Family characteristics, social capital, and college attendance


Material Information

Family characteristics, social capital, and college attendance
Physical Description:
vii, 137 leaves : ill. ; 29 cm.
Smith, Mark Holland, 1954-
Publication Date:


Subjects / Keywords:
College attendance   ( lcsh )
Family -- Sociological aspects   ( lcsh )
Sociology thesis Ph. D
Dissertations, Academic -- Sociology -- UF
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1993.
Includes bibliographical references (leaves 130-136).
Statement of Responsibility:
by Mark Holland Smith.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001912265
oclc - 30301505
notis - AJY7760
System ID:

This item is only available as the following downloads:

Full Text







Dedicated to my beloved wife, Terry.
Without her support and encouragement the completion
of this dissertation would not have been possible.
Her keen perception, quiet wisdom, and selfless love have
inspired me both as a scholar and as a human being.


I wish to express my appreciation to Bo Beaulieu for his

encouragement, support, and patience as supervisor of my

graduate research assistantship and cochair of my doctoral

committee. I wish to thank the rest of my doctoral committee,

cochair Gary Lee and members Leonard Beeghley, John Henretta,

Glenn Israel, and Albert Matheny. Additionally, I wish to

thank Anne Seraphine for her assistance and good humor.




ACKNOWLEDGEMENTS .........................................iii


Introduction .............................. 1
Effects of Race and Sex on
College Attendance........................ 6
Effects of Class on College
Attendance............................... 10
Effects of College Attendance on
Subsequent Status Attainment............ 12
College Attendance and the
"Education Crisis"....................... 17
Effects of Social Interaction on
College Attendance....................... 24


Introduction............................... 28
Family Background in the Study
of College Attendance.................... 30
The Family as "Significant Other"......... 37
Human Capital and Social Capital.......... 47
Research Questions......................... 79

3 METHODOLOGY ............................... 82
Description of the Secondary Data Set..... 82
Statistical Analysis....................... 92

4 FINDINGS................................... 95
Family Background Influences on
College Attendance ...................... 105
Family Structure and Family Process.......110
School Structure and School Process.......113
Community Structure
and Community Process ................... 113

5 CONCLUSIONS ............................... 116
Discussion......... ..................... 116
Significance of the Study................. 124

REFERENCES................... ..oo ................... ...130

BIOGRAPHICAL SKETCH ...................................... 137

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Mark Holland Smith

May, 1993

Chairman: Gary Lee
Major Department: Sociology

The structural transformation of the American economy in

the last two decades, by eliminating many of the high-wage

manufacturing jobs once available to those with only a high

school education, is increasing the importance of a college

education for admittance into competition for middle-class

occupations. Despite the societal ideal of universal access

to higher education, as well as the demonstrable value of

college attendance for subsequent occupational attainment,

many people who are academically qualified nevertheless do not

go to college. This study examines social factors that may

account for attendance behavior.

Theoretically situated in the status attainment

tradition, this study develops a model of educational

attainment that focuses on the role of social interaction

variables in mediating the effects of family background

variables--family income, father's education, and mother's

education--on college attendance. Structural arrangements and

interaction patterns fostering positive relationships are

regarded as social capital, which serves as a resource for the

individual as well as to constrain and enable the individual

in the educational attainment process.

Logistic regression procedures are employed in analysis

of data from the High School and Beyond Longitudinal Study.

In order to evaluate the effects of race and sex, the social

capital model of college attendance is estimated separately

for whites, blacks and Hispanics, with statistically

significant sex interactions.

Results of these analyses indicate that parental

involvement is the most powerful predictor of college

attendance in the model, more influential than family

background variables. Family structural arrangements, such as

single-parent household and whether the mother works, are

found to exert only a modest, if any, negative impact on

college attendance. While social interaction in the family

has the greatest impact on attendance, the integration of the

individual in the community, as measured by number of moves

and by church attendance, is shown to be of some importance in

the educational attainment process.




The purpose of this study is to add to knowledge of the

effects of social interaction on college attendance. More

specifically, the aim is to ascertain whether or not

supportive interpersonal interaction in the family, school,

and community enhances the likelihood of attending college.

Following the work of James Coleman (1988a, 1988b; Coleman and

Hoffer, 1987; Coleman, Hoffer and Kilgore, 1982), such

supportive interaction will be regarded conceptually as

"social capital" and will be treated as mediating the impact

of family background variables on college attendance.

The first chapter is designed to state the research

problem in terms of the importance, both to the individual and

to society, of college attendance. Descriptive statistics of

changes in attendance rates are reviewed in conjunction with

a discussion of the transformation of the American economy, a

transformation serving to highlight the importance of college

attendance in determining subsequent status attainment. Data

pertaining to the effects of family background factors, race,

class, and gender on college attendance rates are analyzed.

This study examines the impact of patterns of student social



interaction in the family and in the community in mediating

the effects of family background on college attendance.

Patterns of College Attendance

Attending, or not attending, college is one of the

crucial decisions in the life of an individual. A watershed

event, college attendance or nonattendance decisively shapes

the subsequent lifecourse, largely determining subsequent

occupational opportunities and income. The role of college

attendance in mediating the distribution of occupations and

incomes shows signs of increasing in importance in the status

attainment process.

Educational attainment has steadily increased throughout

this century, both in high school completion and in college

attendance. Higher educational attainment is seen in American

culture as the tried and true path of upward social mobility,

an article of faith underscored by the fact that, except for

the period between 1932 through 1934, college enrollments

climbed even during the Depression years (Parker, 1971). As

President Ronald Reagan stated in a radio message in 1983,

"Education was not simply another part of American society.

It was the key that opened the golden door. Parents who never

finished high school scrimped and saved so that their children

could go to college" (Mattera, 1990:130).

One spur to growing college attendance was federal

financial work-study assistance under the National Youth


Administration (NYA) in 1935-36, for "financially-needy

students with character and ability" (Parker, 1971:31). The

biggest boost to college attendance increases was the return

of veterans from World War II and later from the Korean War.

College enrollments in the United States increased from

1,364,000 in 1939 to 8,560,000 in 1974, fueled by the post-

World War II GI Bill college benefits and dramatic increases

in college-age youth as the "baby-boom" generation reached

college age (Thomas, Alexander, and Eckland, 1979). The most

dramatic increase in college attendance came during the

tumultuous 1960s, when enrollments grew from around 3,000,000

in 1960 to 7,980,000 in 1969 (Parker, 1971). Over the same

period, the proportion of high school graduates enrolled in

college during the year following graduation increased from

45.1 percent in 1960 to 53.3 percent in 1969, rising to 59.6

percent by 1989 (U.S. Department of Education, 1991:178).

What Parker (1971) refers to as the "enrollment

explosion" of the 1960s can be attributed to the confluence of

several factors. The unprecedented national prosperity and

expanding middle class meant that more families than ever

could afford to send their children to college. Federal and

state grants, guaranteed loans and work-study programs removed

financial barriers to college attendance for low-income

students, while civil rights legislation made college more

accessible to racial minorities and women. Additionally, open

admission policies were adopted at community colleges and many


state universities, creating opportunities for college

attendance for many who in earlier decades would not have

qualified academically.

College Enrollment Patterns

Total degree-credit full-time enrollment grew to

10,473,000 in 1980 but remained flat through the 1980s, rising

only to 10,937,000 by 1988 (U.S. Bureau of the Census,

1990:152). While total enrollment stagnated during the 1980s,

enrollments relative to the size of the college-age population

continued to grow, as can be seen in the fact that the

proportion of the population 18-24 years of age declined by

13.9 percent between 1980 and 1990 (U.S. Statistical Abstract,

1990:16). The percentage of high school graduates enrolling

in college in the October following graduation rose from 50.9

percent in 1980 to 58.4 percent in 1988 (U.S. Department of

Education, 1991).

The end of the explosion in total enrollments can be

attributed to demographic factors--the last of the baby

boomers entering and completing college--as well as to

recession and the decline of federal and state grants and

scholarships for postsecondary education. The extremely rapid

expansion in enrollment levels in the 1960s led to efforts to

find ways to restrict admissions. Most major universities and

liberal-arts colleges became more selective, stiffening

entrance requirements, thus reestablishing the importance of


academic credentials (Thomas et al., 1979). The renewed

emphasis on academic performance reinforced the traditional

view of education as the meritocratic regulator of

occupational status outcomes.

Of those students enrolled in degree credit programs in

1980, about 26 percent attended private colleges; by 1980,

those attending private schools declined to 21 percent. Among

black students, those attending private institutions declined

2.5 percentage points, from 22.5 percent in 1980 to 20 percent

in 1980. White students attending private schools declined 5

percentage points, from 26 percent in 1980 to 21 percent in

1988. Among Hispanic college students, the percentage

attending private schools declined from 21 percent in 1980 to

9 percent in 1988 (U.S. Statistical Abstract, 1990:152).

In 1988, 56 percent of those attending college were

enrolled in four-year colleges and 29 percent in two-year

schools, and 15 percent were attending graduate school (U.S.

Statistical Abstract, 1990:152). We can observe a gradual

shift during the 1980s away from more expensive private

schools to public schools. Since the early 1970s, enrollment

growth in public 2-year schools has exceeded enrollments in

either public or private 4-year schools (U.S. Department of

Education, 1991:76).


Effects of Race and Sex on College Attendance

College enrollment for men reached its peak of 63.2

percent in 1968, dipped as low as 46.9 percent in 1980, and

climbed back to 57.6 percent in 1989. Enrollments for women

students climbed steadily from 47.2 percent in 1967 to 61.6

percent in 1989. Attendance rates for white high school

graduates rose from 53.1 percent in 1967 to 60.4 percent in

1989. Enrollment rates among black high school students

fluctuated considerably while rising from 42.3 percent in 1967

to 52.8 percent in 1989 (U.S. Department of Education,


Table 1-1
Date of first enrollment in postsecondary education among 1982
high school graduates who enrolled before 1986, by


Date of first enrollment in college

10/82 1983 1984 1985
Percent of those enrolled before 1986

non-Hispanic 81.6 10.4 4.7 3.3

non-Hispanic 69.8 18.8 7.4 4.0

Hispanic 73.7 15.1 7.3 3.9

Source: U.S. Department of Education, National Center for
Education Statistics, 1991.

Part of the racial differences in attendance rates for

high school graduates enrolling in college by the following


fall term can be accounted for by the fact that both blacks

and Hispanics are more likely to delay enrollment, as

indicated in Table 1-1. Among 1982 graduating high school

students enrolled before 1986, over 30 percent of blacks

delayed college entry, compared to 26 percent of Hispanics and

18 percent of whites (U.S. Department of Education, 1991:20).

Nearly three out of four adult Americans still do not

possess a college degree (Chronicle of Higher Education,

1989), though the percentage of all high school graduates

enrolled in college or who have completed 1 or more years of

college went up from 40.4 in 1960 to 57.5 in 1988. The rate

of improvement was about the same for whites and blacks, with

the percentage improving from 41 in 1960 to 58.6 in 1988 for

whites. Black high school graduates in college or who have

completed 1 or more years of college went up from 32.5 percent

in 1960 to 46.6 in 1988. Women of both races improved faster

than men, exceeding the rate for men by 1987. The figure for

white women rose from 35.6 percent in 1969 to 59.2 percent in

1988, compared to figures of 41.7 percent and 57.9 percent for

white men. Black women high school graduates in college or

with 1 or more years of college went up from 31.8 percent in

1960 to 49.6 percent in 1988, compared to the figures for

black men of 33.5 percent in 1960 and 42.8 percent in 1988

(U.S. Bureau of the Census, 1990:151).

As a percentage of total enrollment, white enrollment has

been steadily declining, from 82.6 percent in 1976 to 78.8


percent in 1988 (see Table 1-2). The corresponding figures

for blacks also declined, from 9.4 percent of total enrollment

in 1976, to 8.7 percent in 1988. The difference in enrollment

was made up by substantial increases for Hispanics, who

increased their share of enrollments from 3.5 percent in 1976

to 3.8 percent in 1988, and Asians, who enhanced their share

from 1.8 percent in 1976 to 3.8 percent in 1988 (U.S.

Department of Education, 1991:80). The decline in white

enrollments as a percentage of total enrollment reflects the

increase of the proportion of racial minorities among college-

age cohorts. The smaller decline in black enrollment is

concentrated among black males (U.S. Department of Education,


Table 1-2
Percent of total enrollment, by race/ethnicity: Selected years

Year White, Black, American
non-Hisp. non-Hisp. Hispanic Asian Indian

1976 82.6 9.4 3.5 1.8 0.7
1980 81.4 9.2 3.9 2.4 0.7
1984 80.2 8.8 4.4 3.2 0.7
1986 79.3 8.7 4.9 3.6 0.7
1988 78.8 8.7 5.2 3.8 0.7

Source: The Condition of Education, U.S. Department of
Education, National Center for Education Statistics, 1991.

