Академический Документы
Профессиональный Документы
Культура Документы
STUDENT PERSISTENCE
https://mail.go
Table of Contents
ExecuHve
Summary
Project Summary
Background
Process
11
11
12
12
Social capital
12
13
13
Summary
14
Methodology
17
Sample
17
Data Sources
17
Variables
18
Data analysis
20
Results
23
Analysis 1 Results
23
Analysis 2 Results
24
Discussion
29
Conclusions
33
34
Endnotes
35
References
37
ii
Executive Summary
Project
Summary
This
study
was
part
of
a
mixed-methods
research
project,
Socioeconomic
StraAcaAon
of
EducaAonal
AKainment
in
Puerto
Rico
conducted
by
researchers
from
the
Center
for
Applied
Social
Research
at
the
University
of
Puerto
Rico-Mayaguez
(UPRM)
from
January
2009
to
March
2010.
The
project
was
funded
by
the
Puerto
Rico
Higher
EducaAon
Council
(CESPR).
Socioeconomic
factors
are
well-known
to
play
a
role
in
individual
educaAonal
outcomes
in
the
research
literature,
but
empirical
evidence
is
lacking
to
substanAate
this
relaAonship
in
Puerto
Rico.
The
study
used
exisAng
data
from
insAtuAonal
databases
at
UPRM
to
examine
how
socioeconomic
factors
inuence
persistence
in
college
in
Puerto
Rico.
The
n-
dings
of
the
study
and
its
recommendaAons
will
benet
higher
educaAon,
community,
and
non-prot
leaders
in
Puerto
Rico
as
they
aim
to
ensure
all
social
sectors
of
Puerto
Rico
have
access
to
higher
edu-
caAon
and
enjoy
the
social
and
economic
benets
of
advanced
educaAonal
preparaAon.
Background
Since
2004,
the
Center
for
Research
and
DocumentaAon
on
Higher
EducaAon
in
Puerto
Rico
(CEDESP)
of
the
Puerto
Rico
Higher
EducaAon
Council
(CESPR)
has
awarded
$993,000
toward
26
research
projects
on
higher
educaAon
in
Puerto
Rico.
The
current
project
is
the
rst
of
its
kind
for
CESPR,
both
for
its
broad
focus
on
socioeconomic
factors
and
educaAonal
aKainment,
and
for
the
size
of
the
award
($76,000).
The
project
is
also
the
rst
publicly
funded,
mulA-method
research
study
in
Puerto
Rico
to
empirically
examine
the
links
between
socioeconomic
status,
geographic
locaAon,
and
educaAonal
outcomes.
The
principal
invesAgator
for
the
study
is
a
faculty
member
and
researcher
at
the
University
of
Puerto
Rico-Mayaguezs
Center
for
Applied
Social
Research
(CISA).
The
funds
allocated
by
CESPR
provided
relea-
se
Ame
for
the
researcher,
wages
for
undergraduate
research
assistants,
and
materials
and
supplies.
The
administraAve
assistant
for
the
Center
for
University
Access
at
UPRM
provided
administraAve
support.
The
Carvajal
Project
for
the
DemocraAzaAon
of
Knowledge
supported
the
wriAng
and
ediAng
of
the
re-
sults
as
part
of
its
Working
Paper
series.
Process
This
project
included
four
separate
but
related
studies
that
provide
empirical
evidence
for
the
inuence
of
socioeconomic
factors
on
overall
levels
of
educaAonal
aKainment
in
Puerto
Rico.
The
bulk
of
the
research
for
this
study
was
conducted
between
August
and
December
2009,
based
on
the
availability
of
data
and
the
schedule
of
the
researcher.
Data
from
admissions,
enrollment,
and
economic
assistance
databases
at
the
University
of
Puerto
Rico-
Mayaguez
(UPRM)
were
compiled
with
assistance
from
personnel
from
the
UPR
Oce
of
the
Vice-
President
of
Student
Aairs,
the
UPRM
InformaAon
Technology
Center,
and
the
UPRM
Oce
of
InsAtuAonal
Research
and
Planning
to
study
how
socioeconomic
factors
predict
retenAon
in
college.
Two
separate
samples
of
rst
year,
rst
Ame
degree
seeking
students
were
created,
based
on
availability
of
valid
and
reliable
data
on
academic
preparaAon,
family
structure,
and
parent
educaAon
and
income.
High
income
students
whose
parents
aKended
college
were
signicantly
more
likely
to
persist
in
college
than
low
income/rst
generaAon
students.
There
were
some
gender-based
dierences
in
academic
preparaAon
and
socioeconomic
factors
that
predict
retenAon
among
students
who
apply
for
nancial
aid.
Young
mens
persistence
appears
to
be
related
to
academic
preparaAon
and
public
school
aKendance,
while
young
womens
persistence
is
related
to
high
school
GPA,
parents
marital
status,
and
public
school
aKendance.
These
ndings
provide
empirical
evidence
for
educaAonal
aKainment
inequiAes
in
Puerto
Rico.
Even
among
those
who
enroll
in
the
most
selecAve
higher
educaAon
insAtuAon
on
the
island
University
of
Puerto
Rico
-
socioeconomic
and
structural
factors
play
a
role
in
whether
they
conAnue
their
studies
beyond
the
rst
year.
the
Integrated
Postsecondary
Data
System
(IPEDS)
these
data
could
be
provided
to
CESPR,
without
personal
or
insAtuAonally
idenAfying
informaAon.
2. All
students
must
have
access
to
a
college
preparaHon
curriculum
in
middle
and
high
school.
The
ndings
of
this
study
clearly
show
that
previous
academic
preparaAon
is
criAcal
for
students
success
in
college.
Policies
that
track
certain
students
into
less
challenging
curricula
at
an
early
age
discourage
further
study,
and
in
gaps
in
student
knowledge
which
become
apparent
when
students
enroll
in
university-level
classes.
