gender
Table 2. At the start, the gender coefficient is gap. The gender gap is thus not significant
17.3 and significant for the subsidy group for nonsubsidy children at the beginning of
and –0.8 and not significant for the nonsub- Grade 1 or after five years in school.
sidy group (see Table 2, column 1, subsidy
and nonsubsidy panels). This difference is Yearly Growth The focus of the model, as
reflected in the significant gender x subsidy noted, is not on the relative strength of the
status interaction term (see Table 2, the total individual predictors, but on how the individ-
sample). Note that metric coefficients in the ual predictors influence the size of the gender
Table 2. OLS Regression Model Predicting Reading Comprehension at the End of the Fifth Year of School, by Meal-Subsidy Status (Metric
Coefficients Measure Gains in Test Scores over Elementary School)

Total Sample Meal Subsidy No Meal Subsidy


(N = 403) (n = 276) (n = 127)
Early Schooling

1 2 1 2 3 4 5 6 7 1 2 3 4 5 6 7

Reading CAT, Fall Grade 1 .27* .73* .50* .47* .42* .26* .23* .23* .21* .90* .79* .55* .39* .39* .38* .37*
Gender -0.0 0.9 17.3* 18.8* 15.6* 11.6 6.1 2.9 -0.2 -0.8 2.8 1.5 5.7 5.8 2.4 0.9
Race -8.9 -8.6 -16.8* -7.0 -4.7 -4.4 -3.9 -23.3* -22.4* -17.2 -17.2 -15.5 -15.1
Family SES 12 .6* 32.0* 19.5 14.6 12.1 10.7 6.1 28.1* 19.0* 14.7 14.7 15.1* 18.4*
Parent’s Psychological
Support 12.4* 24.5* 12.6* 11.3* 11.0* 11.0* 31.3* 16.8* 16.9* 16.5* 13.7
Reading Mark, End of
Grade 1 17.8* 33.4* 31.3* 30.2* 17.2* 27.1* 27.1* 25.9* 16.1*
Classroom Behavior
Composite, Year 2 15.0 40.2* 34.4* 20.9* -0.9 -19.5 -24.2
Classroom Behavior
Composite, Year 4 21.6* 18.6 21.2 33.6 31.4
Retained in Grade,
Years 1–5 -40.1* -39.6* -49.4*
Meal-Subsidy Status by
Sex Interaction 39.5*

R2 .55 .23 .10 .13 .19 .35 .38 .38 .44 .26 .34 .42 .47 .47 .47 .51

*p < .05.
123
124 Entwisle, Alexander, and Olson

coefficient. In fact, the predictor coefficients tests used for the fourth (or earlier) grade, so
are, for the most part, similar for both subsidy these debits are to be expected. What is impor-
and nonsubsidy students. What is different is tant is that these huge retention coefficients
how they come together over the course of lead to a relatively small decline in the size of
the elementary school years to produce a sig- the gender coefficient. It is the precursors of
nificant gender gap in test scores among sub- retention, like the poor reading marks and
sidy children but not among nonsubsidy chil- classroom behavior seen at the end of Year 2,
dren. The model thus illustrates how, for sub- that change the gender coefficient. Retention is
sidy students, the female advantage in not the root cause of the gender gap; it is the
parental expectations, reading marks, class- lower reading and behavior marks of boys lead-
room behavior, and retention, as laid out in ing to retention that depress their test scores,
Table 1, together translate into lower reading not retention itself. Retention is an administra-
comprehension scores for boys at the end of tive response to unsatisfactory classroom per-
five years of elementary school. Since there formance.
are few differences in these predictors The gender gap in retention rates follows
between male and female nonsubsidy stu- the same trend as the gender gap in test scores
dents (see Table 1), it is not surprising that for (see Table 4). At the end of Year 1, the gender
these students, the parallel regression model difference in retention rates is 8.2 percent for
reveals no relationship between the predic- subsidy children and 0.6 percent for those not
tors and the gender coefficient. What are the on subsidy, and at the end of Year 2, it is still siz-
patterns over time? When we compare annu- able (5.7 percent versus 0.6 percent). These
al growth across subsidy groups (see Table 3), retention figures parallel the large gender gap
the gender gap is negligible for both groups in test scores that emerges in the first two years
in Year 1. At the end of Year 2, however, a sig-
and then continues. By the end of Year 5, the
nificant difference emerges: 23 points favor
gender gap in retention rates has grown to
girls in the subsidy group and 6 points favor
16.7 percent among children on subsidy,
boys in the nonsubsidy group. In Years 3, 4,
whereas the gap among nonsubsidy students is
and 5, the patterns stabilize, probably
negligible (1.8 percent).
because retention rates are much the same by
gender.
The cumulative five-year retention coeffi-
cient (see Table 2) is large for both subsidy and A RICHER PICTURE
nonsubsidy children (-39.6 and –49.4), show-
ing big debits in test scores associated with The multivariate model does a good job of illus-
retention (see Table 2). As we noted, however, trating how the female advantage among sub-
retained children are taking the curriculum and sidy students in parental expectations, reading

Table 3. Gender Gap in CAT Reading Scores over the First Five Years of School, by Meal-
Subsidy Status (N = 392)a

Meal Subsidy (n = 272) No Meal Subsidy (n = 120)

CAT Reading Comprehension Male Female Gap Male Female Gap

Fall, Year 1 274.0 273.1 -0.9 295.4 292.2 -3.2


Spring, Year 1 331.2 336.5 5.3 358.3 361.6 3.3
Spring, Year 2 364.2 387.6 23.4* 410.7 404.8 -5.9
Spring, Year 3 395.4 409.6 14.2* 451.3 454.2 2.9
Spring, Year 4 432.1 455.8 23.7* 495.1 490.9 -4.2
Spring, Year 5 463.0 480.3 17.3* 536.2 530.2 -6.0

aRequiring six annual test scores reduces the case base to 392, different from the analysis sample.
*p < .05.
Early Schooling 125

Table 4. Gender Gap in Retention Rates over the First Five Years of School, by Meal-Subsidy
Status (N = 403)

Meal Subsidy (n = 277) No Meal Subsidy (n = 127)


Male Female Gap Male Female Gap

Year of First Retention (%)


Year 1 25.2 17.0 8.2 5.9 5.3 0.6
Year 2 16.3 10.6 5.7 5.9 5.3 0.6
Year 3 10.4 8.5 1.9 2.0 2.6 -0.6
Year 4 5.9 3.5 2.4 2.0 2.6 -0.6
Year 5 1.5 2.8 -1.3 2.0 0.0 2.0
Cumulative Retention Rate
Through Year 5 59.3 42.6 16.7 17.6 15.8 1.8

aRequiring six annual test scores reduces the case base to 392, different from the analysis sample.
*p < .05.