Table 1-3
Percent enrolled in college in October following high school
graduation, by sex. and race/ethnicity: Selected vears,1967-

Sex Race

Year Total Male Female White Black

1967 51.9 57.6 47.2 53.1 42.3
1970 51.8 55.2 48.5 52.2 48.3
1975 50.7 52.6 48.9 51.2 45.6
1980 49.4 46.9 51.7 49.9 42.6
1985 57.7 58.6 56.9 59.4 42.3
1989 59.6 57.6 61.6 60.4 52.8

Source: The Condition of Education, U.S. Department of
Education, National Center for Education Statistics, 1991

In 1980, female college enrollment overtook that of

males, and with some yearly fluctuations, exceeded male

enrollment substantially by 1989, in an historic reversal of

the traditional female disadvantage in college enrollments

(U.S. Department of Education, 1991:103; see Table 1-3).

Women also display higher rates of attainment of the bachelor

degree. By February, 1986, 18.4 percent of male high school

seniors in 1980 had completed bachelor's degrees, compared to

19.2 percent of females. The advantage is more pronounced for

two-year degrees, with 14.1 percent of female high school

seniors in 1980 completing degrees by 1986, compared to 10.8

percent of males (U.S. Statistical Abstract, 1990:164).

During the 1960s and 1970s, college completion rates for women

ran about 6 percentage points behind men, but by 1990 had


pulled about even (National Center for Education Statistics,

1991: 34). Probably a part of the explanation for the

attainment increases for women is the fact that women

consistently perform better than men, on average, in terms of

high school academic performance (Thomas, et al., 1979).

Effects of Class on College Attendance

Substantial sex differences remain in college major, as

well as in subsequent occupational status and income, but the

statistical evidence suggests that sex plays a very diminished

role in college attendance itself. By contrast, racial

differences in enrollments persist. While the rate increase

in black educational attainment has kept pace, roughly, with

increases in white attainment over the last 30 years, the gap

between average black and white attainment has not narrowed

appreciably. Among students who were high school seniors in

1980, 20.8 percent of whites had completed a bachelors degree

by 1986, compared to 10.1 percent of blacks, 6.8 percent of

Hispanics, 29.7 percent of Asians, and 9.2 percent of American

Indians (U.S. Bureau of the Census, 1990:164).

On the face of it, a plausible explanation for the

persistence of the gap between white and black college

attendance is that--despite civil rights laws and affirmative

action programs giving preference to racial minorities in

admissions and for scholarships--blacks are still subject to

discrimination in college admissions. However, a more


compelling explanation presents itself when the relationship

between race and college attendance is examined with controls

for socioeconomic status.

Analysis of the National Longitudinal Survey (NLS) of the

Class of 1972 by Thomas et al. (1979) revealed that

socioeconomic class accounts for about a third of the total

variance in college attendance,

in contrast to the almost negligible effects of race and
sex. With scholastic aptitude and family origin
controlled, blacks and women experience little direct
disadvantage in terms of the likelihood of attending
college and far less than the disadvantage experienced by
low-status students. In fact, when compared with whites
of comparable status origins and scholastic aptitude,
blacks actually are somewhat more likely to attend
college. Nevertheless, academic credentials were the
major determinants of college access for all groups. (pp.

The lower average socioeconomic status of blacks, combined

with lower measures of scholastic aptitude, accounts for much,

if not all, of the difference in college attendance rates

between blacks and whites. This is not to deny that blacks

experience discrimination in other spheres of economic and

social life, but to suggest that, with respect to access to a

college education, "black aspirations are stymied not because

of their race per se, but because of a lacking of [financial]

ability to realize ambitions" (Thornton, 1977:40).

Undoubtedly, socioeconomic disadvantage among blacks can

be traced to historic patterns of racial discrimination,

though such discrimination is not apparent in college

admissions. In fact, the evidence suggests a small


"affirmative action" effect of black advantage in college


Effects of College Attendance on
Subsequent Status Attainment

Ever since Blau and Duncan's (1967) ground-breaking study

of the American occupational structure, the principle has been

firmly established that educational attainment is a critical

factor in determining occupational status and income. Level

of educational attainment exerts a powerful effect on whether

or not one has a job, the character of the job, and earnings


The first and most fundamental plateau of educational

achievement that students must scale in order to participate

in the public economy is high school graduation. In a study

analyzing census and other large national data sets, Braun

(1991) found that the most powerful predictor of family income

is the percent who have competed a high school education. A

high school education accounts for more than ten times the

variance in family incomes than does race (Braun, 1991:230).

The structural shift toward a service economy in the 1980s has

reinforced the role of the high school diploma as the minimum

criterion for economic participation. As Table 1-4 indicates,

employment rates for 25- to 34-year-old men with only 9-11

years of school have declined from 87.9 percent in 1971 to

75.9 percent in 1990. Among females 25 to 34 years old with

9 to 11 years of school (see Table 1-4), the percent employed


increased from 35.2 percent in 1971 to 44.3 percent in 1990.

This small increase is likely due to secular increases in

female labor force participation. Completion of high school

increased the percentage of women employed only to 43.1 in

1971 but to 67.5 by 1990.

In the 1950s and 1960s, a high school diploma was

sufficient to obtain a high-wage blue-collar manufacturing

job. As manufacturing corporations realized the comparative

advantage of moving plants to developing countries, these jobs

have become few and far between, helping to account for what

Harrington (1984) calls the "new poverty" among young workers,

especially men. Between 1978 and 1983, the percentage of poor

males increased at three times the rate for females, with the

sharpest rate increases of poverty among white males (Braun,

1991:155). While employment among females with high school

educations rose dramatically between 1971 and 1990 (see Table

1-5), employment among males in that category declined from a

high of 93.7 percent in 1972 to a low of 78.6 percent during

the recession of 1983, and then rebounded somewhat to 88.6

percent in 1990. The evidence suggests that this improvement

is due less to the renewed efficacy of a high school diploma

in getting a good job and more to compositional changes in an

economy that is generating more low-wage service sector jobs

and fewer high-wage manufacturing jobs (Braun, 1991; Teixeira

and Swaim, 1991).

Table 1-4
Employment rate of 25- to 34-year-old males, by years of
schooling completed: Selected years. 1971-1990

9-11 years 12 years 4 or more years
Year of school of school of college

1971 87.9 93.6 92.5
1975 78.1 88.4 93.5
1980 77.7 87.0 93.4
1983 69.3 78.6 91.1
1986 73.3 86.2 93.7
1989 77.6 87.8 91.1
1990 75.9 88.6 93.1

Source: The Condition of Education, U.S. Department of
Education, National Center for Education Statistics, 1991

Table 1-5
Employment rate of 25- to 34-year-old females, by years of
schooling completed: Selected years. 1971-1990

9-11 years 12 years 4 or more years
Year of school of school of college

1971 35.2 43.1 56.9
1975 34.5 48.0 66.4
1980 45.6 59.5 75.5
1983 37.1 58.8 79.2
1986 44.1 63.8 80.3
1989 43.0 66.9 82.1
1990 44.3 67.5 83.2

Source: The Condition of Education, U.S. Department of
Education, National Center for Education Statistics, 1991

Increasingly, the ticket to a decent job in the American

economy is inscribed "college graduate." Employment rates for

males with four or more years of college have consistently

exceeded 92 percent since 1971, dropping below that figure


only slightly during the recession years of 1982-84 (see

Tables 1-4 and 1-5). Employment rates among females with 4 or

more years of college climbed sharply from 56.9 percent in

1971 to 83.2 percent in 1990. It is well to note, however,

that 92 percent and 83 percent employment rates for men and

women, respectively, are not as high as one might expect for

college graduates.

Table 1-6
Ratio of annual earnings of male wage and salary workers 25 to
34 years old with 9-11 and 16 or more years of school to those
with 12 years of school, by race/ethnicity: Selected years.

9-11 years of school 16 or more years of school

Year White Black White Black

1975 0.81 0.57 1.18 1.29
1978 0.78 0.74 1.13 1.48
1980 0.80 0.75 1.18 1.33
1983 0.75 0.65 1.34 1.50
1985 0.73 0.70 1.45 1.77
1988 0.70 0.56 1.41 1.37
1989 0.73 0.60 1.45 1.42

Source: The Condition of Education, U.S. Department of
Education, National Center for Educational Statistics, 1991

There are indications that the recession of the late

1980s and early 1990s has been unusually white-collar in the

composition of the unemployed (Braun, 1991), suggesting that

the probability of occupational payoff for college attendance

may be eroding. Examination of earnings patterns reveals even


more clearly the advantage accruing to college attendance and

graduation. During the course of the 1980s (see Tables 1-6

and 1-7), the ratio of annual earnings of those wage and

salary workers with 16 or more years of school to those with

12 years of school increased substantially over the ratio that

was prevalent in the 1970s. In the last five years of the

decade, the earnings advantage of white males was 45 percent,

compared with 54 percent for black males. The earnings

advantage for women for the last five years of the 1980s was

even more substantial--75 percent for white females and 92

percent for black females (U.S. Department of Education,


Table 1-7
Ratio of annual earnings of female wage and salary workers 25
to 34 years old with 9-11 and 16 or more years of school to
those with 12 years of school, by race/ethnicity: Selected
years. 1975-1989

9-11 years of school 16 or more years of school
Year White Black White Black

1975 0.65 0.60 1.74 1.70
1978 0.55 0.48 1.58 1.38
1980 0.63 0.73 1.54 1.65
1983 0.66 0.65 1.69 1.59
1985 0.62 0.66 1.64 1.76
1988 0.53 0.62 1.78 1.93
1989 0.66 0.50 1.89 2.05

Source: The Condition of Education, U.S. Department of
Education, Center for Educational Statistics, 1991


College Attendance and the "Education Crisis"

The structural transformation of the American economy

over the course of the 1970s and 1980s is having the apparent

cumulative effect of highlighting the effect of college

attendance on subsequent status attainment. Opportunities for

secure, high-wage employment are shrinking for those without

college credentials. "Today a college degree is not

necessarily a ticket to rapid social advancement, but without

it one does not stand a chance of escaping the erosion of

living standards," in the words of Mattera (1990). In 1990 on

PBS' MacNeil Lehrer News Hour, financial analyst Paul Solomon

described the emergence of the "upstairs economy" of well-

paying business, technical, managerial and professional

occupations and the "downstairs economy" of low-paying and

low-skill manual labor and service occupations. The critical

dividing line between the upstairs and downstairs economies is

college attendance. According to the William T. Grant

Foundation Commission on Work, Family and Citizenship

(1988:1), some 20 million youth 16-24 years of age are not

likely to attend college. Their report, "The Forgotten Half,"


This nation may face a future divided not along lines of
race or geography, but rather of education. A highly
competitive technological economy can offer prosperity to
those with advanced skills, while the trend for those with
less education is to scramble for unsteady, part-time, low-
paying jobs.


The view expressed above concentrates on the supply-side

of what is widely referred to as the "education crisis."

Educational attainment, as measured by the number of years in

college, has increased dramatically in the last three decades.

Between 1950 and 1988, the percentage of the total population

completing high school increased from 35 percent to over 75

percent, with the proportion graduating from college going up

from 6 percent to 20 percent (Teixeira and Swaim, 1991:24).

However, the rate of increase slowed during the 1980s as

several factors converged to cause concern about the

educational attainment of American workers.

One of these factors generating concern about an

education crisis is the demographic fact that an increasing

proportion of the labor force will be composed of blacks,

Hispanics, and immigrants, groups less likely to have the

educational credentials needed in the labor market of the

1990s. Inner city minorities and urban and rural poor are

more likely to drop out of high school. Additionally, low

educational attainment contributes to problems for workers

displaced by de-industrialization, as such individuals

experience longer periods of unemployment and larger earning

declines when rehired than do laid-off workers with higher

educational attainment (Teixeira and Swaim, 1991:25).

The crux of the concern about a crisis in education lies

in the idea of the "new economy." The new economy is the

postindustrial economy characterized by advanced technology,


information processing, an expanding service sector, and

fierce international competition. Rising occupational skill

requirements, combined with flagging attainment levels, poor

performance on standardized achievement tests, and unfavorable

comparisons on academic performance measures with students in

other countries in tests of math and science knowledge, lead

to an understanding of the education crisis as a crisis in the

supply of high-skill workers in the new economy.

Part of the impetus and support for the supply-side

education crisis thesis came from the Workforce 2000 report

(Johnston and Packer, 1987), which projected skill-level

requirements for the American economy through the year 2000.

The major finding of the report was that the average skill

levels of the fastest-growing occupations are higher than the

required skill levels of the slow-growing or declining

occupational categories. Proponents of the strong version of

the new economy thesis point to projected growth in high-skill

occupations in the service sector such as doctors, lawyers,

technicians, programmers, engineers, and paraprofessionals.