While
not
all
students
will
want
to
aKend
university,
the
fact
that
they
are
prepared
to
do
so
would
be
an
advantage
for
them,
individually,
in
the
labor
market,
and
for
Puerto
Rico
as
a
naAon,
by
having
a
well-educated
public.
3. All
students
in
high
school
must
have
access
to
preparaHon
for
College
Board
tests,
as
well
as
equal
opportuniHes
to
take
the
tests.
Given
the
importance
of
standardized
test
scores
for
student
success
in
college,
as
shown
in
the
ndings
of
this
study,
access
to
test
preparaAon
and
opportunity
to
parAcipate
in
the
test
is
imperaAve
to
increasing
socioeconomic
diversity
in
the
public
university.
The
College
Access
Challenge
Grant
project
as
the
University
of
Puerto
Rico
has
made
important
strides
in
this
direcAon,
by
subsidizing
the
cost
of
taking
the
test
for
all
grade
12
students
in
2009.
4. Higher
educaHon
insHtuHons
must
develop
or
strengthen
eorts
to
retain
lower
SES
students.
Using
insAtuAonal
research
infrastructure,
insAtuAons
should
focus
on
studying
which
students
are
more
likely
to
leave
aper
their
rst
year,
along
with
the
specic
needs
of
those
students
for
extended
orientaAon
or
academic
support.
Then,
resources
must
be
allocated
(or
re-allocated)
to
eorts
that
will
promote
the
persistence
of
those
students
most
likely
to
leave.
Ample
evi-
dence
on
best
pracAces
is
available
in
the
literature
but
it
is
important
to
examine
what
will
work
in
a
parAcular
insAtuAonal
context.
5. Further
research
is
necessary
to
understand
the
barriers
to
college
retenHon
among
low
in-
come
students
in
Puerto
Rico.
Evidence
from
UPRM
suggests
that
the
aKriAon
rates
for
low
in-
come
students
are
twice
as
high
as
their
higher
income
peers;
and
the
ndings
in
the
current
project
showed
that
parent
income
and
educaAon
level
play
a
key
role
in
predicAng
retenAon.
Further
research
is
needed
to
understand
what
barriers
these
students
face,
in
contrast
to
higher
income
peers;
qualitaAve
research
should
play
an
important
role.
Conceptual
frameworks
includ-
ing
social
and
cultural
capital
would
also
help
to
provide
more
theoreAcal
tools
to
understand
educaAonal
inequity
in
PR.
ver,
there
is
limited
published
research
in
Puerto
Rico
that
looks
at
socioeconomic
factors
related
to
per-
sistence
and
performance
in
college.
This
study
addresses
this
gap
through
invesAgaAon
of
the
predicAve
ability
of
academic
preparaAon
and
social
structural
factors
on
retenAon
using
two
dierent
data
sets
compiled
from
the
University
of
Puerto
Rico-Mayaguez
data
bases.
The
analyses
involved
modeling
aca-
demic
preparaAon
and
socioeconomic
factors
as
predictors
of
retenAon.
Slightly
dierent
family
and
school
socioeconomic
measures
are
used
in
each
of
the
two
analyses,
to
ex-
plore
their
uAlity
and
appropriateness.
SelecAon
and
operaAonalizaAon
of
these
variables
was
guided
by
the
ndings
of
the
literature
review.
In
the
rst
analysis,
socioeconomic
status
was
measured
as
the
stu-
dent
belonging
to
one
of
four
family
income-educaAon
groups
based
on
rst
generaAon
status
and
in-
come
category
(low
vs
high).
Based
on
the
known
importance
of
school
structure
variables,
socioecono-
mic
and
achievement
variables
from
the
students
high
school
of
origin
were
also
included.
School
com-
posiAon
variables
comprised
average
English
achievement
and
proporAon
rst
generaAon,
among
stu-
dents
admiKed
to
UPRM.
We
were
unable
to
access
any
neighborhood
level
variables
due
to
the
limited
informaAon
available
in
the
UPRM
database
and
the
limited
Ame
frame
for
compleAon
of
this
study1.
In
the
second
analysis,
family
structural
variables
available
from
nancial
aid
applicaAon
data
(size,
parent
marital
status,
teen
mother),
rst
generaAon
status,
and
parent
income
were
included
as
socioeconomic
background
factors,
while
school
type
(public
vs
private)
was
included
as
a
measure
of
school
socioeco-
nomic
composiAon.
Using
the
Census
and
GIS
database
recently
created
by
CISA
researchers
for
the
larger
project,
in
future
analyses
we
should
be
able
to
add
neighborhood
level
socioeconomic
variables
to
allow
modeling
of
neighborhood
eects.
6
10
11
Social capital
During
the
late
1990s
and
early
2000s,
research
on
educaAonal
aKainment
began
to
incorporate
family
social
capital
variables
as
possible
predictors
of
high
school
graduaAon
and
college
aKainment,
drawing
primarily
from
the
data
available
from
the
NaAonal
EducaAonal
Longitudinal
Study
of
1988
(NELS:88)
and
other
panel
studies
(see
Dika
&
Singh,
2002
for
an
extensive
review).
Results
of
analyses
of
NELS:88
show
that
parental
expectaAons
and
aspiraAons,
parent-teen
communicaAon
and
parent
monitoring,
number
of
moves
(residenAal
stability),
parent
communicaAon
with
school,
and
intergeneraAonal
closu-
re
predict
high
school
compleAon
(Carbonaro,
1998;
Israel
et
al.,
2001;
Muller
&
Ellison,
2001;
Teachman,
Paasch,
&
Carver,
1996;
Yan,
1999).
Other
longitudinal
studies
evidence
the
importance
of
intergeneraAonal
closure,
parents
social
network,
parent
monitoring,
number
of
friends
known
by
pa-
rent,
and
parent
involvement
in
the
school
in
high
school
graduaAon
and
college
enrollment
(Fursten-
12
berg
&
Hughes,
1995;
White
&
Glick,
2000).