marks, classroom behavior, and retention sure of gender preference, which is at least par-
together predict higher reading achievement tially independent of academics, the teachers’
scores for girls over the first five years of ele- choices were strongly imbalanced. The vast
mentary school. This bare-bones portrait can majority of teachers (72 percent) selected more
be considerably enhanced by descriptive data girl than boy helpers, 26 percent selected more
that are available from the BSS. As the model boy helpers, and only 2 percent selected an
demonstrates, classroom behavior (classroom equal number of boy and girl helpers. (Note
fit) is at the heart of the issue. that tasks like cleaning chalkboards or moving
equipment are not female sex typical.) These
preferences matter because girls in classes
The Teacher and Student Fit where teachers preferred girl helpers gained
No doubt boys’ low behavioral ratings reflect more on standardized tests over the fourth year
real differences in how the two sexes behave in than did boys (39 versus 32 points, p = .086),
with race, meal subsidy, parent’s education,
school, but part of this poor “fit” may also
parent’s estimate of student’s ability, parent’s
come from teachers’ bias. Most studies of
expectation of student’s mark, family configu-
teachers’ bias have focused on differences in
ration, school segregation status, and school-
students’ social class or race, rather than on
level parent education all controlled. In classes
gender (see Alexander, Entwisle, and
where teachers preferred boy helpers (a smaller
Thompson 1987), but the margin in conduct group), the difference favored boys but was not
marks favoring girls over boys appears in almost significant (see Entwisle et al. 1997). Probably
every BSS school irrespective of SES or race. when teachers favor one sex over the other,
Hinshaw (1992) suggested that gender bias they act in ways that encourage achievement
may vary from teacher to teacher. If bias springs by children of that sex, such as by smiling at
from personal sources, gender favoritism them, encouraging them to answer, and
should vary from teacher to teacher. To test this choosing them as helpers. In short, subtle kinds
idea, earlier we took a “projective” approach of behavior by teachers could benefit girls more
and asked teachers in Year 4 to name students than boys.
they would pick to carry out nonacademic tasks
(Entwisle et al. 1997). Each teacher named
Parents and Student Fit
three students whom she “could or would rely
on to help” with each of five tasks, such as tak- Parents generally expect girls to get better
ing notes to the office, passing out papers, marks than boys in reading (see, e.g.,
cleaning chalkboards, helping with a television Entwisle and Baker 1983). When BSS children
or other equipment, and so on. On this mea- begin the first grade, their parents expect
126 Entwisle, Alexander, and Olson

similar reading marks, but by the fourth jobs (Barnett 1979; Maccoby and Jacklin
grade, the parents of children on subsidy 1974). Low-SES parents have more dis-
expect boys’ marks to be lower than girls’ crepant goals for sons and daughters than do
(2.5 versus 2.8, where B = 3 and C = 2; see higher SES parents. When BSS parents list the
Table 1), but parents of nonsubsidy boys do jobs that they expect their children to have as
not (3.2 versus 3.1). Expectations, which are adults, the parents of both subsidy and non-
one component of parents’ psychological subsidy children expect their sons’ future jobs
support, thus favor girls among parents of to be almost the same in terms of sex typical-
children who are on subsidy but not among ity. By contrast, for daughters, the parents of
parents of children who are not on subsidy. nonsubsidy students expect jobs to be in
More subtle influences could be driven by fields where about half the participants are
parents’ conceptions of the student role. male, while the parents of subsidy students
Boys’ gender roles are generally thought to expect the large majority of daughters’ jobs
be less consistent with the student role than to be stereotypically female.
are girls’ (see Brophy 1985 for a review), These tendencies could be further reinforced
especially among working-class families. by fathers’ occupations. BSS fathers of children
While both lower-class and middle-class boys on subsidy have a 10th-grade education, and
reject the “effeminate” boy image and/or the 70 percent hold craft, transportation, or labor-
boy with nonmasculine interests, middle-class er jobs—fields in which only 10 percent of the
boys are more accepting than are lower-class workers are women. The fathers of nonsubsidy
boys of the academically oriented boy (Hall children, by contrast, have almost two years of
and Keith 1964; Pope 1953) even in the early postsecondary education, and 42 percent work
grades (see Hess 1970). in professional/technical or managerial jobs—
Two kinds of indirect evidence about role fields in which about half the workers are
conceptions suggest that the BSS boys on women. Such differences in fathers’ jobs could
subsidy (who grew up in the mid-1980s) send different messages to sons about men’s
were encouraged at home to have more tra- work and the relevance of school for their
ditional sex-role ideas than were those not on future work.
subsidy: (1) parents’ chore assignments and All in all, the social climate and atmos-
(2) parents’ expectations for children’s future phere of the home are probably more com-
jobs. patible with nonsubsidy boys’ fit to the stu-
Attitudes about gender roles surface when dent role than are those of boys on subsidy.
parents assign chores (Rogoff 2003). Lower- Role fit, in turn, may be reflected in boys’
income parents assign household chores to conduct and effort to conform to behavioral
sons and daughters in line with traditional sex standards in the classroom and in boys’ effort
roles more often than do better-off parents to master subjects, such as reading. Students’
(Lackey 1989; White and Brinkerhoff 1981). marks, sex-role expectations, and attitudes
BSS data on chores suggest that parents of toward the teacher, as well as parents’ atti-
children on subsidy treat sons in ways that are tudes and teachers’ reactions, probably all
more consistent with traditional ideas about converge to shape the child’s fit to the stu-
male sex roles than do better-off parents; for dent role. Parents’ assignments of chores and
example, parents of subsidy children assign job hopes shape students’ sex-role standards,
chores in meal preparation significantly more and these sex-role standards influence the
often to girls than to boys, but better-off par- “fit” of the student role. Low-SES parents’
ents do not. With regard to baby-sitting, at emphasis on traditional male sex roles could
age 8, only a small fraction of boys baby-sit, affect the classroom behavior of boys on sub-
but starting at age 11, boys on subsidy are sidy. Students’ marks, teachers’ expectations,
less likely than are better-off boys to baby-sit, personal preferences of teachers, and teach-
and by age 13, they are significantly less like- ers’ returning the liking of students all
ly to do so (54 percent versus 44 percent). respond to students’ role fit. For example,
Parents’ sex-role expectations are also teachers’ personal preference for girls over
revealed in parents’ views of children’s future boys, as is apparent in the choice of class
Early Schooling 127

