Martin E. Sandler, Ph.D., Research In Higher Education Article, October 2000

Research in Higher Education, Vol. 41, No. 5, 2000

CAREER DECISION-MAKING SELF-EFFICACY,
PERCEIVED STRESS, AND AN INTEGRATED
MODEL OF STUDENT PERSISTENCE:
A Structural Model of Finances, Attitudes,
Behavior, and Career Development

Martin E. Sandler

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
In response to the extraordinarily diverse adult student population present in college
today, a new structural equation model adapted from Cabrera et al. (1993) integrated
model of student retention was identified with the addition of three variables: career
decision-making self-efficacy (CDMSE), perceived stress and financial difficulty. The
study examined the persistence of students (N = 937) 24 years of age or older study-
ing in two-year and four-year degree programs, by combining data from a survey
questionnaire and institutional records. Twenty-three variables were included, twelve
endogenous variables and eleven exogenous variables, within a nonrecursive struc-
tural equation model. The exogenous variables controlled for the background charac-
teristics of the population of adult students examined. Of the twelve endogenous
variables of a new integrated model of student persistence, CDMSE, a career devel-
opment construct related to the perceived vocational futures and career-related tasks
of adult students has the widest range of influence among the endogenous variables.

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

As a response to the changing demographics of the American work force, the
extraordinarily diverse adult student population present in college today and the
problem of nontraditional student attrition at two-year and four-year urban col-
leges, the construct of career decision-making self-efficacy (CDMSE) is intro-
duced into a new structural model that builds on the integration of the models
of Tinto (1975, 1987, 1993) and Bean and Metzner (1985) as synthesized by
Cabrera, Castaneda, Nora, and Hengstler (1992a), and Cabrera, Nora, and Cas-

Parts of the research were presented at the annual conference of the Association for the Study of
Higher Education in San Antonio, TX, November 1999, and the annual meeting of the American
Educational Research Association in New Orleans, LA, April 2000.
Martin E. Sandler, Assistant Director, The Real Estate Institute, Real Estate Degree Programs,
New York University, School of Continuing and Professional Studies, 11 West 42nd Street, Room
509, New York, NY 10036-8083; martin.sandler@nyu.edu.

537
0361-0365/00/1000-0537$18.00/0  2000 Human Sciences Press, Inc.

538 SANDLER

taneda (1992b, 1993). Although there has been an awareness of the problem of
attrition within undergraduate degree programs, research efforts at examining
this trend affecting adult students have been only moderately successful (Kas-
worm, 1990; Kasworm and Pike, 1994).
The cross-sectional survey research reported in this article introduces a new
and heretofore neglected dimension that proved to be critical to the understanding
of the complexities of adult persistence. It identified a variable, CDMSE, a career
planning and development construct, which previous models of the persistence
phenomenon did not include, thus permitting a richer explanation of the process
of student persistence (attrition’s antonym). CDMSE identifies the degree of con-
fidence students express about their competency or ability (self-efficacy) to em-
bark on informational, educational, and occupational goal planning activities
(Taylor and Betz, 1983; Betz, Klein, and Taylor, 1996; Peterson, 1993a). With
its inclusion in a new structural model of student persistence, the cognitive-initi-
ated career expectations and agentive behavior of adult students can be more
closely examined within the environmental, academic, social, and institutional
systems of undergraduate student life. An attempt is made to examine the per-
ceived vocational futures and career development of adult students (CDMSE),
and, most importantly, to explain the relationships engendered within a new path
model that ultimately lead to the central variable of this inquiry—persistence.
This study adds two other important constructs to provide a more comprehen-
sive examination of adult student persistence. The Cabrera et al. (1993) student
integration/student attrition framework was modified to acquire a nontraditional
footprint so that better testing of an adult undergraduate population could be
achieved (Kasworm, 1990; Kasworm and Pike, 1994; Sandler, 1998). One new
variable, perceived stress, examines the stress experienced by adult students. It
measures the amount of stress adult students perceive due to the energy they
expend and to the amount of work that college requires. Financial difficulty
assesses the attitudes that adult students express about financial difficulty while
attending college, and conceptually explores critical ground beyond the social
and intellectual development needs of students as conceptualized in the struc-
tural synthesis of Cabrera, Nora, et al. (1992, 1993). As this new integrated
model of student persistence verifies, CDMSE is empirically shown to play an
important role in the lives of nontraditional students with respect to their per-
ceived stress, academic integration, social integration, institutional commitment,
intent to persist, and actual persistence in college.