Moving
is
negaAvely
related
to
these
two
outcomes
(Hoerth
Boisjoly,
&
Duncan,
1998).
Family
discussion,
parents
inuence
and
expectaAons,
parent-
school
involvement,
and
parent
monitoring
have
also
been
found
to
predict
years
of
schooling
(Dyk
&
Wilson,
1999;
Kalmijn
&
Kraaykamp,
1996;
Lpez,
1996).
13
use,
studying,
test
taking,
Ame
management);
(f)
knowledge
of
the
labor
and
educaAonal
markets,
and
(g)
problem-solving
knowledge.
Stanton-Salazar
also
idenAes
the
forms
of
insAtuAonal
support
that
facilitate
the
acAvaAon
process,
including
implicit
and
explicit
socializaAon
into
insAtuAonal
discourses,
bridging
connecAons
to
gatekeepers
and
other
social
networks,
advocacy,
modeling
eecAve
coping
with
straAcaAon
forces
(help-seeking,
problem
solving),
emoAonal
and
moral
support,
and
personalized
ad-
vice
and
guidance.
Horvat
(2003)
discusses
the
uAlity
of
Bourdieus
concept
of
habitus
to
understand
how
educaAonal
expe-
riences
and
outcomes
are
inuences
by
structural
factors
namely,
race
and
class.
Horvat
presents
an
example
from
a
longitudinal
study
of
African
American
women
from
college
applicaAon
through
gradua-
Aon.
She
asserts
that
race
and
class
interact
to
shape
how
these
women
interpret
educaAonal
opportu-
nity,
and
that
the
meaning
of
race
and
class
shiped
for
the
students
as
they
moved
through
dierent
elds
of
interacAon
high
school
through
college.
Lews
(2006)
research
on
Korean-American
dropouts
draws
on
parAcipant
observaAon,
interviews,
sur-
vey
and
document
analysis.
He
looked
at
how
SES
background,
social
capital,
and
school
resources
aect
aspiraAons
and
aKainment
of
Korean
American
youth
in
New
York
City.
Lew
references
the
work
of
Stanton-Salazar
(2001),
and
notes
how
economic,
linguisAc,
and
cultural
barriers
limit
access
to
insAtu-
Aonal
support
and
accumulaAon
of
social
capital
for
low
income
Korean
Americans.
He
also
explores
the
ineecAve
learning
environment
and
lack
of
caring
relaAonships
with
school
personnel
as
barriers
to
conAnuing
in
school.
Lew
found
that
these
young
people
were
highly
aware
of
their
otherness
as
mi-
nority
and
low
income
status.
Lew
suggests
that
variability
within
ethnic
groups
must
be
criAcally
exami-
ned
to
understand
barriers
to
developing
and
accessing
resources
in
social
networks.
Summary
The
research
evidence
overwhelmingly
conrms
that
social
structural
factors
restrict
and
facilitate
edu-
caAonal
aKainment.
Research
on
family
structures
and
social
capital
generally
demonstrates
posiAve
links
between
higher
SES
and
educaAonal
aKainment.
Most
of
the
studies
uAlizing
social
capital
have
conceptual
and
methodological
limitaAons,
including
conceptual
validity.
Studies
on
school
and
neighborhood
structural
factors
suggest
that
socioeconomic
segregaAon
and
educaAonal
aKainment
gaps
may
perpetuate
one
another,
and
that
gaps
between
low
income
and
high
income
groups
will
con-
Anue
to
widen.
The
ndings
of
the
review
have
implicaAons
for
the
design
of
research
studies
on
the
eects
of
SES
on
aKainment:
Parent
educaAon
and
income
are
important
variables
that
must
be
modeled
as
part
of
students
so-
cioeconomic
status
(SES).
Social
capital
variables
such
as
interacAon
with
parents
or
with
educaAonal
agents
may
also
help
explain
educaAonal
aKainment;
but
some
researchers
have
criAcized
the
measurement
of
social
capi-
tal
in
educaAonal
research,
ciAng
that
it
primarily
acts
as
a
proxy
for
SES
(see
Dika
&
Singh,
2002
for
a
discussion).
14
15
16
Methodology
Sample
This
study
involves
the
use
of
two
dierent
samples
of
student
data
from
UPRM.
The
samples
were
created
by
the
researchers
from
exisAng
data
based
on
availability
of
valid
and
reliable
data.
While
they
overlap
somewhat,
together
they
represent
an
eleven
year
Ame
span
of
UPRM
entering
freshmen
between
1998
to
2009.
The
sample
for
the
rst
analysis
was
composed
of
the
records
of
5,987
undergraduate
students
selected
from
the
enAre
populaAon
of
rst-Ame
undergraduate
students
entering
UPRM
between
1998
and
2006
(N=20,151).
The
criterion
for
inclusion
in
the
sample
was
based
on
two
extremes
of
family
income:
from
below
$7,499
to
$14,999,
and
$40,000
or
higher.
These
ranges
represent
the
rst
three
and
last
two
income
ranges
on
the
UPR
admissions
form.
The
categories
were
labeled
low-income
and
high-income,
roughly
equivalent
to
the
two
lower
quinAles
and
the
top
quinAle
of
median
family
incomes
in
Puerto
Rico,
based
on
the
2007
American
Community
Survey.
The
sample
for
the
second
analysis
included
records
of
rst-year,
rst
Ame
degree
seeking
students
who
applied
for
nancial
aid
to
study
at
UPRM
between
2005
and
2009
(n=7,006).
Student
records
were
included
in
the
sample
if
they
were
admiKed
to
and
enrolled
in
UPRM.
Data
Sources
This
study
uAlizes
admissions
and
enrollment
data
available
in
the
student
informaAon
system
at
the
insAtuAon.