The implication of the strong version of the new economy

thesis is the view that a substantial, if not major, part of

the problem with the performance of the economy is a failure

of the supply of adequately educated workers.

More recent research, along with a more careful reading

of the Workforce 2000 report, suggests that a weaker version

of the new economy may be more consistent with the facts.


First of all, the projected change in the overall skill level

from 1984 to 2000 is very modest, according to Global 2000

report. Secondly, the occupations that contribute the most to

total employment growth are low-skill service occupations such

as cooks, waiters, household workers, janitors and security

guards (Teixeira and Swaim, 1991:21). Consequently, those who

say America is becoming a nation of "hamburger flippers" and

those who speak glowingly about the emergence of the high-

skill "information workers" are both right, or partly right.

The new economy is likely to be characterized by a fast-

growing high-skill "upstairs economy" and an even faster-

growing "downstairs economy" of low-skill, low-wage jobs.

Educational supply-side problems unquestionably do make

a contribution to the dismal performance of the economy. For

example, fewer Ph.D.s in science and engineering were awarded

in 1985 than in 1970, and we know that a much smaller

proportion of American undergraduates receive degrees in these

fields than is the case for America's principal international

economic competitors, the Japanese and Germans (Braun,

1991:143). A better-educated work force contributes to

productivity and competitiveness.

The exclusive focus on the supply side, however, tends to

obscure the arguably more severe problems with the demand side

of the equation--weak labor-market demand for the middle

ground of high-wage production occupations and the emergence

of a radical split in the service sector distribution of

occupations into low-skill, low-wage and high-skill, high-wage

occupations. Recent studies of the relationship between

educational attainment and economic development in rural areas

found that increased educational attainment by itself does not

create jobs, and the absence of jobs results in the out-

migration of the better-educated (Teixeira and Swaim, 1991;

McGranahan and Ghelfi, 1991).

Declines in real wages and salaries suggest that the

designation "high-wage" is a relative one. In nominal terms,

wages and salaries have steadily increased, over the last

several decades. When adjusting for inflation, however, real

wages and salaries have declined since 1973 so that, on

average, an American worker earns about the same as a worker

in 1961 (Braun, 1991:159). Between 1970 and 1980, average

real wages and salary levels declined in nearly every job

classification among men, with the exception of teachers and

doctors among professional occupations, as well as a couple of

blue-collar occupational categories (Braun, 1991:161).

Between 1973 and 1987, median male income declined $2,851, in

constant 1987 dollars, from $20,603 to $17,752, a decline of

14 percent (Braun, 1991:161).

The decline in real wages and salaries for the majority

of workers is taking place against a backdrop of increasing

income inequality. Much has been made of the fact that

national per capital income rose 16 percent between 1982 and

1987 (Braun, 1991:158), rising 23 percent between 1977 and


1989 (Krugman, 1992:54). However, the gross average of per

capital income obscures the fact that 70 percent of the total

rise in income between 1977 and 1989 was concentrated among

the top 1 percent of the population, 90 percent going to the

top 5 percent of the population (Krugman, 1992:54). The top

1 percent of families increased their after-tax income 102

percent between 1977 and 1989, while the bottom 60 percent of

families lost ground in terms of real income (U.S. News and

World Report, March 23, 1992).

The growing income inequality in the United States is not

due to gender discrimination or to racial discrimination, nor

can it be explained by other demographic factors, but instead

reflects the growing class division in American society. In

an analysis of increasing income inequality, Harrison and

Bluestone found:

The baby boom and the growth in the number of female
workers had no significant impact whatsoever on the increase
in inequality of wages. Indeed, men's and women's wages
actually converged slightly in this period -- owing more to
declines in the average wage of males than to increases in the
wages of women. The wages of white women and women of color
are now almost indistinguishable. Put another way, all of the
increase in inequality since 1975 must have occurred within
age, race, and sex groups, not among them. Inequality is
growing among whites as well as nonwhites, among the old as
well as the young, and among women as well as men (1988:120).

Under the economic conditions that are likely to prevail

in the 1990s--continued loss of high-wage manufacturing jobs

as industry moves to low-wage developing countries,

proliferation of low-wage service-sector jobs, and declining

capacity of government to provide financial assistance where


needed--a college education will be increasingly difficult to

obtain for the working classes and lower middle classes

without taking on huge debt loads. During a decade when real

incomes remained stagnant or declined, real costs for college

continued to climb. Total costs to attend public universities

went up one third while costs at private colleges rose about

50 percent during the 1980s (Mattera, 1990:131).

Declining real incomes and increasing education costs

coincided with declining federal governmental financial

supports for higher education (as part of a larger decline in

funding for a wide array of social programs over the 1980s),

leading to a decrease in the proportion of student aid in the

form of grants, and a steady increase in the proportion of aid

coming in the form of loans (Mattera 1990:132). Student

indebtedness quintupled from the mid-1970s to the mid-1980s,

reaching an annual volume of $10 billion by 1986 (Mattera,

1990:132). These large debt loads can prove to be difficult

to pay off during times of recession and stagnating real

income. It is not uncommon for doctoral students to graduate

with debt loads, and monthly payments, equivalent to the

mortgage on a house, a situation that works against the

likelihood of actually being able to obtain a mortgage on a

house in the near future.

A college diploma is increasingly the ticket to

admittance to the pool of eligible candidates for relatively

high-paying professional, technical, and management


occupations. Yet a four-year degree, or even a master's or

Ph.D. does not guarantee a position commensurate with one's

qualifications in the new economy.

As a personal development strategy, more training and

additional educational attainment are more often than not a

worthwhile investment of time, money and effort, even though

the payoff is not as certain as it once was. As an economic

development strategy for society, increasing educational

attainment, by itself, presents a bleak prospect. Producing

more graduates with 4-year and advanced degrees will benefit

society as a whole only if jobs are available for those


The Effects of Social Interaction on College Attendance

Research has consistently demonstrated that, despite the

public ideal of equal access to higher education, family

stratification position remains the single most important

social predictor of college attendance, rivaling in impact

even high school academic performance (Thomas et al., 1979).

Family socioeconomic status is not the only family

characteristic linked to college attendance, however.

Family structural characteristics--such as the number of

siblings or single parent households--have been found to have

an impact on educational aspirations of high school students

(Hansen and McIntyre, 1989). In addition to family structural

variables, family process variables--such as parent/child


interaction -- have been shown in previous research to exert

an influence on educational aspirations and college attendance

plans of junior and senior high school students (Marini and

Greenberger, 1978; Lomax and Gammill, 1984; Stage and Hossler,

1989). Family process variables, also referred to in the

literature as social integration variables or in terms of the

effects of significant others (Sewell, Haller, & Portes,

1969), are regarded as mediating the effects of family

background variables on educational attainment (Thomas, 1980).

The purpose of this dissertation is to explore the

mediating role of family process variables in transmitting or

modifying the effects of family stratification position on

college attendance, within the context of differences in

family structure. This study also aims to place the

individual and the family within the context of the social

environment of the school and the community, with the

expectation that supportive social interaction in the family,

school and community is positively associated with college


This research will fill gaps in the literature or

otherwise make contributions to sociological knowledge of the

college attendance process in several areas: 1) the research

will be the first to employ social capital theory to account

for differences in college attendance; 2) The study will add

to the body of research that places individuals and families

in a community context; 3) The study will generate a more

complete picture of the role of family process variables in

mediating the effects of family background and family

structure on college attendance; and 4) methodologically, by

disaggregating the SES index, the proposed research presents

the potential of providing more complete understanding of the

individual effects of the concepts comprising the index of

socioeconomic status--parental educational attainment and

family income--on college attendance or nonattendance. This

study will expand knowledge of the factors that make some

students "at-risk" of not attending college. It is hoped that

the analysis may contribute to an expanded conception of being

at risk that goes beyond stratification position and family

structure to include an assessment of family process

variables, as well as school and community social interaction.

In the second chapter, the development of sociological

theories of educational attainment is discussed, these

theories falling under the headings of the status attainment

tradition, human capital analysis, and the social capital

theory of James Coleman. Coleman's social capital conceptual

system is explicated, critiqued, and modified to account for

college attendance. The chapter concludes with a statement of

the model used in the present study and a statement of the

research questions to be addressed.

The third chapter discusses the methodology employed in

the inquiry. That methodology calls for the use of the same

secondary data set--the High School and Beyond Longitudinal

study--and similar statistical procedures utilized by Coleman

in his analysis of high school dropouts. Chapter 3 includes

a description of the data set, a description of the indicators

of the concepts in the model, as well as an explanation of the

statistical procedures to be used in the study.

Chapter 4 presents the findings of the statistical

analysis and a discussion of those findings in terms of the

research questions. Chapter 5 includes a summary of the

findings and offers conclusions concerning the central

questions of the study regarding college attendance.



On a societal level of analysis, the economic importance

of college attendance, as suggested in the preceding chapter,

lies in optimizing the fit of the work-force with a labor

market undergoing compositional shifts, with the aim of

enhancing productivity, economic development, and

international competitiveness. College attendance is part of

the "education pipeline" that allocates the distribution of

occupations and incomes in the society. On an individual

level of analysis, a college diploma is the necessary, but not

always sufficient ticket, for membership in the "middle

class." In the new economy, with rare exceptions, hope for

achieving anything like the "American dream" -- having the

kind of job that makes possible owning a detached single-

family dwelling on a large fenced lot in a quiet and safe

neighborhood with two cars in the garage -- lies in a college

education. In a study of occupational mobility in the United

States, Michael Hout (1988:1391) concludes that the answer to

the old question, "how much schooling does it take to overcome

the disadvantages of low social origins?", is, as a general

rule, "a college degree can do it."



Given the high personal stakes, the question arises as to

why the college attendance rates are not higher than they are,

why some people go to college and others do not. The

predominant sociological analytical framework employed by

researchers seeking answers to this and other questions

concerning educational attainment is the status attainment

model proposed by Blau and Duncan (1967) and augmented by the

"Wisconsin model" of Sewell, Haller, and Portes (1969) and

dozens of variations involving literally hundreds of other

studies. The status attainment model is not so much a theory

of status attainment as it is a set of critical variables

thought to affect educational, occupational, and income status

attainment, as well as the specification of the causal

relationships among the variables. The generic status

attainment model emphasizes the effects of family

socioeconomic status and academic ability mediated through

educational performance and the influence of significant

others. Another substantial body of research on educational

attainment uses human capital or similar econometric

approaches (Stage and Hossler, 1989). Human capital analysis,

introduced by economists (Schultz, 1962; Becker, 1962), frames

college attendance in terms of investments in education. A

third orientation, social capital theory, integrates aspects

of both status attainment and human capital analysis. Though

not previously utilized in the study of college attendance,

social capital theory has been used with promising results by

Coleman and associates in the study of high school dropping

out (Coleman, 1988a; Coleman & Hoffer, 1987; Coleman, Hoffer,

and Kilgore, 1982). The social capital approach is compatible

with both human capital theory and with status attainment

models, adding depth to the former and theoretical content to

the latter.

Family Background in the Study of College Attendance

By general agreement, the single most powerful predictor

of college attendance is the individual-level variable of high

school academic performance, a fact compatible with the

meritocratic ideal. The second most influential factor in

affecting college attendance is the socioeconomic class status

of the family, a fact much less compatible with the

meritocratic ideal. Discussion of the importance of class

raises awkward issues for a society steeped in the myth of the

ever-expanding middle class. President Bush stated in 1988

that "[class is] for European democracies or something -- it

isn't for the United States of America. We are not going to

be divided by class" (Washington Post, November 13, 1988:A27).

The reality, however, is that the United States has a higher

degree of income inequality than any of the European

democracies (Braun, 1991) and an even more lopsided

distribution of wealth, in which the top one-half percent of

the population owns a larger share of the total wealth than


the bottom 90 percent (Mantsios, 1992:99; see Table 2-1). As

Mantsios notes:

The rewards of money...go well beyond those of consumption
patterns and life style. It is not simply that the wealthy
live such opulent life styles, it is that class position
determines one's life chances. Life chances include such far-
reaching factors as life expectancy, level of education,
occupational status, exposure to industrial hazards, incidence
of crime victimization, rate of incarceration, etc. In short,
class position can play a critically important role in
determining how long you live, whether you have a healthy
life, if you fail in school, or if you succeed at work

Table 2-1
Distribution of Wealth in the U.S.

Families Percent of wealth owned

The richest 10% 71.7
(The top 1/2%) (35.1)
Everyone else, or 90% of all families 28.1

Source: Mantsios, 1992: Joint Economic Committee, 1986

Socioeconomic status is linked to educational attainment

in terms of conditioning the environment of support for

aspirations and achievement. Children in families of lower

socioeconomic status are less likely to have supports such as

a private room, a computer in the home, tutoring, or residence

in a district with well-funded schools. Additionally,

children in families with higher socioeconomic status are more

likely to be socialized in their families to value educational

achievement (Wagenaar, 1987).