helpers, is probably influenced by students’ United Kingdom. Moreover, the BSS shows
behavior in class. For all boys, the “fit” to the that the gender and SES of first graders pre-
student role is not as good as the fit for girls, dict educational attainment at age 22
but for nonsubsidy boys, the fit is better than (Entwisle et al. 2005). The findings of these
it is for boys on subsidy. and other studies suggest that taking chil-
Retention and other kinds of low tracking dren’s early school experience into account
are more common for boys on subsidy than could improve our understanding of how dif-
for the other subgroups. The poorer fit of ferences in gender and SES emerge in educa-
boys on subsidy into the student role, which tional attainment, especially reading.
begins to manifest itself in the first grade and U.S. boys and girls generally start their for-
worsens with time, is probably the major rea- mal schooling together. However, concern
son for these boys’ higher retention rate or about boys’ schooling is now on the rise
placement in low tracks. because national data have increasingly
Retention is a handicap in terms of further pointed to deficits in boys’ performance early
schooling for BSS students, and students on in school. In kindergarten, boys have poorer
subsidy pay a higher toll for retention than reading skills than do girls (Tach and Farkas
other students do. At age 22, among 2006; West, Denton, and Reaney 2001), and
retainees who had been on subsidy, 44 per- boys’ basic reading skills, such as letter recog-
cent of the men and 52 percent of the nition and sound-to-letter correspondences,
women were neither at work nor in school, are lower than those of girls at the beginning
while among the nonsubsidy retainees, 21 of the first grade (U.S. Department of
percent of the men and 24 percent of the Education 2002: 125). By the fourth grade,
women were not at work or in school. 42 percent of boys, compared to 32 percent
Actually, being idle at age 22 is not unusual of girls, are “below basic” in reading achieve-
for retainees who had been on subsidy; since ment (U.S. Department of Education 2001,
they had left high school, almost half the time Table 7.1). Even more worrisome, 20.3 per-
they had not been in school or at work, com- cent of boys aged 6–8 are below the modal
pared to about one-fourth the time for other grade for their age compared to 15.6 percent
retainees. Thus, low behavioral ratings and of girls of that age, and by ages 10–11, this
low reading marks in the first grade elicit an gap is even larger (29.2 percent versus 21.7
institutional reaction, namely, retention— percent) (see Hauser 2001:163). That boys
with profound implications. In the BSS, falling are below modal grade for age does not
into the retention track jeopardizes a child’s reflect only retention rates—some children
chance to get any postsecondary education, start school later than do others—but statis-
without which today’s workers are at a severe tics on retention show that modal-age com-
disadvantage in the labor market. In 2000, for parisons underestimate the gender gap. In
example, close to one in five African 1995, for example, 11 percent of second-
Americans with only a high school education grade boys versus 5 percent of second-grade
was unemployed (U.S. Census Bureau 2001, girls repeated a grade (U.S. Department of
Table 260). In addition, among workers with Health and Human Services 1997; see also
a high school education or less, men’s wages Jacob 2002).
are falling (Mishel, Bernstein, and Allegretto Boys’ poorer performance in elementary
2005). school has serious long-term implications.
Boys with low marks in the first grade are
twice as likely to drop out as are boys with
DISCUSSION better marks, even when they have the same
standardized test scores (Ensminger and
Although no national study in the United Slusarcick 1992). Still more ominous is the
States has yet directly linked school experi- higher retention rate of boys. Retention pre-
ence in the primary grades to educational dicts dropout (Alexander et al. 2003; Darling-
outcomes in young adulthood, Kerckhoff Hammond and Falk 1995; Reardon 1996),
(1993) reported such long-term ties in the and dropouts have lower lifetime earnings
128 Entwisle, Alexander, and Olson

and higher unemployment rates and incar- success early in school can depend on inter-
ceration rates and are more dependent on personal dynamics in the classroom. Lower-
welfare (McDill, Natriello, and Pallas 1986; SES parents, however, tend to convey mes-
Royce, Darlington, and Murray 1983). sages to their children about behavior in class
In the BSS data, the male deficit in reading that are vague and lacking in information
skills can be traced to children on subsidy. compared to the messages given by better-
These children’s family backgrounds resemble off parents. For example, when Hess and
those of students in the lowest quartile of Shipman (1965: 877) asked mothers what
national samples in which a similar male they would tell children who were starting
deficit appears (Burbridge 1991b). In the BSS school, a higher-SES mother said: “First of all,
panel, boys on subsidy came from families in I would remind her that she was going to
which the average parent had a 10th-grade school to learn, that her teacher would take
education, which is substantially below the my place . . . and that any questions or prob-
U.S. average. The strong correlation between lems that she might have she could consult
SES and children’s school achievement is with her teacher for assistance.” In contrast, a
widely recognized (see, e.g., Haveman and lower-SES mother said, “Well, John . . . You
Wolfe 1994; Mayer 1997; U.S. Census Bureau must know how to behave the first day at
2003), but the interaction between SES and school. You should be a good boy and should
gender has attracted much less attention. do just what the teacher tells you to do.” The
The patterns laid bare in this article rein- second mother described the student role as
force thinking about the critical nature of the passive and compliant. Her message was
first-grade transition. The retention rate tends vague and lacked information about how to
to be the highest in the first grade, which deal with problems at school except by pas-
then evolves here into the large gender gap sive compliance.
in reading comprehension at the end of Year Six year olds can be noisy and inattentive,
2. The consequences of a rough first-grade but most do not misbehave deliberately. On
transition in terms of test scores are delayed, this account, it is worrisome that gender, a
but the die is cast in the classroom activity of nonacademic characteristic, can be so conse-
Year 1, so the key time for intervention is in quential for the reading skills of children from
Grade 1 or before. lower-SES backgrounds. Just as many BSS
Earlier, we speculated that classroom “fit” boys as girls say they like the teacher “a lot”
responds to sex-role socialization and that when they begin school. The first-grade
sex-role ideas could spring from sources as teachers, however, think that more than
diffuse as hearing about fathers’ and other twice as many of the boys on subsidy as the
men’s jobs or from observing that mothers girls (38 percent versus 17 percent) “need
are more involved with school than are improvement” in conduct in the first grade.
fathers (Lareau, 2002). Fit could respond Teachers also rate boys as less attentive than
even more to direct sources, such as parents girls in the first grade and see the boys on
encouraging boys to adopt traditional male subsidy as becoming more inattentive the
sex-role behaviors. Among higher-status par- longer they are in school. Deportment, work
ents, male sex-role expectations were evolv- habits, and other personal characteristics are
ing away from traditional ideas even by the “noncognitive” and so are seldom seen as a
early 1980s when the BSS panel members major issue in improving school achievement.
were starting school. At that time, however, Even so, in the first grade, inattentive children
traditional male sex-role expectations could do not learn as much as do their more order-
still be functional for men with blue-collar ly peers. It is dangerous, however, to link the
jobs. The precipitous decline in manufactur- noncognitive characteristics of any group
ing and other blue-collar work since then, with their ascriptive characteristics. Layper-
however, makes such roles less functional and sons and policy makers alike would properly
schooling even more important for males object if ethnicity were taken as the root
than was true at that time. cause of low marks. Likewise, to observe that
Most parents sense that their children’s boys on subsidy get lower marks in conduct
Early Schooling 129