There
are
a
limited
number
of
socioeconomic
variables
available
in
these
databases;
essenAally,
parent
income
and
educaAon;
based
on
the
informaAon
that
students
provide
during
the
admissions
process.
The
income
measure
is
truncated
at
the
higher
end,
as
the
nal
category
includes
incomes
of
$50,000
or
higher.
We
obtained
data
from
the
nancial
aid
applicaAon
process
(FAFSA)
for
the
second
set
of
analyses,
specically
to
allow
us
to
model
social
structure
and
socioeconomic
variables
that
are
not
available
in
the
admissions
and
enrollment
data;
for
example,
family
size,
parent
marital
status,
and
exact
family
income;
which
have
shown
to
be
important
predictors
of
aKainment
in
previous
research.
Personnel
from
the
UPR
Oce
of
the
Vice-President
of
Student
Aairs,
the
UPRM
InformaAon
Technology
Center,
and
the
UPRM
Oce
of
InsAtuAonal
Research
and
Planning
were
instrumental
in
obtaining
and
preparing
the
data
from
various
databases.
The
data
les
were
provided
in
Excel
format,
and
merged
les
containing
the
variables
of
interest
were
created.
These
les
were
exported
to
Minitab
(Analysis
1)
and
SPSS
(Analysis
2)
to
carry
out
the
descripAve
and
staAsAcal
analyses.
17
Variables
All
variables
used
in
the
analyses,
along
with
names,
descripAons,
and
descripAve
staAsAcs,
are
shown
in
Tables
1
(Analysis
1)
and
2
(Analysis
2).
Previous
academic
achievement.
Students
scores
on
standardized
university
admission
tests
and
high
school
GPA
are
used
as
measures
of
previous
academic
achievement.
The
achievement
and
apAtude
tests
(Prueba
de
Evaluacin
y
Admisin
Universitaria,
or
PEAU)
were
developed
by
the
College
Board
for
Puerto
Rico,
and
have
the
same
scoring
as
the
SAT
(range
200
to
800).
English
achievement
and
mathemaAcs
apAtude
are
included
in
Analysis
1,
along
with
Spanish
verbal
apAtude
in
Analysis
2.
Both
analyses
included
measures
of
high
school
GPA
(four
point
scale).
These
academic
achievement
variables
were
chosen
because
they
represent
the
components
of
the
admission
index
for
the
University
of
Puerto
Rico.
Individual
socioeconomic
factors.
In
Analysis
1,
the
variable
family
income-educaAon
group
was
created
by
combining
a
students
income
status
(low=
below
$7,499
to
$14,999
and
high=$40,000
or
higher)
with
rst
generaAon
status.
While
the
rst
and
last
categories
of
the
admissions
income
scale
(below
$7,499
and
$50,000
or
higher)
roughly
represent
the
top
and
boKom
quinAles
of
median
income
in
Puerto
Rico,
we
included
addiAonal
categories
on
either
end
to
have
a
sucient
number
of
cases
for
analysis.
While
rst
generaAon
status
is
open
dened
as
neither
parent
having
completed
a
four-year
degree,
we
dened
rst
generaAon
in
Analysis
1
as
students
for
whom
neither
parent
had
aKained
any
more
years
of
schooling
beyond
a
high
school
degree.
This
deniAon
was
considered
to
be
more
appropriate
for
the
Puerto
Rican
context,
as
any
years
of
schooling
beyond
high
school
generally
represent
a
signicant
dierence
in
income
and
occupaAonal
opAons.
First
generaAon
status
was
computed
by
recoding
mothers
and
fathers
educaAon
variables,
then
compuAng
a
single
parent
educaAon
variable.
In
Analysis
2,
parents
gross
income,
number
of
family
members,
parents
marital
status
(1=married,
0=other),
whether
mother
was
teen
mother
(19
or
younger),
and
rst
generaAon
status
were
all
included
as
structural
factors.
In
the
second
analysis,
the
more
tradiAonal
deniAon
of
rst
generaAon
was
used;
that
is,
students
were
dened
as
rst
generaAon
if
neither
parent
had
completed
a
bachelors
degree.
In
both
analyses,
students
who
are
not
rst
generaAon
are
called
conAnuing
generaAon.
School
composiHon.
For
Analysis
1,
the
average
English
achievement
score
(school
level
English
achievement)
and
the
proporAon
of
rst
generaAon
students
(school
level
rst
generaAon)
at
the
students
high
school
of
origin
were
included
as
measure
of
school
composiAon.
The
esAmaAon
of
these
school
level
values
is
limited
because
it
could
only
be
calculated
for
students
admiKed
to
UPRM.
While
these
variables
do
not
provide
an
accurate
representaAon
of
the
composiAon
of
the
students
schools
of
origin,
we
can
use
them
in
a
cauAonary
way
to
invesAgate
the
relevance
of
school
composiAon
to
retenAon.
In
Analysis
2,
school
type
(public
vs
private)
was
the
sole
measure
of
school
composiAon.
RetenHon.
The
dependent
variable
for
both
of
these
analyses
is
retenAon,
measured
as
a
dichotomous
variable
(0=student
did
not
re-enroll
in
rst
semester
of
second
year;
1=student
re-enrolled
in
rst
semester
of
second
year).
18
Table
1.
DescripAons
of
previous
academic
achievement,
individual
socioeconomic
factors,
and
school
composiAon
factors
modeled
in
the
predicAon
of
retenAon
in
Analysis
1
Variable
High
school
GPA
Mean
SD
3.56
0.40
594.28
103.96
645.97
87.11
--
--
585.77
55.20
0.26
0.17
19
Table
2.