The social stratification position of the family, then,

is regarded as the starting point and most influential social

factor in models of educational and occupational attainment

(Blau and Duncan, 1967; Sewell et al., 1969). Socioeconomic

class status, however, does not represent a single dimension,

as is the case with sex and race, but is comprised of three

dimensions: income, educational level, and occupational

status. In their groundbreaking study of occupational

mobility, Blau and Duncan (1967) account for sons'

occupational status attainment with a causal model beginning

with father's education and father's occupation. As depicted

in Figure 2-1, father's education and father's occupation were

shown to exert substantial separate direct effects on son's

education, which in turn exerts the largest direct effect on

son's first job (path coefficient=.440) and main job (path

coefficient .394). The family income dimension of

socioeconomic status is not incorporated into the model. With

occasional exceptions, most subsequent research in educational

and occupational status attainment has employed composite

measures of socioeconomic status that typically include

fathers' and mothers' education, fathers' occupation, family

income, and often also a measure of household possessions.

Because these components of socioeconomic status tend to "go

together," and are correlated with each other, using a

composite SES measure usually does not present a problem,

though it is important to remember that socioeconomic status,


or class, is not a single "thing", but instead represents a

complex combination of three or more dimensions.

Figure 2-1
Blau and Duncan's 1967 Model of Male Occupational Attainment
Source: Blau and Duncan, 1967.

Blau and Duncan's study of occupational mobility was one

of the first studies in the social sciences to use path

analysis, and is notable also for inaugurating the status

attainment research paradigm (see Figure 2-1). Within two

years of the publishing of The American Occupational

Structure, Sewell et al. (1969) introduced an expanded model

of educational attainment and early occupational attainment,

which, along with several similar variations (Alexander,


Eckland, and Griffin, 1975), has come to be known as the

"Wisconsin model" of status attainment. The Wisconsin model

incorporated measures of mental ability, academic performance,

significant others' influence, educational aspirations, and

occupational aspirations. Instead of looking at the separate

effects of fathers' education and fathers' occupation, the

Wisconsin model utilized a composite measure of SES which

included fathers' education, mothers' education, fathers'

occupation, and a measure of students' perceptions of family

economic status.

By viewing college attendance as an important point of

achievement in the "socioeconomic life-cycle" of individuals

(Alexander, et al., 1975:324), status attainment research has

consistently found empirical support for the causal primacy of

family class status in educational, occupational, and income

attainment. Since the first studies using the Wisconsin model,

well over 500 subsequent papers have been published either

replicating, extending or disputing the status attainment

model (Campbell, 1983:47), with the result of "a body of

research characterized by an unusual degree of coherence and

cumulativeness" (Alexander et al., 1975:324). Some have gone

so far as to argue that the status attainment tradition comes

as close to a "Kuhnian paradigm" as is to be found in the

social sciences (Campbell, 1983; Bielby, 1981).


In a review of the status attainment literature, Campbell

(1983:47) observes that the Wisconsin model seeks answers to

the following questions:

1. What are the relative impacts of family background
and schooling on subsequent attainments?
2. What is the role of academic ability in the
attainment process?
3. How do aspirations and motivation determine
attainment, and what is the role of family and
school in providing support for aspirations? Do
social psychological variables merely transmit the
effects of family background and/or ability or do
they have an impact of their own?

The Wisconsin model (see Figure 2-2) offers a number of

empirical generalizations in answer to these questions.

Father's education, mothers's education, father's occupation

(it is still uncommon to include mother's occupation in

measures of SES), and family income each influence the status

attainment process at every stage. Academic aptitude likewise

influences each stage of the process, including academic

performance, aspirations, interaction with significant

others, as well as directly affecting educational,

occupational and income attainment. Academic performance

directly influences interactions with significant others,

aspirations and attainment, but is itself influenced by

academic ability and family background. Significant others --

peers, teachers, and parents -- exert direct effects on

educational and occupational aspirations, as well as on

measures of attainment. In the words of Sewell et al.


We assume 1) that certain social structural and psychological
factors -- initial stratification position and mental ability
-- affect both the sets of significant others' influences
bearing on the youth, and the youth's own observations of his
ability; 2) that the influence of significant others, and
possibly his estimates of his ability, affect the youth's
levels of educational and occupational aspiration; 3) that the
levels of aspiration affect subsequent levels of attainment;
4) that education in turn affects levels of occupational

Fathers Education

Mothers Education

Fathers Occupation

Family Income/

Figure 2-2
The Wisconsin Model of Status Attainment

Source: Alexander, Eckland, and Griffin,


The body of status attainment research reveals that

academic performance is the single most powerful predictor of

educational aspirations and educational attainment, suggesting


that a large proportion of non-attendance in college is due to

low academic ability and poor school performance. Yet the

research also reveals that a substantial proportion of low-SES

students with the academic ability to do well in college

nevertheless do not attend college. The explanation suggested

by the status attainment model is that "their parents do not

provide the psychological or financial support for mobility"

(Campbell, 1983:59). Financial supports are determined by the

family socioeconomic class status. Psychological supports are

provided by the influence of the "significant other." Though

friends and teachers are sometimes treated as significant

others, this designation is associated predominantly with the

influence of the family.

The Family as "Significant Other"

The inclusion in the status attainment model of

indicators of significant others influences is consistent with

research in psychology and education using social-

psychological variables. As early as the 1920's,

psychologists were aware of important statistical associations

between family structural variables and individual abilities

(Marjoribanks, 1972). Researchers determined that sibsize --

the number of children in the family -- correlates about -.3

with the mean cognitive ability scores of the children in the



Sibsize was taken up and incorporated with sibling

spacing, birth order and parent's intellectual level to form

the confluence model (Zajonc, 1976; Zajonc & Marcus, 1975) to

account for variation in cognitive development. The

explanation of the connection between family size and

intellectual development is based on the dilution hypothesis,

stated by Blake (1981:422) as "the more children, the more

parental resources are divided...and hence, the lower the

quality of output." The importance of family configuration

to educational attainment is in the nature of the familial

environment of support; whether parental attention is divided;

whether older brothers or sisters are around to read to the

younger children and encourage them, and so forth.

In addition to family configuration and family

socioeconomic status, patterns of family interaction are

linked to educational attainment and subsequent occupational

attainment. Studies pointing to the importance of family

interaction variables (Kent & Davis, 1957; Marjoribanks, 1972)

suggest that "about half the variance in verbal ability can be

accounted for by sociopsychological assessments of the family

environment" (Walberg & Marjoribanks, 1976:532-534). However,

Widlak & Perrucci's 1988 study has been one of the few to

examine family interaction along with family configuration in

seeking to understand the relationships between family

environment and intellectual development. The authors found

empirical support for the hypothesis that cognitive

development is positively related to parental and sibling

support and encouragement.

A substantial body of research exists to suggest that

interaction with parents can have a substantial effect on

formation of educational goals and aspirations (Haller and

Portes, 1973; Kandel and Lesser, 1969). Parental

encouragement for college is positively associated with plans

for college and college enrollment (Murphy, 1981).

The basic theoretical perspective views educational goals as
one outcome of the socialization process and the family as a
major agent of socialization. It posits a strong future
orientation for parents as they view their maturing children,
and assumes that the intimate interaction between parent and
child is the context within which parental views of the future
are transmitted to the child. Research based on this
perspective has provided considerable evidence of parental
influence on the child's goals (Kerckhoff and Huff 1974).

That socioeconomic class status exerts an independent

direct effect on educational and occupational aspirations is

a "sociological truism" (Sewell and Shah, 1968:559).

Aspirations have been found to be sensitive to the influence

of parental encouragement of "high educational and

occupational goals" (Sewell and Shah, 1968:560), and children

of higher socioeconomic status are more likely to encourage

high levels of attainment. Parental support, or non-support,

of high attainment is regarded in the status attainment

tradition as an intervening variable transmitting and

mediating the effects of family socioeconomic class



As noted previously, psychological research has

demonstrated that number of children in the family is

negatively related to cognitive development. The dilution

hypothesis suggests that parental attention is diluted in

larger families, thereby reducing opportunities for supportive

parental interaction with children. In a similar manner, two-

parent and single-parent families provide differential

opportunities for parent-child interaction.

A considerable body of literature extending back several

decades has examined the psychological and social effects of

single-parent families on children. For example,a study

conducted by Deutch and Brown (1964) concluded that, among

blacks, differences of eight IQ points were attributable to

father absence from the home. Examining school participation,

Burchinal (1964) found that:

Lower and similar school-activity scores were observed for
boys who lived with their mothers only or with mothers and
stepfathers. Girls from unbroken homes were clearly more
active in school activities than other girls. Girls living
with their mothers only were the next most active (Bales,

In a study of the effects of the home environment on the

academic performance of "disadvantaged" boys, Peterson,

Debord, Peterson and Livingston, (1966) hypothesized that "the

nuclear family with few children (three or less) will provide

the most stable environment and thus be positively associated

with academic achievement (Bales, 1979 :149). Coleman (1966)

developed measures of the structural "integrity" of the

family, based on the presence of an intact or a "broken" home.

In the 1990s, use of this kind of terminology might be

construed as implying moral criticism of single mothers and/or

absent fathers. Because so many economic and social factors

beyond individual control work against the maintenance of

nuclear families, no such moral criticism is warranted, even

though two-parent families are clearly the ideal. The

important point is that the presence or absence of a two-

parent household does make a difference in the structure of

opportunities for parent-child interaction.

Expanding the Significant Other

Much research on effects of family environment on

educational attainment has proceeded upon the reasonable

assumption that the "home produces the first and perhaps most

subtle influence on the mental development of the child"

(Marjoribanks, 1972:324). However, the family environment is

only part of the "total network of forces" acting upon the

individual, a network that includes the home, the school and

the community (Marjoribanks, 1972:324). According to Wagenaar

(1987), attention to school and community structural variables

is useful in helping to situate "individual level correlates

within a larger context, thereby showing how individual

decisions can be affected substantially by social structure"

(Wagenaar, 1987:174).

Wehlage and Rutter (1986) reported that school size and

other organizational aspects of school contribute to a sense


of alienation and estrangement. The finding that number of

students in the school affects the social environment for

learning bears similarity to the findings concerning the

effects of family size and presence of parents. The structure

of the social environment conditions the process of


Social Interaction and College Attendance

Early on, status attainment studies included measures of

peer and teacher interaction with students (Sewell et al.,

1969; Alexander et al., 1975). Sewell et al. (1969), in the

first formulation of the Wisconsin model of status attainment,

employed an "index of significant others' influence" that

included youth's report of parental and teacher encouragement

for college as well as friends' college plans. The effects of

interaction with family, teachers, and friends on occupational

attainment are explained in terms of support and

encouragement, or lack thereof, of higher educational and

occupational aspirations.

Another way of thinking about the effects of social

interaction on educational attainment is to regard certain

kinds of social interaction as contributing to the integration

of the individual into the society -- a society that values

and requires evidence of educational attainment. Eckstrom,

Goertz, and Pollack (1986), and Wehlage and Rutter (1986)

found that dropouts are more alienated than school stayers.

Wagenaar (1987) found that dropouts are characterized by

normlessness and social isolation. Stinchcombe (1965)

suggested that students who do not participate in

extracurricular school activities are more alienated from the

academic programs. Rehberg and Schafer (1968) found that

educational expectations were positively influenced by

participation in school athletics. Hanks and Eckland (1976),

on the other hand, found little effect of athletic

participation on educational attainment; they did, however,

find stronger effects for participation in other

extracurricular activities. These findings confirmed those of

Otto (1976) and Spady (1970), who developed early formulations

of the proposition relating social integration to status

attainment. Otto operationalized adolescent social

integration as participation in high school extracurricular

activities, including athletics, band, chorus, dramatics,

debate, 4-H or FFA, the school paper, student government, or

hobby clubs.

Participation in extracurricular activities, or more

generally, social integration, is seen as mediating the

effects of background socioeconomic status. Hanks and Eckland

(1976:292) suggest that, "participation in other

extracurricular activities...serves an important integrative

function in school and college by fostering the acquisition

and transference of status across adolescent and adult social

systems." Additionally, social participation "encourages


compliance" (p. 292) with the norms and attitudes associated

with educational attainment.