and then to conclude that gender (an ascrip- ably benefit from more, rather than less,
tive characteristic) is the basic reason why recess time because children, especially boys,
these boys consistently make less progress are more attentive after recess than before,
than girls in elementary school is dangerous. whether breaks are outdoors or indoors, are
The parallel between school inequity linked to short or long (Pellegrini et al. 2002). Inner-
gender and school inequity linked to minori- city boys, especially, use their facility with
ty-group membership is obvious: Boys do get games (ball games and chase and jumping
lower marks than girls in elementary school, games) as a way to adjust to early schooling.
but being male does not, by itself, “produce” Their success in games at recess predicts bet-
these low marks any more than ethnicity, by ter adjustment in the classroom. Curiously,
itself, produces low marks. Our analyses sug- students in secondary schools have 5–10
gest that children’s adaptation to the class- minute breaks every hour to move from one
room, as shaped by teachers’ and parents’ classroom to another, but frequent breaks are
views of children, undercuts the schooling of not the rule in elementary school.
boys on subsidy. Still another approach would be to give
Part of the remedy could be for schooling disadvantaged children more explicit instruc-
in the primary grades to be restructured to fit tion and practice in classroom behavior early
the profiles of all children. As Kimmel (2000) in the first grade or in kindergarten. In a time-
said, the facile “boys will be boys” model use study, Berkeley (1978) showed that
needs to be challenged. If teachers find boys’ kindergarten teachers in a suburban middle-
conduct to be problematic or the conduct of SES school spent more time teaching class
poor children of both sexes to be more prob- routines—not calling out and lining up prop-
lematic than that of better-off children, as erly—than did teachers in a working-class or
they seem to, these issues flag problems that inner-city school. Teachers in the latter
need solutions. schools seemed less aware of the need to
One change that may be effective address- accustom children to classroom routines.
es both classroom behavior and sex-role Gender differences that are prejudicial to
issues: recruit more male teachers for the ear- males no doubt have been downplayed
liest grades, especially in schools where many because of the zeitgeist that prevailed
students come from disadvantaged families. through most of the 20th century, when men
(Of the BSS teachers, none in the first grade with a minimal education could obtain good
and 2 percent in the second grade are men.) jobs in mines, factories, construction, and the
Nationally, the proportion of male teachers in skilled trades. Now, however, the labor mar-
elementary schools has been dropping (from ket increasingly seeks workers with more
16 percent in 1982 to 14 percent in 2003; see schooling, and women have responded more
U.S. Census Bureau 2004–05, Table 238), and than have men. In 1970, the ratio of college
these male teachers are concentrated in the enrollments favored men by a wide margin
upper, rather than the lower, grades. Boys’ (59 to 41), but now the ratio favors women
associating closely with a teacher who is an (54 to 46) (Riordan 2003). In framing policy,
example of a white-collar man may, in itself, however, it is important to acknowledge that
help reinforce male sex-role standards that the roots of this gender imbalance in higher
are more consistent with the student role. education extend back to the early grades
Also, we suspect that male teachers may (Bellisimo et al. 1995; Cameron and Wilson
interact with boys differently from female 1990; Entwisle et al. 2005). Moreover, while
teachers and be more tolerant of restlessness, dropout issues are often cast in terms of sec-
moving about, noise levels, boisterousness, ondary school structure or organization, edu-
and the like (see Dee 2005). cational trajectories begin much earlier
Another change, and one relatively easy to (Brooks-Gunn, Duncan, and Maritato 1997;
make, is in class schedules. In the United Ensminger and Slusarcick 1992). A simple
States, recess policies vary widely, but in the logic and steadily mounting evidence suggest
early grades, recess time is declining that tackling the problems early in life that
(Pellegrini and Bohn 2005). Boys would prob- lead to dropout and labor market disability
130 Entwisle, Alexander, and Olson

holds more promise (see Reynolds, Ou, and measure of classroom behavior entered prior
Topitzes 2004). to reading mark and parental support, even
though time ordering compels the order of
variables in the original model. The alterna-
NOTE tive model shows, as the reviewer suspected,
a larger coefficient for classroom behavior. In
1. At the suggestion of a reviewer, we esti- fact, by itself, this coefficient is large enough
mated an alternative model with a combined to explain the gender gap.

APPENDIX

Individual Child and Family Variables


Race. Race is 0 for white, 1 for African American.
Sex. Sex is 0 for boys, 1 for girls.
Meal-subsidy status. Data on child's eligibility for free or reduced-price meals came from
school records. Children from families with incomes that were less than 1.85 times the feder-
al poverty level were eligible for subsidies.
Family SES composite. Family SES is the average of five z-scores: mother’s and father’s edu-
cational and occupational levels as obtained from the parent questionnaires and whether or
not the student qualified for a federal meal subsidy, also obtained from school records. The
average education of the mothers (of 95 percent of the students) was 11.7 years and of the
fathers (of 67 percent of the students), 12.2 years. For SES, there was 99.5 percent coverage,
with all but 0.5 percent of the students providing data for at least one component. The alpha
reliability (5 items) is .86 and ranges from .74 to .84 for three items, available for over 95 per-
cent of the sample.
Parent's psychological support. Each year before the first report card, a parent (mostly the
mother) guessed the child's first marks in reading and math (4 for excellent, 3 for good, 2 for
satisfactory, and 1 for unsatisfactory) and how far the child would go in school (1 for not fin-
ish high school, 2 for finish high school, 3 for 1–2 years of college, 4 for finish 4 years of col-
lege, and 5 for more than college). Parent’s psychological support is the mean of these three
z-scores.
Parents’ occupational aspirations for the student. Beginning in Year 1, the parents were asked:
“What kind of occupation or job would you like your child to have when he [or she] grows
up?” and “What kind of job do you expect him [or her] to get?” These occupational aspira-
tions and expectations were assigned sex-typicality ratings (percentage of female job incum-
bents) derived from the job titles of the Bureau of Labor Statistics (U.S. Department of Labor
1984). Thus, a rating of 80 percent would be a job in which 80 percent of the workers are
female.
Home chores. In Years 3, 6, 7, 8, and 9, parents were asked about their children's chores:
makes own bed, helps with meals, runs errands to store, baby-sits younger children, mows the
grass, washes the car, and helps with other housecleaning (vacuuming, sweeping, and so
forth).
California Achievement Test scores (CAT). Children's CAT Reading Comprehension scores
(Form C) in October and May of each school year came from the school records. See California
Achievement Test (1979).
Marks. Students’ marks in reading, obtained from report cards, were coded 4 for excellent,
3 for good, 2 for satisfactory, and 1 for unsatisfactory.
Classroom behavior ratings. The student’s classroom behavior rating for each quarter is the