DescripAons
of
previous
academic
achievement,
individual
socioeconomic
factors,
and
school
composiAon
factors
modeled
in
the
predicAon
of
retenAon
in
Analysis
2
Variable
High
school
GPA
Mean
3.67
SD
0.34
562.01
112.59
621.27
87.50
579.81
71.97
--
--
$30,074.16
$29,877.38
Parent
income
Number
of
family
members
3.85
1.38
Parents
married
(dummy
variable)
--
--
--
--
--
--
Teen
mother
(dummy
variable)
Data
analysis
DescripAve
staAsAcs,
including
correlaAons,
as
well
cross-tabulaAons
and
ANOVA
were
conducted
prior
to
conducAng
logisAc
regression
analyses,
to
determine
the
relaAons
among
variables
in
the
models
and
to
make
the
nal
selecAon
of
variables
to
include
in
the
models.
LogisAc
regression
was
used
because
the
dependent
variable,
retenAon,
is
dichotomous
(yes/no).
While
Microsop
Excel
was
used
to
compile
the
databases,
data
were
exported
to
other
programs
for
analysis.
The
staAsAcal
packages
Minitab
(Analysis
1)
and
SPSS
(Analysis
2)
were
used
to
run
all
analyses.
All
staAsAcal
tests
were
evaluated
at
the
=.05
level.
20
21
Results
Analysis
1
Results
The
descripAve
staAsAcs
(means,
standard
deviaAons)
for
the
variables
used
in
this
analysis
are
shown
in
Table
1.
The
correlaAons
between
the
variables
(see
Appendix
C)
were
all
posiAve
and
staAsAcally
signicant
at
the
p<.001
level.
The
strongest
correlaAon
among
the
individual
variables
was
between
English
achievement
and
math
apAtude
(r=.31).
Next,
a
series
of
ANOVAs
were
conducted
to
determine
if
dierences
in
academic
achievement
existed
among
the
four
family
income
and
educaAon
(FIE)
groups
(Appendix
C).
High
school
GPA
was
signicantly
higher
for
low
income
students,
regardless
of
parent
educaAon
level.
This
result
seems
to
reect
the
known
phenomenon
that
GPA
are
higher
for
public
vs
private
school
students
at
UPRM,
and
that
low
income
students
are
more
concentrated
in
public
schools
island-wide.
English
achievement
was
signicantly
higher
for
high
income
students,
regardless
of
parent
educaAon
level;
and
English
achievement
was
signicantly
higher
for
low
income/conAnuing
generaAon
students
than
for
low
income/rst
generaAon
students.
College
academic
achievement
(rst
year
GPA)
was
signicantly
higher
for
high
income/conAnuing
generaAon
students
compared
to
all
other
groups,
but
there
were
no
dierences
among
the
other
3
groups.
These
results
indicate
that
socioeconomic
factors
clearly
seem
to
be
a
factor
in
college
academic
achievement
for
this
sample.
The
logisAc
regression
model
is
shown
in
Table
3.
Three
individual
level
factors
were
staAsAcally
signicant
for
predicAng
retenAon:
high
school
GPA
(z=14.30,
p<.001),
standardized
math
apAtude
(z=3.37,
p<.01),
and
family
income
and
educaAon;
as
expected,
high-income/conAnuing
generaAon
students
were
more
likely
to
persist
to
second
year
than
low-income/rst
generaAon
students
(z=4.01,
p<.001).
None
of
the
school
factors
were
signicant
for
predicAng
retenAon.ii
23
Table
3.
LogisAc
regression
analysis
of
retenAon
from
rst
to
second
year
of
college,
Analysis
1
(n=5,987)
Predictor
High
school
GPA
SE B
Odds RaHo
1.50
0.11
14.30
.00
4.49
-0.00
0.00
-1.02
.31
1.00
0.00
0.00
3.37
.00
1.00
-0.01
0.00
-1.67
.10
0.99
0.00
0.00
0.97
.33
1.00
0.12
0.13
0.92
.35
1.13
-0.03
0.21
-0.15
.88
0.97
0.50
0.13
4.01
.00
1.66
Analysis
2
Results
Means
and
standard
deviaAons
for
the
variables
in
Analysis
2
are
presented
in
Table
2.
Prior
to
conducAng
the
logisAc
regression
analyses,
exploratory
cross-tabulaAons
were
conducted.
First,
gender
was
crossed
with
rst
generaAon
status
and
retenAon.
The
results
indicated
that
both
of
these
factors
are
not
independent
of
gender
in
the
sample.
First
generaAon
students
are
more
likely
to
be
women
(2=27.77,
p<.001),
as
are
students
who
persist
to
the
second
year
of
studies
(2=22.85,
p<.001).
Based
on
these
ndings,
it
was
decided
to
run
separate
regression
models
for
men
and
women
to
see
how
well
the
selected
factors
predict
retenAon
based
on
gender.
The
other
cross-tabulaAons
indicate
that
retenAon
status
is
not
independent
of
rst
generaAon
status,
school
type,
whether
the
mother
was
19
or
younger
at
students
birth,
and
whether
parents
are
married.
These
results
suggest
that
the
variables
selected
for
the
model
are
appropriate
ones
for
predicAng
retenAon.
The
logisAc
regression
analyses
for
men
and
women
are
shown
in
Tables
5
and
6.
Standardized
scores
of
the
conAnuous
variables
in
the
model;
test
scores,
GPA,
income,
and
number
of
family
members;
were
used
to
aid
in
the
interpretaAon
of
odds
raAos.
Four
predictors
were
staAsAcally
signicant
in
the
predicAon
of
young
mens
retenAon
in
college
high
school
GPA
(2=131.468,
p<.001),
mathemaAcs
apAtude
(2=5.238,
p<.05),
Spanish
apAtude
(2=3.67,
p<.05)
and
school
type
(2=3.380,
p<.05).
Young
men
from
private
schools
are
1.27
Ames
more
likely
to
persist
to
the
second
year
of
studies
than
those
from
public
schools.
24
Table
5.