Typically, the influence of social interaction,

participation, and social integration are examined in the

context of the family and the school, and less often at the

level of the community:

The tradition growing out of [status attainment] research has
concentrated on the individual's mobility as determined by
socioeconomic and ethnic background. Unfortunately, most of
these studies overlook the significance of another factor: the
characteristic of community in which one resides and carries
out most activities. As a result, the significance of the
social context affecting status attainment, and the
consequences for individuals, are also ignored (Semyonov,

Studies measuring community effects on status attainment

have produced mixed results. A number of studies found little

or no effect of community size on occupational status

mobility. Hauser and Featherman (1977:269), in a study

analyzing four size categories of urban communities, concluded

that "contextual differences varying concomitantly with city

size do not alter the process of stratification in significant

ways." Other studies agree that individual-level variables

exert greater influence on the status attainment process than

do community-level variables (Muller, 1974). On the other

hand, a number of studies have suggested that size of

community is "systematically associated with other contextual

characteristics of a locality such as...level of

industrialization, economic composition [and] the occupational

structure" that do affect the status attainment process

(Semyonov, 1981:361; Blau and Duncan, 1967; Lane, 1968). Blau

and Duncan (1967) advanced the view that, due to the greater

degree of functional differentiation in larger communities,

opportunities for enhanced status attainment are fewer in

smaller communities. Lane (1968:741) noted the connection

between the socioeconomic class structure and the size and

complexity of communities: "A system of stratification

interpenetrates with community structure sufficiently to

produce divergent patterns of mobility."

Table 2-1
Relationship Between Place of Residence and College Plans:

Place of Percentage of students
Residence Going to College

Farm 21.5
Village 27.9
Small city 33.9
Medium city 37.0
Large city 42.4

Rural 24.7
Urban 37.3

Source: Sewell, 1964:34.

Size of place effects are more pronounced for educational

attainment. For over forty years, studies have consistently

demonstrated that rural students have lower educational and

occupational aspirations and expectations than small-town and

urban students (Sewell, 1964; Cobb, McIntire, and Pratt,


1989). During the 1950s and 1960s, the relationship between

size of place and college aspirations was positive and linear,

as evidenced in Table 2-2.

Table 2-2
Response of High School Sophomores to the Question, "How Far
in School Do You Think You Will Get?, by Place of Residence:

Educational expectations Urban Suburban Rural

Less than high school .7 .3 .8
High school grad only 14.1 13.7 22.8
Less than two years at
business or voc. sch. 5.8 6.4 10.2
Two years or more at
business or voc. sch. 11.9 10.3 12.8
Less than two years
of college 3.2 2.8 2.8
Two or more years of
college degree with
Associate Degree 12.3 12.6 12.6
Finish college with
Bachelors 26.1 27.8 22.6
Master's or equivalent 13.1 14.2 9.0
Ph.d., M.D. or equivalent 12.9 11.8 6.3

Source: Cobb, et al., 1989; High School and Beyond, 1980.

The economic decline of central cities and the

corresponding movement of high SES families into suburban

areas over the last two decades has produced a curvilinear

relationship between educational aspirations and size

community of residence, as suggested by Table 2-2. The lowest

levels of aspirations continue to be in rural areas, but

suburban areas now exceed those of urban areas. Rural


students also perceive less support for college from their

parents, teachers and guidance counselors, report lower

occupational status aspirations, and have less confidence in

their ability to do college work (Cobb, et al., 1989; High

School and Beyond, 1980).

Density of population affects the occupational

opportunity structure and the demand for education. Recent

research suggests that characteristics of the community may

also affect the opportunity structure of social interaction

(Coleman et al., 1982; Coleman and Hoffer, 1987; Coleman,

1988a; Smith, Beaulieu, and Israel, 1992). Coleman and his

associates found that student integration into the community

and participation in community organizations was associated

with lower dropout rates. According to Coleman, participation

in community activities, particularly church, is a measure of

what he referred to as "social capital."

Human Capital and Social Capital

Along with the status attainment framework, the principal

research orientation to educational attainment has been human

capital theory. Human capital theory was first developed by

the economists Schultz (1962) and Becker (1962) to account for

increases in productivity that could not be explained by

improvements in technology or financial capital. The idea

behind human capital is that the skills, talents and knowledge

of people amount to a kind of "capital" analogous to financial

assets. The theory suggests that, assuming that people are

rational, individuals make investments in their human capital

stock with the expectation of realizing benefits -- higher

income and a better job -- in the future. The principal

avenues of human capital enhancement are formal and informal

schooling and job training.

Human capital theory, like status attainment research,

has generated hundreds of studies. Its appeal lies in the

reasonableness and demonstrability of its fundamental

proposition, that increased investments in education and

training lead to higher incomes and greater productivity.

Human capital theory has several weaknesses, however. The

"more narrowly individualistic focus" of the human capital

approach leads to a "neglect of social structural factors"

(Fligstein, Hicks, and Morgan, 1983:291). The concern of

Fligstein et al. is with the neglect of the fit of the

individual with the structure of the labor market.

The view of education as an investment with calculable

returns carries with it the assumption that the individual has

the resources necessary to carry out rational educational

investment decisions. One implication of this assumption is

that low status occupational attainment and low income can be

interpreted as a failure to invest on the part of individuals.

The role of family class background in determining which

students are able to make higher level educational investments

tends to be obscured through inattention. Human capital


analysis can predict the economic payoff for each additional

increment of education, but cannot explain why some people

choose to make investments in their education and some do not.

One reason for this is human capital theory's

questionable assumption of more-or-less equal available

resources for investment. According to one line of research,

another reason for the "failure to invest" in human capital is

the lack of "social capital." Social capital theory draws

upon human capital theory by way of parallel conceptual

structure, yet is consistent with the status attainment

tradition that places family stratification position foremost

in the causal sequence. While human capital consists of

individual skills, talents and knowledge, social capital is

comprised of the social resources available to individuals in

the form of interaction and networks of interaction. "If

physical capital is wholly tangible, being embodied in

observable material products, and human capital is less

tangible, being embodied in the skills and knowledge acquired

by an individual, social capital is still less tangible, for

it exists in the relations between persons" (Coleman,


The conceptualization of social relations as a kind of

"capital" or resource was developed by James Coleman and his

associates (Coleman, 1988; Coleman & Hoffer, 1987; Coleman,

Hoffer & Kilgore, 1982). Using the massive High School and

Beyond longitudinal survey following some 30,000 high school


sophomores in 1980 and 1982, with somewhat smaller samples in

1984 and 1986, Coleman concluded that supportive interpersonal

relations on both the family and the community levels reduces

the risk of dropping out and enhances prospects for

educational attainment.

At the level of the family, social capital reflects the

nature of the relations that exist among family members. The

child's access to the parents' human capital--that is, parents

educational level, as distinct from the income dimension of

socioeconomic status--depends in part on the physical presence

or absence of the parents in the home, and in part on the

quantity and quality of the interaction between parents and

child. A family can have high human capital, yet if the

parents do not interact with the children, the human capital

is less effective.

At the community level, social capital exists in the

norms, social networks, and interactions between adults that

facilitate or support educational attainment. The form of

interaction most conducive to the enhancement of social

capital is referred to as interQenerational closure by

Coleman. Intergenerational closure is a relationship

structure in which "a child's friends and associates in school

are sons and daughters of friends and associates of the

child's parents" (Coleman, 1990:318). In such a situation,

other adults in the community are available to reinforce norms

and values consistent with educational attainment.


This formulation of the characteristics of social capital

raises certain questions. On the one hand, the argument is

advanced that social capital is a resource to assist the

individual in the acquisition of human capital. This

formulation is consistent with the economic behavior of the

rational actor. On the other hand, social capital is

conceptualized as constraining and enabling individual

behavior in the manner of a structural variable. Not only is

social capital highly intangible, but it embodies what might

be regarded as a paradigm clash.

In his definitive statement of the social capital

formulation in the article, "Social Capital in the Creation of

Human Capital" (1988:895), Coleman discusses these clashing


There are two broad intellectual streams in the description
and explanation of social action. One, characteristic of the
work of most sociologists, sees the actor as socialized and
action as governed by social norms, rules, and obligations.
The principal virtues of this intellectual stream lies in its
ability to describe action in social context and to explain
the way action is shaped, constrained, and redirected by the
social context.
The other intellectual stream, characteristic of the work
of most economists, sees the actor as having goals
independently arrived at, as acting independently, and as
wholly self-interested. Its principal virtue lies in having
a principle of action, that of maximizing utility."

Coleman sees both of these intellectual streams as being

defective as an approximation of reality. Over-emphasis on

structural constraints leads to an "over-socialized"

conception of the individual, in which the individual has no

autonomy or volition, a problem explored also by Wrong (1961).


For its part, the rational actor framework "flies in the face

of empirical reality," inasmuch as "persons' actions are

shaped, redirected, constrained by the social context; norms,

interpersonal trust, social networks, and social organization

are important in the functioning not only of the society but

also of the economy" (Coleman, 1988a:487).

Coleman advocates a synthesis of these divergent

intellectual streams that "accepts the principle of rational

or purposive action," connecting this principle with

"particular social contexts" (1988a, S96). Coleman envisions

a synthesis that "can account not only for the actions of

individuals in particular contexts but also for the

development of social organization" (1988a:S96), an

accomplishment that would amount to the discovery of the

heretofore "missing link" between the micro and the macro.

Coleman offers the conceptualization of social capital as part

of the synthesis, and while he does not quite succeed in

theoretically synthesizing the rational actor with social

structure, there is merit in his approach.

As a bridge between the rational actor and the social

structure, the social capital conceptualization is consistent

with the concept of "embeddedness" adopted by Granovetter

(1985). Granovetter's concern, similar to Coleman's, was to

find a theoretically happy medium between the

"undersocialized" individual, or "atomized-actor", on the one

hand, and the "oversocialized" conception of the individual as


criticized by Wrong (1961) and by Duesenberry (1960) on the

other, who observed that "economics is all about how people

make choices; sociology is all about how they don't have any

choices to make" (Granovetter, 1985:485).

In his article, "Economic Action and the Social

Structure: The Problem of Embeddedness," Granovetter (1985)

examines the importance of trust and malfeasance in

maintaining and disrupting market behavior while arguing for

an "embeddedness" conceptualization:

"The embeddedness argument stresses...the role of concrete
personal relations and structures (or "networks") of such
relations in generating trust and discouraging
malfeasance.... The embeddedness approach...threads its way
between the oversocialized one of impersonal, institutional
arrangements by following and analyzing concrete patterns of
social relations. Unlike each makes no
sweeping (and thus unlikely) predictions of universal order or
disorder but rather assumes that the details of social
structure will determine which is found" (1985:490).

Granovetter's aim is to retain the rational actor but to

superimpose social structure. Coleman's aim is to retain

social structure but to bring in the rational actor. He does

this by arguing that social capital is a resource for the

individual and that this resource inheres in social structure

itself (1988a:S98):

If we begin with a theory of rational action, in which each
actor has control over certain resources and interests in
certain resources and events, then social capital constitutes
a particular kind of resource available to an actor.
Social capital is defined by its function. It is not a
single entity but a variety of different entities, with two
elements in common: they all consist of some aspect of social
structures, and they facilitate certain actions of
actors...within the structure. Like other forms of capital,
social capital is productive, making possible the achievement
of certain ends that in its absence would not be possible.


Like physical capital and human capital, social capital is not
completely fungible but may be specific to certain activities.
A given form of social capital that is valuable in
facilitating certain actions may be useless or even harmful
for others.
Unlike other forms of capital, social capital inheres in
the structure of relations between actors and among actors.

Clearly, Coleman is working from a strong structuralist-

functionalist perspective:

The value of the concept of social capital lies first in the
fact that it identifies certain aspects of social structure by
their functions, just as the concept "chair" identifies
certain physical objects by their function, despite
differences in form, appearance, and construction. The
function identified by the concept of "social capital" is the
value of these aspects of social structure to actors as
resources that they can use to achieve their interests

Though he does not use the terminology, Coleman's

treatment of social capital suggests that, in many cases it

represents a latent function of social organizations. For

example, church organization provides access to a network of

social supports in addition to its primary function of

spiritual nurturance.

As examples of aspects of social structure that serve as

resources for actors, Coleman points to research by De Graf

and Flap (1988) that demonstrates how "informal social

resources are used instrumentally in achieving occupational

mobility" (Coleman 1988a:S102). He suggests that trust and

trustworthiness are aspects of social structure that serve as

social capital, inasmuch as more can be accomplished in social

relations where such properties are present. Still another

form of social capital is norms:

When a norm exists and is effective, it constitutes a
powerful, though sometimes fragile, form of social capital.
Effective norms that inhibit crime make it possible to walk
freely outside at night in a city and enable old persons to
leave their houses without fear for their safety. Norms in a
community that support and provide effective rewards for high
achievement in school greatly facilitate the school's
task....effective norms can constitute a powerful form of
social capital. This social capital, however, like the forms
described earlier, not only facilitates certain actions; it
constrains others. A community with strong and effective
norms about young persons' behavior can keep them from "having
a good time" (1988a:S105)

Social capital, then, is a resource to be used

instrumentally by individuals, but is also a structural factor

that enables and constrains individual behavior -- "keeps them

from having a good time." It enforces norms that encourage

and support educational attainment as well as enforcing norms

constraining inappropriate behaviors. The most valuable

social capital in this regard is produced by certain kinds of

social structures more than others, the leading example being

what Coleman refers to as intergenerational closure.