continued
Early Schooling 131

APPENDIX CONTINUED

mean of at least 4 of 6 possible ratings—conduct, effort, pays attention, class participation,


completes assignments, and works independently, coded 2 for satisfactory and 1 for needs
improvement. The yearly rating is the mean of the quarterly ratings. Internal reliability
(Cronbach’s alpha) for the classroom behavior composite is .92 for both Years 2 and 4.
Satisfaction with teacher. Beginning in the first grade, students were asked: “Do you like
your teacher a lot, think he [or she] is just OK, or not like him [or her] much at all?” as adapt-
ed from a 4-item School Satisfaction Scale developed from the Epstein and McPartland (1977)
Quality of School Life Scale.

School Tracking Variables


Reading groups. Teachers reported reading-group placements in the fall and spring of
Grades 1 and 2. Teachers reported the number of groups in a class, and to make comparisons
across classrooms with a different number of reading groups, students were ranked as follows:
1 = lowest group, 2 = intermediate-level group, or 3 = highest-level group. The reading-group
level for Grade 2 is available only for students who did not repeat Grade 1.
Reading instructional level. Report cards show the instructional level of the reader used by
the student every quarter.
Retention. Grade retention (from school records) was coded 1 for retained and 0 for pro-
moted. A cumulative measure for Grades 1 through 5 was 1 for retained at least once and 0
for never retained. Retention rates were nearly 17 percent in Grade 1 and 10 percent in Grade
2, with 37 percent the cumulative rate through Grade 5.

Teacher Variables
Gender bias in classroom chore assignment. In Year 4, the teachers were asked to name three
students in their class “on whom you could or would rely to help” with each of five classroom
chores: take notes to the office; pass out or collect papers; clean chalkboards, put up posters,
and fix bulletin boards; help with a television or other equipment; escort visitors around the
school. The students’ names were categorized by gender, and the total number of girls and
boys who were chosen as helpers was tallied for each teacher.

Attrition
Table A2 compares the analysis sample (N = 403) to the original random sample (N = 790).
After five years, the sample included more African Americans (62 percent), but otherwise the
composition of the two samples was similar; for instance, beginning test scores in verbal com-
prehension were 281 and 280 in the two samples, average reading marks were also close
(about 1.9 or just under a C average), retention rates matched well (37 percent versus 40 per-
cent), and demographic characteristics (except for race) were close.
Table A1. Definitions of Variables, Means, and Standard Deviations of the Primary Variables Used in the Regression Analysis (N = 403)
132

Variable Mean (SD) Description

Background Characteristics
Student’s gender .54 (.50) 1 = Female, 0 = Male
Student’s race .62 (.49) 1 = African American, 0 = white

Family Socioeconomic Status Composite -.13 (.74) Average of both parents’ educational level, a ranking of mother’s and father’s
occupational status (TSEI2; see Featherman and Stevens 1982), and participa-
tion in the federal meal-subsidy program (all measures converted to z-scores)

Proportion on Meal Subsidy .68 (.47) Participation in the federal meal-subsidy program
0 = No subsidy, 1 = Meal subsidy
Meal Subsidy
Mother’s years of education 10.75 (1.98) Socioeconomic Characteristics of Families, by Meal-Subsidy Status
Father’s years of education 10.95 (1.97) *Examples of TSEI2 job status scores:
Mother’s TSEI2 job status* 25.90 (9.92) 25.23 Typist, 25.37 Salesclerk
Father’s TSEI2 job status* 25.37 (9.64) 23.96 Bartender, 24.47 Machinist
Proportion resident father .43 (.50)
No Meal Subsidy
Mother’s years of education 13.29 (2.65)
Father’s years of education 13.28 (2.82)
Mother’s TSEI2 job status* 41.23 (17.95) 41.58 Sales manager; 42.69 Photographer
Father’s TSEI2 job status* 38.98 (19.51) 38.43 Building managers; 39.20 Radiology technicians; 39.43 Surveyors
Proportion resident father .79 (.41)

Parental Support
Reading mark expectations 2.69 (.78) Parent’s expectations for student’s reading and math marks, beginning of Grade 1:
Math mark expectations 2.67 (.76) 1 = unsatisfactory, 2 = satisfactory, 3 = good, 4 = excellent

Educational Expectations 3.04 (1.13) Parent’s expectations for student’s educational attainment:
1 = not finish high school, 2 = finish high school, 3 = 1–2 years of college,
4 = 4 years of college, 5 = > 4 years of college

Parental Support Index -.01 (.78) Composite is mean of z-scores for parent’s expectations for student’s reading and
math marks and educational attainment
continued
Entwisle, Alexander, and Olson
Table A1. Continued

Variable Mean (SD) Description

Student Achievement
CAT reading score, Fall of Grade 1 279.69 (40.00) California Achievement Test, Form C, Reading Comprehension, scale scores,
Beginning of Grade 1 (Fall 1982)
Early Schooling

CAT Reading Score, Spring of Year 5 486.85 (73.14) California Achievement Test, Form C, Reading Comprehension, scale scores,
End of Year 5 (Spring 1987)

Reading Mark, Grade 1, Quarter 4 2.25 (.93) Reading of students in Grade 1, quarter 4 (spring 1983):
1 = unsatisfactory, 2 = satisfactory, 3 = good, 4 = excellent

Retention, Years 1–5 .40 (.49) Any retention in first 5 years of school; from school records, coded:
0 = not retained
1 = Retained
Student Classroom Behavior Ratings
Beginning Year 2 (Quarter 1)
Conduct 1.74 (.44) Classroom Behavior ratings from report cards, each category coded
Effort 1.74 (.44) 1 = needs improvement, 2 = unsatisfactory
Attention 1.66 (.47)
Class participation 1.78 (.41) A Classroom Behavior rating is created for each quarter of Years 2 and 4 by taking
Completes assignments 1.74 (.44) the mean of at least 4 of the 6 possible ratings (e.g., conduct, effort) recorded
Works independently 1.62 (.49) quarterly.