LogisAc
regression
analysis
of
mens
retenAon
from
rst
to
second
year
of
college,
Analysis
2
Walds
2
Predictor
B
SE
B (df=1)
p
Odds
RaHo
Z-
Individual
Spanish
ApAtude
0.120
0.062
3.761
.052
1.128
0.133
0.058
5.238
.022
1.142
-0.012
0.062
0.036
.850
0.988
0.567
0.049
131.468
.000
1.764
0.042
0.068
0.393
.531
1.043
-0.051
0.062
0.685
.408
0.905
Parent
educaAon
(0=rst
generaAon;
1=conAnuing
generaAon)
Parent
marital
status
(0=married;
1=not
married)
Mothers
age
when
S
born
(0=teen
mother;
1=not
teen
mother)
School
type
(0=public;
1=private)
-0.057
0.112
0.263
.608
1.059
-0.049
0.129
0.144
.704
0.952
-0.035
0.223
0.025
.874
0.965
0.235
0.120
3.380
.050
1.265
Table
6.
LogisAc
regression
analysis
of
womens
retenAon
from
rst
to
second
year
of
college,
Analysis
2
Walds
2
Predictor
B
SE
B (df=1)
p
Odds
RaHo
Z-
Individual
Spanish
ApAtude
0.113
0.067
2.803
.094
1.119
0.044
0.067
0.425
.514
1.045
-0.069
0.068
1.015
.314
0.933
0.443
0.051
74.654
.000
1.558
0.034
0.071
0.230
.631
1.034
-0.102
0.068
2.231
.135
0.903
0.233
0.125
3.501
.061
1.263
-0.349
0.134
6.783
.009
0.705
0.379
0.199
3.608
.057
1.460
0.310
0.141
4.860
.027
1.364
25
The
logisAc
regression
model
to
predict
young
womens
persistence
in
college
showed
a
slightly
dierent
paKern
of
staAsAcally
signicant
predictors
-
high
school
GPA
(2=74.654,
p<.001),
parents
marital
status
(2=6.783,
p<.01),
and
school
type
(2=4.860,
p<.05).
Young
women
whose
parents
are
married
are
1.4
Ames
more
likely
to
persist
than
those
whose
parents
are
not
married
(1/.705).
Finally
young
women
who
aKended
private
school
are
1.3
Ames
more
likely
to
conAnue
to
second
year
studies
than
young
women
who
studied
at
public
schools.
The
logisAc
regression
analyses
in
this
second
analysis
revealed
some
gender-based
dierences
in
academic
preparaAon
and
socioeconomic
factors
that
predict
retenAon.
Young
mens
persistence
appears
to
be
related
to
academic
preparaAon
(College
Board
scores,
high
school
GPA),
and
aKending
private
school,
while
young
womens
persistence
is
related
to
high
school
GPA,
parents
marital
status,
and
private
school
aKendance.
In
an
exploratory
vein,
an
addiAonal
logisAc
regression
analysis
was
conducted,
leaving
school
type
out
of
the
model,
to
see
if
any
of
the
other
socioeconomic
factors
would
become
signicant
predictors.
None
of
the
socioeconomic
factors
were
signicant
predictors
of
mens
retenAon.
However,
rst
generaAon
status
and
whether
mother
was
a
teenage
mother,
were
staAsAcally
signicant,
favoring
young
womens
whose
families
have
higher
educaAon
levels
and
women
whose
mother
was
over
19
years
when
she
was
born.
26
27
Discussion
There
is
scarce
research
available
on
factors
that
predict
college
student
retenAon
in
Puerto
Rico.
RetenAon
is
an
important
indicator
of
educaAonal
aKainment,
as
it
approximates
the
likelihood
to
graduate
from
college.
This
study
included
several
measures
of
two
factors
idenAed
in
the
research
literature
socioeconomic
indicators
and
academic
preparaAon
to
determine
whether
these
factors
are
important
in
the
predicAon
of
student
persistence.
In
the
rst
set
of
analyses,
high
school
GPA
and
math
apAtude
were
shown
to
be
important
predictors
of
staying
in
college,
as
well
as
the
family
income
and
educaAon.
These
results
have
parAcular
implicaAons
for
policy
and
pracAce
at
the
UPR
and
other
universiAes
in
Puerto
Rico.
Both
high
school
GPA
and
math
apAtude
are
included
in
the
admission
index
for
UPR,
thus,
the
results
demonstrate
that
those
requirements
are
appropriate
ones
for
adming
students
that
are
most
likely
to
complete
university
studies.
However,
knowing
that
socioeconomic
factors
parent
educaAon
and
income
-
aect
a
students
chances
of
conAnuing
and
compleAng
university
studies
suggests
that
insAtuAons
must
have
resources
and
supports
in
place
to
assist
those
students.
In
parAcular,
rst
generaAon
students
are
a
group
that
insAtuAons
should
target
for
improved
and
extended
orientaAon
services.
It
is
in
both
the
individual
and
public
interest
that
the
retenAon
and
graduaAon
rates
of
our
insAtuAons
improve,
and
insAtuAons
must
make
conscious
eorts
to
improve
retenAon
to
ght
impressions
that
solo
los
duros
pueden.
The
second
set
of
analyses
were
undertaken
to
incorporate
several
social
structural
variables
and
determine
whether
these
would
result
in
a
more
suitable
model
for
predicAng
retenAon.
The
decision
to
analyze
the
models
separately
for
men
and
women
was
made
aper
examining
the
rst
generaAon
and
retenAon
dierences
between
these
groups.
Contrary
to
expectaAons,
the
parental
income
variable
(gross
income)
and
the
number
of
family
members
did
not
make
signicant
contribuAons
to
the
predicAon
of
retenAon
for
either
men
or
women.
The
only
socioeconomic
variable
that
was
important
for
predicAng
retenAon
for
both
groups
was
aKendance
at
a
public
high
school;
that
is,
that
students
from
public
schools
are
less
likely
to
persist
at
UPRM
than
from
private
schools.