CONCEPTUAL FRAMEWORK AND RELATED LITERATURE
An extensive literature on student persistence and attrition exists prior to the
1970s. However, few systematic investigations were performed that utilized
conceptual models. Attempts to describe and explain the phenomenon of the
higher education dropout process were simply descriptive (Pantages and Cree-

FINANCES, ATTITUDES, BEHAVIOR, AND CAREER DEVELOPMENT 539

don, 1978; Feldman and Newcomb, 1969). Two theorists emerged to provide
comprehensive frameworks in which to better explain college departure and
withdrawal behavior. Cabrera et al. (1993) have aptly referred to these two
models as the “student integration model” of Tinto (1975, 1987) and the “stu-
dent attrition model” of Bean and Metzner (1985). Both models have been vali-
dated across various settings and populations (Cabrera, Castaneda, et al., 1992).
Although the student attrition model has been validated empirically for tradi-
tional populations, its designers’ focus has been on nontraditional student popu-
lations (Cabrera et al., 1993).
A third seminal model, “the integrated model of student retention” (Cabrera
et al., 1993) surfaced in the early 1990s to bring the two competing earlier
models into focus in a single synthesized structural framework. Through the
merger of the student integration model of Tinto (1975, 1987) and the student
attrition model of Bean and Metzner (1985), Cabrera et al. (1993) examined the
underlying factors that have been utilized for both traditional and nontraditional
student populations. By merging these models, a more comprehensive under-
standing of the complex complementary and divergent capacity of each compet-
ing theory was yielded in a new explanatory framework.
Each variable construct defined in the study section below has either origi-
nated in the student integration model of Tinto (1975, 1987, 1993), or the stu-
dent attrition model of Bean and Metzner (1985), the synthesis of Cabrera, Cas-
taneda, et al. (1992) Cabrera et al. (1993), or in the new integrated model of
student persistence by this author. Through the synthesis of the integrated model
of student retention of Cabrera et al. (1993), variable constructs are interrelated
with antecedent reference made to each respective investigator and their pro-
posed system of relationships. A hypothetical model for this study unfolds as a
system of twelve variable constructs and their purported relationships, while
providing for the effects of eleven background variables.
The theory of planned behavior is employed as a conceptual bridge for model
integration in the present study. Model variables deployed in the integrated
model of student persistence address precursor theories noted above, and test
new relationships that have not been examined before in an elaborated frame-
work (Braxton, Sullivan, and Johnson, 1997). These variables are conceptually
explored in innovative ways, in large part due to the plurality of outcomes that
adult student behavior can bring to the context of undergraduate learning (Kas-
worm, 1990). As a rationale for their inclusion in a new model, these variables
are investigated with nontraditional students to examine the degree of model
integration that stems from social cognitive theory and attitude-behavior interac-
tions (Bandura, 1997; Braxton et al., 1997; Eagly and Chaiken, 1993). In turn, a
new model of adult adjustment that examines adult career development, student
integration, and student persistence is introduced which sustains an elemental,
if not conceptual, allegiance to the integrated model of retention of Cabrera et
al. (1993), its variables and foundation theories synthesized within.

540 SANDLER

Some researchers have argued, nevertheless, that factors used in the investiga-
tion of traditional student populations may not be appropriate for models of
nontraditional adult students because of the diverse—if not confusing—results
obtained with adult undergraduates (Kasworm, 1990). In addition, Kasworm and
Pike (1994) questioned the efficacy of employing some constructs related to
the student integration model (Pascarella, 1980) in the testing of nontraditional
students. Specifically, empirical evidence was obtained that challenged the ap-
propriateness of the inclusion of academic performance in a traditional model
of student satisfaction (Kasworm and Pike, 1994).
In light of the findings of Kasworm and Pike (1994), careful attention was
given in the development of a new model to the “five domains of reality” of
Kasworm (1990), critical perceptual frameworks used in past research on adult
undergraduate students. The integrated model of student retention of Cabrera et
al. (1993) was modified in this research investigation to include new constructs
germane to nontraditional students. In this manner, critical assumptions that
researchers of adult undergraduate students have historically addressed were
assessed for inclusion in an elaborated model with the following additional con-
structs: career decision-making self-efficacy, perceived stress, and financial dif-
ficulty. An examination of critical path relationships within the model and the
explained variance of related constructs are evaluated to assess the relative im-
portance of selected constructs deployed in this investigation. A reexamination
of the conclusion of Kasworm and Pike regarding the appropriateness of a tradi-
tional model of performance is made.