Beginning Year 4 (Quarter 1)


Conduct 1.72 (.45)
Effort 1.66 (.48)
Attention 1.61 (.49)
Class participation 1.73 (.45)
Completes assignments 1.75 (.43)
Works independently 1.64 (.48)

Classroom Behavior Composite, Year 2 1.73 (.30) A Classroom Behavior Composite rating for each year was constructed by taking
the mean of the available quarterly ratings.
Classroom Behavior Composite, Year 4 1.70 (.30) (alpha reliability is .92 for both years)
133
134 Entwisle, Alexander, and Olson

Table A2. Attrition Analysis: Characteristics of Original BSS Sample and OLS Regression
Sample

Original Sample Regression Sample


(N = 790) (N = 403)

Characteristic Mean SD Mean SD


Proportion Female .51 .50 .54 .50
Proportion African American .55 .50 .62 .49
Family SES Index -.04 .80 -.13 .74
Mother’s Years of Education 11.7 2.55 11.6 2.50
Proportion on Meal Subsidy .66 .47 .68 .47
Proportion Two-Parent Family .56 .50 .55 .50
Reading CAT, Beginning Grade 1 281 40.8 280 40.0
Math CAT, Beginning Grade 1 292 31.9 293 31.0
Reading Mark, Beginning Grade 1 1.88 .71 1.85 .70
Proportion Retained in Grade, Yrs 1-5 .37 .48 .40 .49

REFERENCES Observational Study of Kindergarten in Three


Social Settings.” Unpublished Ph.D. disserta-
Alexander, Karl L., and Doris R. Entwisle. 1988. tion, Johns Hopkins University, Baltimore.
“Achievement in the First Two Years of School: Bianchi, Susan M. 1984. “Children’s Progress
Patterns and Processes.” Serial No. 218. Through School: A Research Note.” Sociology
Monographs of the Society for Research in Child of Education 57:184–92.
Development 53(2). Bossert, Steven T. 1981. “Understanding Sex
Alexander, Karl L., Doris R. Entwisle, and Susan L. Differences in Children’s Classroom Exper-
Dauber. 2003. On the Success of Failure: A iences.” Elementary School Journal 81:255–
Reassessment of the Effects of Retention in the 66.
Primary Grades (2nd ed.). New York: Bourdieu, Pierre, and Jean-Claude Passeron. 1977.
Cambridge University Press. Reproduction in Education, Society, Culture.
Alexander, Karl L., Doris R. Entwisle, and Maxine S. Beverly Hills, CA: Sage.
Thompson. 1987. “School Performance, Brooks-Gunn, Jeanne, Greg J. Duncan, and Nancy
Status Relations and the Structure of Maritato. 1997. “Poor Families, Poor
Sentiment: Bringing the Teacher Back In.” Outcomes: The Well-Being of Children and
American Sociological Review 52:665–82. Youth.” Pp. 1–17 in The Consequences of
Baker, David P., and David L. Stevenson. 1986. Growing up Poor, edited by Greg J. Duncan
“Mothers’ Strategies for Children’s School and Jeanne Brooks-Gunn. New York: Russell
Achievement: Managing the Transition to Sage Foundation.
High School.” Sociology of Education Brophy, Jere E. 1985. “Interactions of Male and
59:156–66. Female Students with Male and Female
Barnett, Rosalind C. 1979, March. “Parent Child- Teachers.” Pp. 115–42 in Gender Influences in
Rearing Attitudes: Today and Yesterday.” Classroom Interaction, edited by Louise C.
Paper presented at the meeting of the Society Wilkinson and Cora B. Marrett. New York:
for Research in Child Development, San Academic Press.
Francisco. Brophy, Jere E., and Thomas Good. 1970.
Bellisimo, Yolanda, Colin H. Sacks, and John R. “Teachers’ Communication of Differential
Mergendoller. 1995. “Changes Over Time in Expectations for Children’s Classroom
Kindergarten Holding Out: Parents and School Performance.” Journal of Educational
Contexts.” Early Childhood Research Quarterly Psychology 61:365–74.
10:105–222. Burbridge, Lynn C. 1991a. “Academic
Berkeley, Muriel V. 1978. “Inside Kindergarten: An Achievement: The Role of Race, Sex, and
Early Schooling 135

Socioeconomic Status.” Research Report Attainment of Young Children.” Sociology of


(Wellesley, MA: Wellesley College Center for Education 54:34–50.
Research on Women), 10(1). ——1982. Early Schooling: Cognitive and Affective
——. 1991b. “The Interaction of Race, Gender, Outcomes. Baltimore, MD: Johns Hopkins
and Socioeconomic Status in Education University Press.
Outcomes.” Working paper No. 246. Epstein, Joyce L., and James M. McPartland. 1977.
Wellesley, MA: Center for Research on The Quality of School Life Scale. Boston:
Women, Wellesley College. Houghton Mifflin.
California Achievement Test. 1979. California Farkas, George, Robert Grobe, David Sheehan, and
Achievement Tests: Technical Bulletin 1, Forms C Yuan Shuan. 1990. “Cultural Resources and
and D, Levels 10–19. Monterey, CA: CTB/ School Success: Gender, Ethnicity and Poverty
McGraw Hill. Groups Within an Urban School District.”
Cameron, Mary Bridget, and Barry J. Wilson. 1990. American Sociological Review 55:46–61.
“The Effects of Chronological Age, Gender and Featherman, David L., and Gillian Stevens. 1982. A
Delay of Entry on Academic Achievements and Revised Socioeconomic Index of Occupational
Retention: Implications for Academic Status: Application in Analysis of Sex Differences
Redshirting.” Psychology in the Schools in Attainment. Pp. 141–82 in Social Structure
27:260–63. and Behavior: Essays in Honor of William Hamilton
Darling-Hammond, Linda, and Beverly Falk. 1995. Sewell, edited by Robert M. Hauser, D.
“Using Standards and Assessments to Support Mechanic, Archibald O. Haller, and T. S. Hauser.
Student Learning: Alternatives to Grade New York: Academic Press.
Retention.” In Report to the Chancellor’s Gates, Arthur J. 1961. “Sex Differences in Reading
Committee on Grade Transition Standards. Ability.” Elementary School Journal 61:431–34.
National Center for Restructuring Education, Grant, Linda, and Suet-ling Pong. 2003.
Schools and Teaching. New York: Teachers “Perspectives on Critical Issues.” Sociology of
College, Columbia University. Education 76:368.
Dee, Thomas S. 2005. “Teachers and the Gender Hall, M., and R. A. Keith. 1964. “Sex Role
Gap in Student Achievement.” Working paper Preference Among Children of Upper and
11660. Cambridge, MA: National Bureau of Lower Social Classes.” Journal of Social
Economic Research. Psychology 62:101–10.
Ensminger, Margaret E., and Anita L. Slusarcick. Hart, Betty, and Todd R. Risley. 1995. Meaningful
1992. “Paths to High School Graduation or Differences in the Everyday Experience of Young
Dropout: A Longitudinal Study of a First-Grade American Children. Baltimore, MD: Paul H.
Cohort.” Sociology of Education 65:95–113. Brookes.
Entwisle, Doris R., and Karl L. Alexander. 1996. Hauser, Robert M. 2001. “Should We End Social
“Family Type and Children’s Growth in Promotion? Truth and Consequences.” Pp.
Reading and Math Over the Primary 151–78 in Raising Standards or Raising Barriers?
Grades.” Journal of Marriage and the Family Inequality and High Stakes Testing in Public
58:341–55. Education, edited by Gary Orfield and Mindy
Entwisle, Doris R., Karl L. Alexander, and Linda S. L. Kornhaber. New York: Century Foundation.
Olson. 1997. Children, Schools and Inequality. Hauser, Robert M., Devah I. Pager, and Solon J.
Boulder, CO: Westview Press. Simmons. 2000, October 18–19. “Race-
—– 2005. First Grade and Educational Attainment Ethnicity, Social Background, and Grade
by Age 22: A New Story. American Journal of Retention.” Paper presented at the National
Sociology 110:1458–1502. Invitational Conference “Can Unlike Children
——. 2006. “Educational Tracking Within and Learn Together? Grade Retention, Tracking,
Between Schools: From First Grade Through and Grouping,” Alexandria, VA.
Middle School and Beyond.” Pp. 173–97 in Haveman, Robert H., and Barbara L. Wolfe. 1994.
Developmental Contexts in Middle Childhood, Succeeding Generations: On the Effects of
edited by Aletha C. Huston and Marika N. Investments in Children. New York: Russell Sage
Ripke. New York: Cambridge University Press. Foundation.
Entwisle, Doris R., and David P. Baker. 1983. Hess, Robert D. 1970. “Social Class and Ethnic
“Gender and Young Children’s Expectations Influences on Socialization.” Pp. 457–558 in
for Performance in Arithmetic.” Developmental Carmichael’s Manual of Child Psychology, edited
Psychology 19:200–09. by Paul H. Mussen. New York: John Wiley &
Entwisle, Doris R., and Leslie A. Hayduk. 1981. Sons.
“Academic Expectations and the School Hess, Robert D., and Virginia C. Shipman. 1965.
136 Entwisle, Alexander, and Olson