Figure 2-3 graphically represents a structure without

closure and with closure. In Figure 2-3a, the children, B and

C, are friends, but their parents, A and D, do not know each

other. In Figure 2-3b, "the parents' friends are the parents

of their children's friends" (1988a, S106):

The consequence of this closure is...a set of effective
sanctions that can monitor and guide behavior. In the
community in figure [2]-3b, parents A and D can discuss their
children's activities and come to some consensus about
standards and about sanctions. Parent A is reinforced by
parent D in sanctioning his [or her] own child, C, but also
for the other child, B. Thus, the existence of
intergenerational closure provides a quantity of social
capital available to each parent in raising his [or her]

children -- not only in matters related to school but in other
matters as well (1988b, S107)

E |A

D- F

B (C


Figure 2-3
Network involving parents (A D) and child )

and with (b) intergenerational closure.

SOURCE: Coleman, 1988a.

While it was once the norm, intergenerational closure is

no longer very common, Coleman argues, due to the geographical

mobility and individualism of modernity (Coleman and Hoffer,

1987; Coleman, 1990; Coleman, 1988b:388):

The social capital of intergenerational closure exists in some
isolated small towns and rural areas where the social
relations among adults are restricted by geographic distance
and residential mobility less important. Intergenerational
closure exists in schools based in a religious community, such
as Catholic schools, although the social relations which make

---- ( D c i d e ( C i .

up the community are more narrowly focused around a single
dimension of social life, the religious institution....
intergenerational closure does not now exist in most modern
public schools or in most non-religiously-based private
schools. The absence of social capital represents the loss of
a resource for young persons.

Coleman and his associates argued that lower dropout rates

among Catholic high school students are attributable to the

intergenerational closure associated with the communities

surrounding Catholic high schools.

Substantial theoretical grounds exist in support of

Coleman's pronouncement of the demise of intergenerational

closure surrounding public schools. The theme of the

"terminal eclipse" (Wilkinson, 1986, p. 4) of the local

functional community is based on the view that in modern mass

society individuals' functional dependence on local community

has been replaced by attachments to outside corporations,

national culture and international markets (Warren, 1978).

Electronic and satellite technology have contributed to

"territory-free networks" of social interaction (Wilkinson,

1990, p. 155). Territorially-based interaction represents

only one pattern of community, "a pattern that becomes less

and less evident over the course of American history" (Bender,

1978, p. 6).

The "demise of community" thesis has not gone

unchallenged, however. Wilkinson (1990, p. 154) argues that:

"unless suppressed by barriers to authentic social

interaction, community always occurs where people live

together, whether or not they realize it and whether or not


they like it." Individuals have many connections to the

larger society, yet these connections are made principally

through interactions at the local level. Smith et al. (1992)

suggest that, granting that modern social conditions tend to

erode a sense of local community, the process is nevertheless

historical and transformative.

Besides the decline in community social capital, Coleman

is concerned about the loss of social capital in the family.

The primary reason for this loss is the increase in single-

parent (usually single-mother) families and the increase in

mothers who work outside the home, seen as a problem for the

family and as contributing to the decline of intergenerational


In schools where there are a dense set of associations among
parents, these associations are commonly the result of a small
number of persons, ordinarily mothers, who do not hold a full-
time job outside the home. These mothers themselves
experience only a small part of the benefits of the social
capital surrounding the school. If one of the mothers decides
to abandon these activities, perhaps to take a full-time job,
the action may be entirely reasonable from a personal point of
view and even from the point of view of that family itself.
The benefits of the new activity to the individual and to the
family may far outweigh the losses which arise from the
decline in associations with other parents whose children are
in the school, but the withdrawal of these activities
constitutes a loss to all those other parents whose
associations and contacts were dependent on them (Coleman,

Coleman refers to the physical absence of the father or

mother as a structural deficiency: "Single parent families and

families in which the mother worked before the child entered

elementary school represent two forms of structural


deficiency" (1988b:385). Social capital deficiencies are seen

as having not only a structural, but also a functional,


Functional deficiency in the family refers to the absence of
strong relations between children and parents despite the
physical presence of the family members in the household and
the opportunity for strong relations. Functional deficiencies
may result from the child's embeddedness in a youth community,
from the parents' embeddedness in relationships with other
adults which do not cross generations, or from other sources.
Whatever the source, the child does not profit from the human
capital of the parents because the social capital is
The distinction between the human capital existing in the
family and the social capital existing in the family
constitutes the critical difference between what may be called
the "traditional disadvantage" of background and what I have
termed "family deficiencies." Disadvantaged background
ordinarily refers to the absence of resources embodied in the
parents, represented primarily by the parents' education but
also by other variables, such as low economic level or the
status of a racial-ethnic minority, which stand as surrogates
for low levels of human capital. By family deficiencies, I
mean the weakness of the links between the adult members of
the family and the children constituting an absence of social
capital" (1988b:385).

Critique of the Coleman Analysis

While the social capital conceptualization developed by

Coleman makes an important contribution to the study of

educational attainment, the conceptual structure is not

without its shortcomings. Aside from the dubious

appropriateness of treating economic status and race/ethnicity

as merely "surrogates" for the educational level of the

parents, there are other theoretical problem-areas. First,

the claim that social capital provides a theoretical linkage

between the micro and the macro levels of analysis may


overstate the value of the concept. Second, the rigidly

structural-functionalist formulation of social capital used by

Coleman leads to several related problems with understanding

the effects of social interaction on educational attainment.

Coleman claims that social capital serves as a "resource"

for the individual in his or her efforts to increase human

capital stock, but also indicates that social capital

functions to enforce norms and values that are consistent with

educational attainment, therefore linking micro-level economic

behavior and social structure. Where the dependent variable

is occupational attainment, it is easy to see how the

individual might use informal social networks instrumentally

to obtain a job.

Where the dependent variable is completion of high

school, or college attendance, the instrumental use of social

capital is less probable, though it is possible, as in the

situation of a marginal student who has decided that it is

important to stay in high school and seeks advice from an

adult member of the family, the school or the community. The

more typical operation of social capital is in the supportive

family, school and community social environment which

encourages and enforces pro-educational norms and behaviors.

It should be noted, however, that capital of any variety

has a structural dimension as well as an individual one. For

example, while the financial capital of the family is known to

have a positive effect on the educational attainment of

children, financial capital may also affect the educational

attainment process contextually: "Higher social class parents

generally live in wealthier neighborhoods, in which more money

is available for education. Such increased funds help attract

and retain more highly skilled teachers and help provide more

specialized services, more instructional resources, more field

trips, and better facilities" (Wagenaar, 1987:169-70).

Additionally, it might be argued, the human capital, not only

of the family, but of the community as a whole, influences the

intellectual environment for achievement.

In the case of social capital, the structural dimension

is, perhaps, even more salient. While it is possible to think

of social norms, for example, as a resource to be used by

students, norms and the sanctions that enforce them are more

typically considered as factors that constrain and condition

the individual. From the individual resources perspective the

student trying to decide whether or not to go to college may

seek out supportive advice from an adult friend of the

parents. From the structural perspective, a network of adults

acts as "sentinels" monitoring the activities of children and

apprising parents of inappropriate behavior, as well as

encouraging appropriate behavior. The principal value of

social capital is not so much in serving as a resource for a

student who has already made a decision about educational

attainment, but is rather in shaping the decision itself.

Unlike investments in financial capital or investments in

human capital, the individual does not "invest" in social

capital in the same manner as with financial capital and human


There is a property of social capital that differentiates it
from both physical capital and human capital. This property
of social capital has serious implications for the social,
psychological, and cognitive growth of young persons in the
United States. In western society in general, physical
capital is ordinarily a private good, in that the person who
invests in physical capital may capture the benefits produced
by the capital through his or her property rights in the
capital. The incentive to invest in physical capital is not
depressed; there is not a suboptimal investment in physical
capital because those who invest in it are able to capture the
benefits of their investments [though there may be a
suboptimal distribution of the financial capital required to
invest in physical capital, ed]. For human capital (at least
human capital of the sort that is produced in schools), the
person who invests the time and resources in building up this
capital reaps its benefits in the form of a higher-paying job,
more satisfying or higher status work...
Social capital of the sort that is valuable for a young
person's education is not a private good. The kinds of social
structures which make social norms possible and the sanctions
that enforce them, do not benefit primarily the person or
persons whose efforts would be necessary to bring them about.
The benefits extend to all those who are part of such a
structure (Coleman, 1988b:388-389).

These two frameworks of analysis--rational actor and

structural--can be seen as analogous to an optical figure-

ground reversal. When the attention is focused on the

individual figure, the background loses focus. Conversely,

when the focus shifts to the background, the foreground

becomes blurred. It is easy to shift back and forth between

figure and ground, but difficult to hold both in focus at the

same time.

Coleman's solution (1988a:S105), not entirely

satisfactory as a formal theoretical formulation, treats the

individual as "embedded" in social structure, attempting to

retain "the conception of rational action but to superimpose

on it social and institutional organization." Superimposition

of structure on rational action does not necessarily clarify

the theoretical linkages between the two, however. The concept

of "social capital," then, does not go a long way toward a

theoretical synthesis of micro and macro-level analysis.

Given this caveat, though, the theoretically "looser"

formulation used in the present study represents the

individual student as purposive and relatively rational and

embedded in a structure of relationships in the family, school

and community. That structure of relationships enables

individuals, while it also constrains individuals. The micro

and macro levels are thus integrated to the extent that both

levels are taken into account, yet no claim is made here that

anything approaching a formal theoretical synthesis is being


The abstract progression from physical to financial to

human to social capital involves an elegant parallel

conceptual structure, to be sure. However, in the sense of

theory as a set of interrelated propositions, from which

testable hypotheses are deduced as part of process leading to

the development of scientific laws, the social capital

framework is not a full-fledged, formal "theory." "Much of

what is considered theory in the social sciences consists of

conceptual frameworks that direct systematic empirical work"

(Nachmias and Nachmias, 1981:42). The social capital

formulation is a conceptual framework, or what Turner refers

to as an "analytic scheme" (1991:9). Though the social

capital model provides a useful arrangement of concepts and

understanding of the relationships among those concepts, it is

prudent not to over-reach in making claims concerning the

theoretical sophistication of the formulation.

Structure and Function

The other problematic aspect of Coleman's theoretical

explication of the role of social capital in the educational

attainment process lies in the rigid structural-functional

approach used. As noted above, Coleman suggests that social

capital "identifies certain aspects of social structure by

their functions, just as the concept 'chair' identifies

certain physical objects by their function" (1988a:S101). This

formulation leaves the implication that enforcing and

promoting norms and values of educational attainment is a

"function" of the family; that staying home and socializing

the children to value educational attainment is a "function"

of mothers; that it is a "function" of parents to monitor the

behavior of their children's friends and to encourage them to

stay in school, get good grades and go to college.


Families with single parents--usually single mothers--and

families in which the mother worked before the child enters

elementary school, are categorized as "structurally deficient"

in the Coleman analysis. The working mother is doubly

dysfunctional in this view, causing a structural deficiency

both in the family and in the community, in the latter

breaking the intergenerational closure by "abandoning" the

"dense set of associations among parents" when she takes a

full time job (she is triply guilty if she is also single).

Social Capital and "Family Values"

Because of its intrinsic concern with system maintenance

and stability, structural-functional analysis can be

interpreted in a manner consistent with conservative social

and political ideology. In the terms of the contemporary

public debate over "family values," Coleman's formulation

could be construed as implying that the only "functional"

family is a two-parent family in which the father works

outside the home and the mother stays home to take care of the

children, even though Coleman takes pains to acknowledge that

the mother's abandonment of her function "may be an entirely

reasonable action from a personal point of view and even from

the point of view of that household with its children"


Granting that the ideal family situation for children

includes two parents who love them, Coleman's formulation


places too much responsibility for building and sustaining

social capital on the mother, thus neglecting the mutual

responsibilities of fathers and mothers to "be there" for

their children. The economic fact of life in the 1990s is

that in most families both the father and the mother must

participate in the labor market. Even in the extremely

unlikely event of a major turnaround in the economy such that

one salary could support a middle-class lifestyle as in the

1950s and 1960s, while some women would probably return

willingly to their traditional role as housewives (like Rosie-

the-Riveter cheerfully giving up her factory job for returning

World War II soldiers), educated modern women are unlikely to

give up the financial and personal autonomy that comes from

participation in the labor market.

In any event, there is nothing inherent in the concept of

social capital that compels the a priori assumption that a

father working while the mother stays home with the children

is any more functional than a family in which the mother works

and the father stays home to take care of the children or a

family in which both parents work and take turns arranging to

spend time with the children.