Given
that
the
College
Board
scores
for
public
school
students
tend
to
be
lower,
this
may
parAally
explain
this
result.
However,
College
Board
scores
were
included
in
the
model,
so
there
are
possibly
other
factors
at
play.
Given
that
about
60%
of
the
incoming
student
populaAon
comes
from
the
public
sector,
UPRM
must
do
a
beKer
job
of
trying
to
understand
why
this
is
happening
and
to
determine
how
to
improve
the
retenAon
of
these
groups.
The
most
interesAng
result
of
the
second
analysis
was
indeed
the
dierences
between
the
models
for
29
men
and
women,
and
how
family
background
characterisAcs
may
aect
them
dierently
at
the
moment
of
conAnuing
their
educaAon.
AddiAonal
research
using
these
data
should
explore
other
combinaAons
of
the
socioeconomic
variables,
for
individuals,
schools,
and
neighborhoods
perhaps
incorporaAng
census
tract
data
developed
in
Study
#2.
These
results
raise
quesAons
about
how
boys
and
girls
are
socialized,
and
whether
their
educaAonal
trajectories
and
moAvaAons
are
similar
as
they
aspire
to
complete
a
college
educaAon.
Research
uAlizing
data
from
other
public
and
private
insAtuAons
in
Puerto
Rico
would
be
an
important
next
step
to
understand
whether
these
ndings
are
insAtuAonally-specic,
or
indicaAve
of
a
broader
cultural
trend.
Given
that
we
know
that
retenAon
rates
tend
to
be
lower
for
men,
it
also
suggests
that
insAtuAons
should
be
working
to
gure
out
which
young
men
we
are
losing
and
how
to
beKer
retain
them,
perhaps
through
mentoring
programs.
While
these
results
were
obtained
from
analyses
of
data
from
a
single
insAtuAon
in
Puerto
Rico,
they
shed
light
on
class-based
dierences
in
the
experiences
of
higher
educaAon
in
Puerto
Rico.
If
other
insAtuAons
are
doing
these
types
of
analyses
for
insAtuAonal
research
purposes,
these
results
should
be
shared
more
widely
so
that
insAtuAons
may
work
together
to
use
limited
resources
to
improve
educaAonal
aKainment
island-wide.
Further,
there
should
be
more
basic
data
available
for
analysis,
from
both
the
public
and
private
higher
educaAon
sectors,
to
allow
a
more
comprehensive
analysis
of
the
eects
of
socioeconomic
factors
on
higher
educaAon
aKainment
in
Puerto
Rico.
30
31
Conclusions
Prior
to
this
study,
there
had
been
few
publicly
available
studies
on
socioeconomic
factors
and
educaAonal
aKainment
in
Puerto
Rico.
The
most
important
ndings
of
the
study
were
as
follows:
High
income
students
whose
parents
aKended
college
were
signicantly
more
likely
to
persist
in
college
than
low
income/rst
generaAon
students.
There
were
some
gender-based
dierences
in
academic
preparaAon
and
socioeconomic
factors
that
predict
retenAon
among
students
who
apply
for
nancial
aid.
Young
mens
persistence
appears
to
be
related
to
academic
preparaAon
and
public
school
aKendance,
while
young
womens
persistence
is
related
to
high
school
GPA,
parents
marital
status,
and
public
school
aKendance.
These
ndings
provide
empirical
evidence
for
educaAonal
aKainment
inequiAes
in
Puerto
Rico;
even
among
those
who
enroll
in
the
most
selecAve
higher
educaAon
insAtuAon
on
the
island,
socioeconomic
and
structural
factors
play
a
role
in
whether
they
conAnue
their
studies
beyond
the
rst
year.
The
limited
availability
of
research
on
higher
educaAon
in
Puerto
Rico
became
apparent
during
the
course
of
the
study.
While
we
were
aware
of
several
dierent
research
and
outreach
iniAaAves,
both
large
and
small,
across
dierent
insAtuAons
in
Puerto
Rico,
very
liKle
informaAon
was
available
about
the
outcomes
and
ndings
of
this
work.
Publically
available
research
reports
and
peer-reviewed
published
research
on
higher
educaAon
in
Puerto
Rico
are
scarce.
Even
the
CESPR
research
reports
are
not
published
online.
Also
during
the
course
of
this
project,
Puerto
Rico
experienced
the
largest
government
layo
in
history
over
17,000
employees
and
the
CESPR
was
drasAcally
downsized
like
many
other
Puerto
Rican
public
agencies.
In
a
Ame
when
liKle
informaAon
is
available
to
understand
challenges
facing
higher
educaAon
in
PR,
fewer
resources
are
being
allocated
to
this
eort.
33
project
showed
that
parent
income
and
educaAon
level
play
a
key
role
in
predicAng
retenAon.
Further
research
is
needed
to
understand
what
barriers
these
students
face,
in
contrast
to
higher
income
peers;
qualitaAve
research
should
play
an
important
role.
Conceptual
frameworks
includ-
ing
social
and
cultural
capital
would
also
help
to
provide
more
theoreAcal
tools
to
understand
educaAonal
inequity
in
PR.
Endnotes
i.
This
working
paper
includes
secAons
of
the
project
report
Socioeconomic
StraAcaAon
of
EducaAonal
AKainment
wriKen
by
the
papers
author
(Dr.
Sandra
Dika)
as
submiKed
to
the
Puerto
Rico
Council
of
Higher
EducaAon
(CESPR)
in
June
2010.
The
co
Principal
InvesAgators
of
the
project;
Drs.
Rima
Brusi
and
Walter
Daz;
provided
editorial
suggesAons.
ii. As
a
follow
up
to
the
logisAc
regression
in
Analysis
1,
a
stepwise
regression
was
conducted
to
explore
how
well
the
individual
and
social
level
variables
predict
achievement
in
college.