“Early Experience and the Socialization of Mayer, Susan E. 1997. What Money Can’t Buy:
Cognitive Modes in Children.” Child Family Income and Children’s Life Chances.
Development 36:869–88. Cambridge, MA: Harvard University Press.
Hinshaw, Stephen P. 1992. “Externalizing Behavior McDill, Edward L., Gary Natriello, and Aaron M.
Problems and Academic Underachievement in Pallas. 1986. “A Population at Risk: Potential
Childhood and Adolescence: Causal Relation- Consequences of Tougher School Standards
ships and Underlying Mechanisms.” Psycholog- for Student Dropouts.” American Journal of
ical Bulletin 111:127–55. Education 94:135–81.
Huston, Aletha C. 1983. “Sex-Typing.” Pp. Mickelson, Roslyn A. 2003. “Gender, Bourdieu, and
387–467 in Handbook of Child Psychology, Vol. the Anomaly of Women’s Achievement
4, edited by Paul H. Mussen. New York: John Redux.” Sociology of Education 76:373–75.
Wiley & Sons. Mikulecky, Larry. 1996. Family Literacy: Parent and
Jacob, Brian A. 2002. Where the Boys Aren’t: Non- Child Interactions. Washington, DC: U.S.
Cognitive Skills, Returns to School, and the Department of Education.
Gender Gap in Higher Education. Working Mishel, Lawrence, Jared Bernstein, and Sylvia
paper No. 8964. Cambridge, MA: National Allegretto. 2005. The State of Working America
Bureau of Economic Research. 2004–2005. Ithaca, NY: Cornell University
Kerckhoff, Alan C. 1993. Diverging Pathways: Social Press.
Structure and Career Deflections. New York: Pellegrini, Anthony D., and Catherine M. Bohn.
Cambridge University Press. 2005. “The Role of Recess in Children’s
Kimmel, Michael. 2000, Spring–Summer. What Cognitive Performance and School
About the Boys? Research Report. Wellesley, Adjustment.” Educational Researcher 34(1):
MA: Wellesley Center for Research on Women. 13–19.
Lackey, Pat N. 1989. “Adults’ Attitudes About Pellegrini, Anthony D., K. Kato, P. Blatchford, and
Assignment of Household Chores to Male and E. Baines. 2002. “A Short-Term Longitudinal
Female Children.” Sex Roles 20:271–81. Study of Children’s Playground Games Across
Lamont, Michele, and Annette Lareau. 1988. the First Year of School: Implications for Social
“Cultural Capital: Allusions, Gaps, and Competence and Adjustment to School.”
Glissandos in Recent Theoretical Develop- American Educational Research Journal
ments.” Sociological Theory 6:153–68. 39:991–1015.
Lareau, Annette. 1987. “Social Class Differences in Pope, B. 1953. “Socio-Economic Contrasts in
Family-School Relationships: The Importance Children’s Peer Culture Prestige Values.”
of Cultural Capital.” Sociology of Education Genetic Psychology Monographs 48:157–220.
60:73–85. Porche, Michelle V., Stephanie J. Ross, and
——. 1992. “Gender Differences in Parent Catherine E. Snow. 2004. “From Preschool to
Involvement in Schooling.” Pp. 207–24 in Middle-School: The Role of Masculinity in
Education and Gender Equality, edited by Julia Low-Income Urban Adolescent Boys’ Literacy
Wrigley. London: Falmer Press. Skills and Academic Achievement.” Pp.
——. 2002. “Invisible Inequality: Social Class and 338–60 in Adolescent Boys: Exploring Diverse
Child Rearing in Black Families and White Cultures in Boyhood, edited by Niobe Way and
Families.” American Sociological Review Judy Y. Chu. New York: New York University
67:747–76. Press.
Lavy, Victor. 2004. Do Gender Stereotypes Reduce Powell, Walter W., and Kaisa Snellman. 2004. “The
Girls’ Human Capital Outcomes? Evidence from Knowledge Economy.” Annual Review of
a National Experiment. Working paper No. Sociology 30:199–220.
10678. Cambridge, MA: National Bureau of Powell, William. 1967. “The Nature of Individual
Economic Research. Differences.” Pp. 1–17 in Perspectives in
Leaper, Campbell, Kristin J. Anderson, and Paul Reading: Organizing for Individual Differences.
Sanders. 1998. “Moderators of Gender Effects Newark, DE: International Reading
on Parents’ Talk to Their Children: A Meta- Association.
Analysis.” Developmental Psychology 34:3–27. Prawat, Richard, and Robert Jarvis. 1980. “Gender
Maccoby, Eleanor, and Carol Jacklin. 1974. The Differences as a Factor in Teachers’
Psychology of Sex Differences. Stanford, CA: Perceptions of Students.” Journal of Edu-
Stanford University Press. cational Psychology 72:743–49.
Martin, Karin A. 1998. “Becoming a Gendered Reardon, Sean. 1996, April. “Eighth-Grade
Body: Practices of Preschools.” American Minimum Competency Testing and Early High
Sociological Review 63:494–511. School Dropout Patterns.” Paper presented at
Early Schooling 137