Another problem arising from the strong structural-

functional paradigm Coleman uses is the assertion that social

capital inheress" in social structure itself. A certain kind

of social capital is said to exist simply because both parents

are in the household, for example. The absence of one parent

results in a structural deficiency in the family social

capital. However, in the situation wherein both parents are

in the household, if the parents do not interact

constructively with their children, a functional deficiency is

said to exist. Even if we say that the structure is sound but

is dysfunctional, social capital can hardly be said to

"inhere" in the mere fact of the presence of structure.

Financial capital does not inhere in the structure of the

bank, but must be deposited there. Human capital does not

inhere in the individual, but must be instilled therein.

Similarly, social capital does not inhere in the structure of

the family or the community, but is invested in the structure

through social interaction. The size, shape and other similar

properties of the social structure determine the opportunities

for relationships and interaction within the structure.

Social capital exists in the relationships between people

rather than in the social structure itself, though the

structure influences the patterns of relationships. The

stability of the family structure may be seen as facilitating

nurturance of relationships, therefore assisting in the

development of social capital. Investments in social capital

are made through the efforts involved in relationships as well

as through supports for relationships.

Rather than viewing social capital as intrinsic to social

structure and certain kinds of interaction as functions of

that social structure, the approach adopted in the present


study is to distinguish between structure and process as

complementary components of social capital. Two-parents or

one-parent in the household, and whether or not the parents

work, each of these family structural factors influence the

process of interaction by affecting the density of

relationships and frequency of interaction between parents and

children. The physical structure of the presence of family

members, then, sets the opportunity pattern for social

interaction -- process -- in the family.

Though the process of interaction is required in the

generation of social capital, social capital does not inhere

in the process of interaction per se. This is evident in

Coleman's category of functional deficiencies, interaction

which does not facilitate productive behavior or enforce norms

and values consistent with educational attainment. In other

words, since not all interaction qualifies as social capital,

social capital cannot be said to inhere in interaction. Both

structure and process may be regarded as necessary but not

sufficient conditions for the existence of social capital.

The social capital is present, not in the structure of

relations, as such, nor in the process of social interaction,

as such, but in intangible relationships. The physical

structure of the family provides the opportunities and the

stability for the growth of relationships. Whether the

specific qualities of the relationship qualify as social

capital is determined by the process of interaction.



Figure 2-4
A network with closure

Source: Coleman and Hoffer, 1987.

Coleman and Hoffer note:

In a diagram like that of Figure [2-4], representing relations
between four persons, A, B, C, and D, the human capital
resides in the nodes. Social capital and human capital are
often complementary. For example, if B is a child and A is an
adult parent of the child, then in order for A to be useful
for the cognitive development of B, there must be capital in
both the node and the link, human capital held by A, and
social capital in the existence of the relation between A and
B (Coleman and Hoffer, 1987:222).

The social capital does not exist in the nodes of the

structure, the individuals, but are seen as existing in the

lines linking the nodes. What do these lines represent?

According to Coleman and Hoffer (1987) the lines represent the

relationship between the individuals. Is that relationship

the same thing as interaction, though? If interaction is the

same thing as a relationship, and is a repository of social

capital, then social capital could be operationalized simply

as interaction, but the principle has already been established

that not all interaction qualifies as, or contributes to the

development of, social capital.

Consider what happens when interaction is not taking

place. After an episode of interaction concludes and a parent

and child go their separate ways, does that means that the

social capital ceases to exist? Interaction is discrete and

episodic, whereas relationships exhibit duration and

evolution. That is why a loss takes place when a family moves

-- relationships developed over some considerable length of

time through manifold interactions are terminated.

Opportunities for new patterns of interactions may present

themselves, but relationships take time to build.

Relationship requires structure -- that is, individuals at the

nodes -- and requires interaction, but relationship -- as well

as social capital -- is not exactly the same thing as the

structure of relations nor is it the same thing as

interpersonal interaction.

The social capital in a relationship is, as Coleman

observed, highly intangible, existing at the levels of meaning

as well as of feelings. Trust, love, caring, responsibility,

communication, sharing and respect are all qualities of

relationships that facilitate the transmission of values,

norms, aspirations and expectations, and contribute to the

intellectual and social development of children. The abstract

and intangible nature of social capital means that in the

present study of college attendance, social capital is treated

as an unmeasured concept. Measured instead are more tangible

indicators of the necessary conditions for the presence of

family social capital: the physical structure of the family

and the process of interaction. It is assumed that where the

structural and processual prerequisites for the existence of

social capital are present, there is a certain unknown but

positive likelihood of the presence of social capital. Such

an approach is not as neat and clean as simply defining social

capital as a certain kind of structure of relations as well as

a certain sort of interpersonal interaction, but makes up for

these shortcomings by virtue of an enhanced theoretical and

logical rigor.

To venture, then, a definition of social capital: Social

capital is an intangible quality of relationships in families

and communities which constrains and facilitates purposive

individual behavior in a manner consistent with the interests

of both individuals and the social structure, that is, higher

educational attainment. Social capital is invested in

relationships through the process of interpersonal

interaction. Frequency, duration, density and opportunities

for interpersonal interaction are determined by the structure


of relations in the family, the school and the community.

This definition is abstract enough to apply to a the broad

domain of educational attainment, and, ultimately, status

attainment, while addressing the dual nature of social capital

as a resource or benefit for the purposive individual and also

acting as a structural variable enabling and constraining


Bringing Status Attainment Back In

FuUwaf Odue@to

Mothers edusaiolo

Mtihrs oedupathmn

Famthi ocIupmton
r-niiy IOOMHHI


momlal ompital

ent *,r slal eMapital

*oelal pital

Figure 2-5
Modified status attainment model incorporating and emphasizing

elements of the social capital model

The value of the Coleman social capital framework is not

so much as a theoretical alternative to the status attainment


MOph epbow

elements---[ of.hes........a m de


approach but more as an innovative extension of the status

attainment model. The Coleman framework provides an

elaboration of the influence of the "significant other" in a

manner that takes account of both structure and social

interaction in the family as well as in the community. With

some modification, Coleman's social capital scheme contributes

unifying themes to the status attainment analytical framework.

Superimposing the social capital model over the status

attainment model clarifies the role of social capital

variables in mediating the influence of family background

variables on educational attainment, as illustrated in Figure


This model is not so very different from the model

developed by Coleman and Hoffer (1986) to account for

droppingout behavior among high school students. Coleman and

Hoffer's model includes several measures of family background,

a composite index of socioeconomic status -- family income,

father's education and occupation, mother's education, and

household possessions -- and indicators of race and ethnicity.

Number of siblings, both parents in the household, and mother

worked while child was young are included as indicators of

family structure. Mother's expectations for college, and talk

with parents are employed as indicators of family process or

interaction (function). Number of moves since grade 5 is a

proxy measure of community social capital, inasmuch as a

student who changes schools because of family moves breaks the


social linkages in the school and community that develop

social capital.

Table 2-3
Dropout rates between spring, grade 10. and spring, grade 12,
for students whose families differ in social capital,
controlling for human capital and financial capital in the

Percentage Difference
Dropping in Percentage
Out Points

1. Parents' presence:
Two parents 13.1
Single parent 19.1 6.0
2. Additional children:
One sibling 10.8
Four siblings 17.2 6.4
3. Parents and children:
Two parents, one sibling 10.1
One parent, four siblings 22.6 12.5
4. Mother's expectation
for child's education:
Expectations of college 11.6
No expectation of college 20.2 8.6
5. Three factors together:
Two parents, one sibling,
mother expects college 8.1
One parent, four siblings,
no college expectations 30.6 22.5

SOURCE: Coleman, 1988a:S112.

Coleman (1988a) calculated the probabilities of dropping

out, for students whose families differ in social capital. As

indicated in Table 2-3, family structure and family

interaction combine to produce substantial differences in

dropout rates. Between students from families with two

parents, one sibling and in which the mother expects the

student to go to college, and students from families with one

Table 2-4
Predicted dropout rates in the South between Spring., Grade 10
and Spring, Grade 12. for students whose families and
communities differ in social capital, controlling for human
and financial capital.


Low 47.7% 11.9%

High 15.2 2.6%

SOURCE: Smith, Beaulieu, and Israel, 1992:84.

aHigh family social capital is defined as: (1) two parents
present; (2) one sibling; (3) mother did not work when child
was young; and (4) mother expects child to go to college. Low
family social capital is defined as: (1) one parent present;
(2) four siblings; (3) mother worked full-time when child was
young; and (4) mother has no expectations for college.
High community social capital is defined as: (1) child has
never changed schools since grade 5 because of a family move;
and (2) child participates actively in church activities. Low
community social capital is defined as: (1) child has changed
schools 3 or more times since grade 5 because of family moves;
and (2) child does not participate in church activities.

parent, four siblings, and no college expectations, Coleman

found a 22.5 percent difference in the probability of dropping

out. Emulating the Coleman model, but with an additional

measure of community social capital, church attendance, Smith

et al. (1992) found that community social capital exerts a

substantial separate effect on dropping out and that family

and community social capital combine to produce surprisingly

large effects, as can be seen in Table 2-4. When both

community and family social capital are low, the probability

of dropping out is 47.7 percent, odds of about 50-50. When


family social capital is high and community social capital is

low, or community social capital is high and family social

capital is low, the predicted probability of dropping out is

about the same, between 12 and 15 percent. This suggests that

high levels of either family or community social capital can

compensate to a large degree for low levels of the other.

However, when both community and family social capital are

high, students are virtually assured of graduating.

The Model Used in the Present Study

Table 2-5
Family background and social capital variables in the social
capital model used in the analysis of college attendance, with
examples of indicators

Family Background

Financial Capital
Family income

Family Human Capital
Father's Education
Mother's Education


Social Capital Variables

Family Structure
mother works
Family Process
parents encourage college
parents monitor activities
School Structure
public or private school
School Process
school social environment
Community Structure
community size (rurality)
Community Process
participation in community

The present analysis extends the work of Coleman in two

ways: First, the social capital model is applied to college

attendance instead of high school completion; and second, the

conceptualization is extended to reflect the logical

implications of the notion of "embeddedness" of the individual

in social structure.

Table 2-5 presents the family background variables and

social capital variables used in the analysis. While Coleman

used a composite SES measure, the model used here

disaggregates SES and looks separately at the effects of

family financial capital (family income) and family human

capital (educational level of the parents). Out of concern

for parsimony and possible multicollinearity, parents'

occupational status is not included in the model. Race and

gender are included as important control variables.

Though Coleman discussed the role of community social

capital, he made no systematic effort to measure its effects,

nor did he give attention to the effects of school structure

and school social interaction on educational attainment. In

the present study, utilizing the same dataset used by Coleman,

the social capital variables reflect the view of the

individual as progressively embedded, or nested, in the

family, the school, and the community. At each level of

social organization, the presence or absence of social capital

is seen as dependent upon structural and process variables.

At each level -- the family, the school, and the community --

the indicators of the necessary conditions for the presence of

social capital mediate the effects of family background

variables on the process of college attendance.

Social Capital Sbucfural Vardable

Social Capital Proces Varfab.le

Figure 2-5
Conceptual model of college attendance emphasizing
relationship between structure and process variables

Figure 2-5 graphically depicts the theoretical

relationships among the principal concepts of the social

capital model used in the present analysis. The arrows show

the logical causal direction of the relationships, though no

path coefficients will be estimated. The theoretically

important -- theory used here in the broad sense of a logical

arrangement of concepts -- family background variables are the

family economic resources and the parents educational level.

Both of these factors are seen as exerting direct influence on


college attendance, but also exert indirect effects through

family, school, and community structural and process

variables. Family background variables influence the nature

of the family structure as well as influencing the type of

school and community. The structural variables influence the

quality and frequency of the interaction represented by the

process variables.

The family, school, and community structural variables

are regarded as exerting separate effects on college

attendance through their intervening effects on family, school

and community process variables, thereby mediating the effects

of family income and parental education on college attendance.

While structural variables are seen as exerting substantial

influence over interaction in family, schools and communities,

as represented by the process variables, the process variables

are not completely determined by the structural variables.

Family, school and community process variables are viewed as

exercising independent effects on college attendance,

mediating the impact of family, school and community


Research Questions

Research Question: Do indicators of family, school and

community structure exercise significant

effects on college attendance net of

Family Background




Family Structure



Family Process


School Structure


School process



family background factors, i.e., family

income, and parental education?

Family income is positively associated

with college attendance.

Father attending college is positively

associated with child attending college.

Mother attending college is positively

associated with college attendance.

Both parents in the household is

positively associated with college


Mother working is negatively associated

with college attendance.

Parental interest in their children's

school work and encouragement of academic

attainment is positively associated with

college attendance.

School type is negatively related to

college attendance.

Student report of problems with high

school social interaction climate is


Community Structure



Community process


negatively associated with college


Participation in school extracurricular

activities is positively related to

college attendance.