Achievement
was
measured
using
rst
year
GPA.
Alpha
to
enter
and
remove
variables
was
set
at
.15.
Overall,
the
model
explained
35%
of
variance
in
rst
year
GPA,
illustraAng
the
appropriateness
of
the
variables
selected.
High
school
GPA
made
the
strongest
contribuAon
(26%),
echoing
the
results
of
the
logisAc
regression
model
for
retenAon.
While
neither
of
the
school
composiAon
variables
contributed
to
the
predicAon
of
retenAon,
school
level
English
achievement
explained
an
addiAonal
7%
of
the
variance
in
rst
year
GPA.
The
other
factors
that
contributed
to
the
predicAon
of
grades
included
math
apAtude,
English
achievement,
and
family
income-educaAon.
High
income/conAnuing
generaAon
students
had
higher
rst
year
GPA
than
low
income/rst
generaAon
students,
while
the
low
income/rst
generaAon
students
had
beKer
results
than
high
income/rst
generaAon
students.
35
References
Baker,
J.
G.
(2009).
Class,
ability,
and
mobility:
Economic
and
academic
paths
from
middle
school
to
early
adulthood.
Journal
of
EducaAon
Finance,
34(4),
355-371.
BeKs,
J.
R.,
Rueben,
K.
S.,
&
Danenberg,
A.
(2000).
Equal
re-
sources,
equal
outcomes?
The
distribuAon
of
school
resources
and
student
outcomes
in
California.
San
Francisco:
Public
Policy
InsAtute
of
California.
Biddle,
B.
J.
&
Berliner,
D.
(2003).
What
research
says
about
unequal
funding
for
schools
in
America.
Available
online
at:
hKp://www.wested.org/online_pubs/pp-03-01.pdf.
Accessed
January
2010.
Borman,
G.
J.,
&
DAgosAno,
J.
V.
(1996).
Title
I
and
student
achievement:
A
meta-analysis
of
federal
evaluaAon
results.
EducaAonal
EvaluaAon
and
Policy
Analysis,
18(4),
309-326.
Bourdieu,
P.
(1986).
The
forms
of
capital.
In
J.
Richardson
(Ed.),
Handbook
of
theory
and
research
for
the
sociology
of
educaAon
(pp.
241258).
Westport,
CT:
Greenwood
Press.
Bowen,
W.
G.,
Kurzweil,
M.,
&
Tobin,
E.
(2005).
Equity
and
excel-
lence
in
American
higher
educaAon.
CharloKesville,
VA:
Univer-
sity
of
Virginia.
Braxton,
J.
M.,
Duster,
M.,
&
Pascarella,
E.
T.
(1988).
Causal
modeling
and
path
analysis:
An
introducAon
and
an
illustraAon
in
student
aKriAon
research.
Journal
of
College
Student
Devel-
opment,
29(3),
263272.
Brown,
R.,
Copeland,
W.
E.,
Costello,
E.
J.,
Erkanli,
A.,
&
Worthman,
C.
M.
(2009).
Family
and
community
inuences
on
educaAonal
outcomes
among
Appalachian
youth.
Journal
of
Community
Psychology,
37(7),
795-808.
Brusi,
R.
(2009).
Si
te
dejas
llevarUniversidad,
geograa
y
desigualdad.
Primer
Cuaderno
de
Trabajo,
Proyecto
Carvajal
para
la
DemocraAzacin
del
Conocimiento,
57
pp.
Mayagez,
PR:
Centro
Universitario
para
el
Acceso.
Available
online
at:
hKp://blogs.uprm.edu/cuacceso/documentos-invesAgacion-y-di
fusion/.
Accessed
November
2009.
Carbonaro,
W.
(1998).
A
liKle
help
from
my
friends
parents:
IntergeneraAonal
closure
and
educaAonal
outcomes.
Sociology
of
EducaAon,
71,
295313.
Cabrera,
A.
F.,
&
La
Nasa,
S.
M.
(2000).
Three
criAcal
tasks
Amer-
ica's
disadvantaged
face
on
their
path
to
college.
New
DirecAons
for
InsAtuAonal
Research,
27(3),
23-29.
Coleman,
J.
(1988).
Social
capital
in
the
creaAon
of
human
capi-
tal.
American
Journal
of
Sociology,
94
(Issue
Supplement),
S95
S120.
Daz,
W.
(in
press).
Pobreza,
clase,
y
logro
educaAvo
en
Puerto
Rico.
Segundo
Cuaderno
de
Trabajo,
Proyecto
Carvajal
para
la
DemocraAzacin
del
Conocimiento.
Mayagez,
PR:
Centro
Uni-
versitario
para
el
Acceso.
(available
online
at:
hKp://blogs.uprm.edu/cuacceso/documentos-invesAgacion-y-di
fusion/
in
May
2010).
37
Levitz,
R.,
Noel,
L.,
&
Richter,
B.
(1999).
Strategic
moves
for
re-
tenAon
success.
New
DirecAons
for
Higher
EducaAon,
108,
31-
49.
39
40
Acknowledgments
The
author
acknowledges
the
contribuAon
of
Dr.
David
Gonzlez,
Professor
of
Industrial
Engineering
at
UPRM,
in
the
data
preparaAon
and
analysis
for
the
rst
set
of
analyses
described
in
this
secAon.
Dika
and
Gonzlez
presented
a
preliminary
version
of
this
work
at
the
2009
AssociaAon
for
InsAtuAonal
Research
(AIR)
Forum
(see
references).
As
always,
any
error
in
the
text
is
the
sole
responsibility
of
the
author.
The
research
presented
in
this
working
paper
was
funded
by
Fundacin
Carvajal
and
the
University
of
Puerto
Rico-Mayagez.
41
Centro Universitario para el Acceso Universidad de Puerto Rico - Mayagez e centro.accesso@upr.edu w cua.uprm.edu