the meeting of the American Educational ——. 2002. The Big Payoff: Educational Attainment
Research Association, New York. and Synthetic Estimates of Work-Life Earnings,
Rehberg, Richard A., and Evelyn R. Rosenthal. Current Population Reports (Series P-23, No.
1978. Class and Merit in the American High 210). Washington, DC: U.S. Government
School. White Plains, NY: Longman. Printing Office.
Reynolds, Arthur J., Suh-Ruu Ou, and James W. ——. 2003. Statistical Abstract of the United States:
Topitzes. 2004. “Paths of Effects of Early 2003. Washington, DC: Author.
Childhood Intervention on Educational ——. 2004–05. Statistical Abstract of the United
Attainment and Delinquency: A Confirmatory States: 2004–2005. Washington, DC: Author.
Analysis of the Chicago Child-Parent Centers.” U.S. Department of Education. 2001. The Condition
Child Development 75:1299–328. of Education 2001. NCES 2001-072.
Riordan, Cornelius. 2003. “Failing in School? Yes; Washington, DC: U.S. Government Printing
Victims of War? No.” Sociology of Education Office.
76:369–72. ——. 2002. Digest of Education Statistics, 2002.
Robinson, Helen M. 1955. “Factors Which Affect Washington, DC: Author.
Success in Reading.” Elementary School Journal ——. 2004. The Condition of Education, 2004. NCES
55:266–69. 2004077. Washington, DC: U.S. Government
Rogoff, Barbara. 2003. The Cultural Nature of Printing Office.
Human Development. New York: Oxford U.S. Department of Health and Human Services.
University Press. 1997. Trends in the Well-Being of American
Royce, Jacqueline M., Richard B. Darlington, and
Children and Youth. Washington, DC: Author.
Harry W. Murray. 1983. “Pooled Analyses:
U.S. Department of Labor. 1984. “Table 18:
Findings Across Studies.” Pp. 411–59 in As the
Employed Civilians by Detailed Occupation,
Twig Is Bent: Lasting Effects of Preschool
Sex, Race, and Hispanic Origin.” Washington,
Programs, edited by the Consortium for
DC: U.S. Department of Labor, Bureau of
Longitudinal Studies. Hillsdale, NJ: Lawrence
Labor Statistics.
Erlbaum.
Useem, Elizabeth L. 1991. “Student Selection into
Sadker, Myra, and David Sadker. 1994. Failing at
Course Sequences in Mathematics: The
Fairness: How America’s Schools Cheat Girls.
Impact of Parental Involvement and School
New York: Charles Scribner & Sons.
Policies.” Journal of Research on Adolescence
Schwartz, Wendy, and Katherine Hanson. 1992.
Equal Mathematics Education for Female 1:231–50.
Students. Technical Report No. ED 344977. West, Jerry, Kristin Denton, and Lizabeth M.
ERIC Clearinghouse on Urban Education. Reaney. 2001. The Kindergarten Year: Findings
Stanchfield, Jo M. 1973. Sex Differences in Learning from the Early Childhood Longitudinal Study.
to Read. Bloomington, IN: Phi Delta Kappa Technical Report No. NCES 2001-023.
Educational Foundation. Washington, DC: National Center for
Stroud, J. B. and E. F. Lindquist. 1942. “Sex Education Statistics.
Differences in Achievement in the Elementary White, Lynn K., and David B. Brinkerhoff. 1981.
and Secondary Schools.” Journal of Educational “The Sexual Division of Labor: Evidence from
Psychology 33:657–67. Childhood.” Social Forces 60(1):170–81.
Tach, Laura Marie, and George Farkas. 2006. Wilson, Frank T., Agnes Burke, and C. E. Fleming.
“Learning-Related Behaviors, Cognitive Skills, 1939. “Sex Differences in Beginning Reading
and Ability Grouping When School Begins.” in a Progressive School.” Journal of Educational
Social Science Research 35:1048–79. Research 32:570–82.
U.S. Census Bureau. 1983. Census of Population: Zemsky, Robert, Daniel Shapiro, Maria Iannozzi,
1980: Vol. 1. Characteristics of the Population. Deter Capelli, and Thomas Bailey. 1998. The
Washington, DC: U.S. Government Printing Transition from Initial Education to Working Life
Office. in the United States of America. Washington,
——. 2001. Statistical Abstract of the United States: DC: U.S. Department of Education, Office of
2001. Washington, DC: Author. Educational Research and Improvement.
138 Entwisle, Alexander, and Olson

Doris R. Entwisle, Ph.D., is Research Professor of Sociology, Johns Hopkins University. Her main area
of interest is the sociology of human development over the life course, with an emphasis on issues
of inequality. Her current research concerns the ways in which members of the Beginning School
Study (BSS) panel make the transition to adulthood in the third decade of life. Her book (with Karl
L. Alexander and Linda S. Olson), Children, Schools, and Inequality (Westview-Perseus, 1977)
explores the ways in which social structure shapes the paths that children take through elementary
school.

Karl L. Alexander, Ph.D., is John Dewey Professor of Sociology and Chair of the Department of
Sociology, Johns Hopkins University. Since 1982, he and Doris Entwisle have been directing the BSS,
an ongoing long-term study of youth development. His interests center on schools and problems of
educational stratification that can be addressed via organizational, social-psychological, and life-
course perspectives. With Doris Entwisle and Susan Dauber, he published a revised, updated ver-
sion of their 1994 evaluation of grade retention, On the Success of Failure (Cambridge University
Press, 2003). He is working on several projects that examine the lasting imprint of early home and
schooling experiences on the transition to adulthood.

Linda S. Olson, MA, is Associate Research Scientist, Department of Sociology, Johns Hopkins
University. She has been associated with the BSS for the past 20 years. Her interests center on sta-
tistical modeling and the effects of social structure on schooling outcomes.

This research was supported by Grant 200300057 from the Spencer Foundation and a Mellon
Fellowship No. 30600638 to Doris R. Entwisle. Address correspondence to Doris R. Entwisle,
Department of Sociology, Johns Hopkins University, Merganthaler 530, 3400 North Charles Street,
Baltimore, MD 21218; e-mail: entwisle@jhu.edu.

You might also like