Size of place is curvilinearly related to

college attendance, with suburban

students experiencing higher college

attendance than rural and urban students.

Low number of family moves is positively

associated with college attendance.

Church attendance is positively related

to college attendance.



Description of Secondary Data Set

Answers to the research questions are sought through the

use of the High School and Beyond (HS&B) data set. The High

School and Beyond longitudinal study was conducted by the

National Opinion Research Center on contract with the National

Center for Education Statistics, U.S. Department of Education.

The base-year survey, conducted in the Spring of 1980,

involved a two-stage, national probability sample. In the

first stage, 1,122 schools were selected for the sample out of

a sampling frame of 24,725 (NCES, 1991). Certain strata of

schools were oversampled to facilitate analysis of sub-

populations of interest, for example, alternative public,

Hispanic Catholic and public, high performance private, and

other non-Catholic private schools. In the second stage, 36

sophomores and 36 seniors were randomly selected from each

school. Ultimately, over 58,000 sophomores and seniors from

1,015 public and private high schools took part in the study

(84 percent of the eligible sample).

Each student completed a set of questionnaires that were

designed to elicit information on individual/family background

characteristics, high school experiences, work experiences,



and future plans. Students in both cohorts also took a series

of timed cognitive tests. The sophomore and senior

questionnaires shared about 75 percent of questions in common.

Questionnaires were administered on school premises by NORC

survey representatives (NCES, 1991). Base year survey

instruments included: (1) a sophomore questionnaire, (2) a

senior questionnaire (3) cognitive tests for both sophomores

and seniors (4) a school questionnaire, (5) teacher

questionnaire, and (6) a parent questionnaire.

A follow-up study conducted during the early part of 1982

was targeted to all 1980 sophomores (now seniors) who

participated in the 1980 survey. The intent of the follow-up

was to continue documentation of the secondary school

experiences of high school students. For persons who remained

in school, a near-duplicate version of the survey instrument

administered two years earlier was employed. As in the case

of the 1980 survey procedures, in-school group administration

of survey instruments took place (NCES, 1991).

The 1984 and final 1986 waves focused attention on post-

high school experiences: college attendance, work experience,

marriage, and future plans. For both 1984 and 1986, a sample

of 15,000 participants in the 1982 wave were mailed

questionnaire packets. Follow-up by telephone interview and

personal visits yielded completion rates of 91 percent in 1984

and 90.6% in 1986 (NCES, 1991).


Other High School and Beyond files include the High

School Transcript File, which includes information on all

courses taken during the four years of high school, grade

point averages, standardized test scores, and days absent.

Course offerings and enrollment data for 957 schools are

available in the Offerings and Enrollment File. The HS&B HEGIS

and PSVD File contains data on post-secondary educational

institutions attended by HS&B respondents. This data includes

type of institution, degrees offered, admission requirements,

enrollment, and tuition (NCES, 1991).

Because certain strata -- such as racial and ethnic

minorities -- are oversampled for policy-relevant reasons,

weights are calculated for each wave. Weights are calculated

as the inverse of the probability of selection in a survey,

adjusted to compensate for "unit" nonresponse, such as failure

to complete a whole questionnaire or combination of data

elements (NCES, 1991). The inclusion of relevant sophomore

cohort weights in the variable list "projects" to the

population of 3,781,000 high school sophomores in 1980 (NCES,


Measurement of Variables

Table 3-1 outlines the variables to be examined and

specifies the manner in which they are to be measured. Coding

for several variables follows Coleman's coding approach,

though in some cases a somewhat different coding scheme is


Table 3-1
Variables used in the analysis, and their measurement.


Coding Scheme

Dependent Variable

College attendance

1= attended 2-year or 4-year
college by Feb. 1984; 0= did
not attend college by Feb. 1984

Independent Variables

Family Financial Resources

Family Income

4=less than $8,000; 11.5=8,000
-$14,000; 17.5=$15,000-19,999;
22.5=$20,000-24,999; 27.5=
$25,000-$29,999; 35.0=$30,000-
$39,999; 45.0=$40,000-$49,000;
50.0=$50,000 or more; 0=missing




Family Human Capital Resources

l=father attended college;
0=high school graduate or less

1=missing on FATHERED

l=mother attended college;
0=high school graduate or less

l=missing on MOTHERED


Father's Education

FEDX (BB038)

Mother's Education

MEDX (BB039)

Background Control Variables

Race (contrasts with white)

1= female; 0= male


Table 3-1, continued.

Family Social Capital

Family Structure

Both Parents

Table 4-1. continued.

1= both parents in household; 0= one
parent in household

Mother Worked

l=mother worked full-time while
respondent was in high school;
0=mother did not work full-time
while respondent was in high school

Family Process Variables

Parental Involvement

BB046A-B; BB046C)

Additive scale constructed by
summing: 1= mother expects college
for respondent; 0= mother does not
expect college; 1= father expects
college for respondent; 0= does not
expect college; 1= father monitors
respondents's school work; 0= father
does not monitor school work; 1=
mother monitors school work; 0=
mother does not monitor school work;
1= "true" response to statement, "my
parents always know what I'm doing";
0= "false" response

School Social Capital

School Structural Variables

Type of High School

l=private school, religious or
otherwise; 0=public school

School Process Variables

School Environment
INTERSCH (YB019A-YB019F) Additive scale constructed by
assigning 2 points for response,
"Often happens", and 1 point for
"Sometimes happens", for following
measures: students do not attend

Table 3-1, continued.

school; students talk back to
teachers; students do not obey
instructions; students fight with
each other; students attack teachers

Community Social Capital
Community Structural Variables

Size of Place (contrasts with suburban)
SCHURB1 l=urban;
SCHURB2 l=rural;
(SCHURB) 0=suburban

Table 4-1. continued.

Number of moves
since grade 5 l=respondent has changed schools 2 or 3
MOVES (BB011) times since grade 5 due to family moves;
0=respondent has changed school 1 time or
less due to family moves since grade
schools since starting 5th grade

Church Attendance

1= respondent attends church;
respondent does not attend church

Note: Variable names are given in all-capital letters.
Original tape-names are given in parentheses if different.

Discussion of the Variables and Coding Scheme

The choice of variables and coding scheme represents an

extension and refinement of the model used by Coleman and his

associates. The present study retains several of the social

capital variables used by Coleman, such as number of parents

in the household and mothers' work status, though the coding

differs to some extent. Additional measures have been

included to reflect the parallel conceptual arrangement of


structural and processual social capital precursors at the

level of the family, the school and the community.

Disaggregation of SES

An innovation of the study is that of disaggregating the

socioeconomic status variable (SES). Typically, researchers

using the High School and Beyond data set utilize a composite

SES measure made up of five components, including father's

occupation, father's education, mother's education, family

income, and scale of eight household items. Such indexes are

useful, but can result in uncertainties in interpretation,

since three or more dimensions -- education, income,

occupation, and possessions -- are treated as a single

dimension --socioeconomic status. Coleman drew the conceptual

distinction between family financial capital and family human

capital, but used the composite measure in his model.

Disaggregation of the measure will allow for the examination

of the effects of family human capital (FATHERED and MOTHERED)

separate from family financial capital (INCOME). FATHERED and

MOTHERED are both coded as either attending college or not

attending college. Approximately 25 percent of respondents

declined to answer the INCOME question, and about 16 percent

indicated that they did not know how much education their

parents had. In order to retain these cases in the data set,

dummy variables -- FAMINCX, FEDX, and MEDX -- were coded as 1

when cases were missing on INCOME, FATHERED and MOTHERED,

respectively, and were coded 0 otherwise.

Family Structure and Family Process Variables

Family structure variables are consistent with those used

by Hansen and McIntyre (1989) and Coleman (1988), with the

exception that the mother work measure in Coleman's research

looked at mother working prior to elementary school, whereas

the present study examines the effects of mother working

during elementary school and during high school (MAWORK). The

family structure variables -- mother working, number of

parents in household (BOTHPAR) -- determine the opportunity

structure for interactions within the family. Mother working

means she may have fewer opportunities to interact, while

single parent household likely means fewer opportunities for

parent-child interaction.

While family structure indicators attempt to measure

opportunities for family social interaction, the family

process indicator is intended as a measure of family

interaction itself, particularly parent-child interaction that

encourages college attendance or otherwise reinforces norms

and values of achievement. Several questionnaire items were

combined into an additive scale to measure family process

(INTERPAR): father and mother monitor school work, parents

keep track of what the respondent is doing, father and mother

expect respondent to go to college. Social capital measures

are not indicators of social capital as such. Measures of the

structure of family relations and the process of family


interaction are treated as indicators of the precursors of

social capital.

School Structure and School Process Variables

Type of high school -- public or private high school

(HSTYPE) -- is included in the model as an indicator of school

structure. Coleman and Hoffer (1987) found that students in

private schools had lower dropout rates. This was especially

true for Catholic high schools, but was also the case for

other religious-oriented high school. The effect is

attributed by Coleman and Hoffer, not to the religious content

of the schooling, but to the functional community they argue

envelopes the school which provides opportunities for

intergenerational closure. The present inquiry incorporates

type of school in order to ascertain whether that school

effect carries over from dropping out to college attendance.

The measure of school social process is INTERSCH, an additive

scale constructed by summing reports of a number of school

problems such as students fighting in school, students

threatening teachers and chronic absenteeism.

Community Structure and Community Process

Community social capital variables represent the degree

to which students are socially integrated into the community

as well as the structure of opportunities for social

interaction. Number of moves and size of place are

incorporated into the model of college attendance as measures

of community structure. Number of moves since grade 5

reflects the student's access to community social capital.

Students who move from school to school do not stay in a

community long enough to become integrated into the social

structure by establishing long-term relationships. Size of

place in which the school is located is included because human

and financial resources are most likely to be constrained in

certain places such as inner cities and rural areas. People

living in rural counties are more likely to suffer from

poverty, unemployment and underemployment, and are less likely

to have a high level of education (Beaulieu 1988; O'Hare 1988;

Reid 1989; Tweeten 1988). Rural students are more likely to

report that they would be satisfied with less education than

suburban or urban students, as well as reporting lower levels

of aspirations to professional occupations and higher levels

of aspirations to achieve lower-status occupations such as

craftsperson or service occupation (Hansen and MacIntyre,

1989). Differences in the density of population and the

spatial dispersion of residential and business activities

might plausibly be seen as leading to different patterns of

social interaction and hence development of social capital.

Since Durkheim's Suicide, church attendance has been used

as in indicator of social integration. As noted by Coleman,

churches are 'multiplex' organizations which serve functions

other than those originally intended, such as providing

opportunities for intergenerational closure. Church

attendance, then, is indicative of the youth's integration

into the community's social structure.

Statistical Analysis

Because the dependent variable is the dichotomous

question of college attendance or non-attendance, an

appropriate statistical procedure appears to be use of

logistic regression (Alexander et al., 1987). Multiple

logistic regression is considered to be the preferred method

for estimating, for a given individual, the probability of a

certain event occurring, in this case attending college. A

major advantage of logistic regression is that the independent

variables can be categorical, ordinal, interval, or a

combination of all three. The parameter estimates obtained

through multiple logistic regression are similar in

interpretation to those of multiple linear regression. The

overall fit of the model is indicated by the model chi-square,

degrees of freedom, and probability levels. A large model chi-

square and small probability level indicate that the model is

a significant improvement over an intercept-only model.

The research questions are to be answered by estimating

probabilities based on selected levels of the independent

variables in the fitted model. These probabilities are

derived from the logits (log odds) of attending college given

one or more characteristics, while controlling for the effects

of other factors. The latter are set at their mean or most


common value in calculating the logits. The form of the

equation used to calculate the probability of attending

college at different levels of the independent variables is as


probability (pi) = ea+blx1+...+bkxk / [1 + ea+blx+...+bkxk]

Alternately, the equation can be written in this manner:

log(pi/l pi) = a + blxl + ... + bkxk

Taking the antilog of both sides of the equation yields the

following equation:

pi/l pi = ea+b1x1 ...+ bkxk

The implications of the exponential form of the right side of

the equation is that each x level of the independent variables

has a multiplicative effect of eb on the probability of

falling into the 1 category of the dichotomous dependent

variable, in this case attending college (Agresti and Finlay,


The model is analyzed using SAS's logist procedure on

mainframe computer. Participation flags are selected for each

wave and the appropriate weight variable is included in order

to correct for oversampling in certain strata -- such as race

and school type -- and for non-response bias. Use of

participation flags includes in the dataset only those 1980

sophomores who participated in each of the subsequent waves.

This stipulation resulted in the reduction of the sample from

14,825 to 8,062 cases. Of these 8,062 cases, 5225 respondents

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E6R74V6KG_JACY8W INGEST_TIME 2012-08-30T16:09:52Z PACKAGE AA00011773_00001