Adult Undergraduates in Higher Education:
Five Domains of Perceptual Reality
In a review of the research, Kasworm (1990) elucidated that past research
perspectives on adult undergraduate students have been influenced largely by
beliefs that pertain to the traditional student body. A broad spectrum of issues
pertaining to adult undergraduate education was delineated and examined by
means of a meta-analysis (Kasworm, 1990). “Specifically, research on the adult
undergraduate has been highly diverse in its assumptions, beliefs, and perspec-
tives of adults within the academic tradition of undergraduate education” (Kasw-
orm, 1990).
By means of content analysis, five domains of perceptual reality were sum-
marily cited by Kasworm (1990) about past adult undergraduate research: (1)
image of implied deficiency, (2) image of student entry and adaptation, (3)
image of description and characterization, (4) image of psychosocial develop-
ment, and (5) image of equity and outcome. The nominal listing of the five
domains gives a cursory explanation of the perceptual categories addressed.
Readers may refer to Kasworm for definitional reference regarding each.


In the building of a new model, the five domains of perceptual reality are
revisited in the selection of endogenous and exogenous variables. Although the
student integration model of Tinto (1975, 1987) is not considered in the meta-
analysis of Kasworm (1990), it is considered within the context of the the pres-
ent study as one that is relevant to adult undergraduate students. Elements of
the student integration model are assessed, albeit in a hybrid context borne from
the integrated model of student retention of Cabrera et al. (1993).
With the testing of a new person-environment model, an attempt is made “to
capture the reality of the transactional relationship between adult students and
the undergraduate institution,” by examining adult development in a student
learner context (Kasworm, 1990). In turn, relationships are drawn between the
undergraduate experience and other key life roles that adult students encounter.
The constructs, background variables, path relationships, and theories employed
bear antecedent reference to the five perceptual domains addressed by Kasworm.
With the assistance of another meta-analysis conducted by Braxton et al. (1997),
an enriched scope of empirical considerations regarding the student integration
model of Tinto (1975, 1987, 1993) and the integrated model of student retention
of Cabrera et al. (1993) are considered. Braxton et al. compiled evidence that
attests to the increased capacity of the student integration model of Tinto with
commuter students, and, in part, nontraditional populations. The testing of the
new model presented here attempts to bridge elements of conflict pertaining to
the conclusions of the meta-analyses of Kasworm and Braxton et al. by introduc-
ing an elaborated framework and new evidence. In this manner, the new model
attempts to broaden the scope of inquiry for adult undergraduate students. Nev-
ertheless, the research findings presented here defy simple codification and a
single brush stroke explanation. The diversity of results reached by researchers
of adult students remains a conundrum, largely due to the heterogeneity of char-
acteristics of adult students and the variety of responses pluralistic societies like
ours can engender in the context of higher education.

The Tinto Model—Student Integration Model
Tinto provides an explanatory model or predictive model of student integra-
tion in which he explains the process by which individuals leave colleges and
universities before graduating (1975, 1987). By employing the notion of person-
environment fit (Lewin, 1935; Pervin, 1968) in the student integration model,
Tinto (1975, 1987, 1993) identifies attrition as the lack of congruency between
students and institutions. Tinto (1975, 1987) anecdotally incorporated the related
notions of the rites of passage to adulthood and suicide as described and investi-
gated by Van Gennep (1960) and Durkheim (1951), respectively.
The designers of the student integration theory match the motivation and
academic ability of the student with the academic and social characteristics of

542 SANDLER

the institution to establish two essential commitments: commitment to the edu-
cational goal and commitment to remain at the institution (Cabrera et al., 1993).
Tinto emphasizes the constructs of goal and institutional commitment that stu-
dents acquire or develop because of their interaction with individuals and subse-
quent integration into the institution (Tinto, 1975). Integration refers to the “ex-
tent of shared normative attitudes and values of peers and faculty in the
institution and abides by the formal and informal structural requirements for
membership in the community” (Pascarella and Terenzini, 1991). Students enter
college with varying backgrounds, dispositions, intentions, and goals, which in
turn are “modified and reformulated on a continuing basis through a longitudinal
series of interactions between the individual and institutional structures and
members of the academic and social systems of the institution” (Pascarella and
Terenzini, 1991). Satisfaction derived from these structures and academic and
social systems by students are theoretically presumed to lead to integration and
subsequently to retention (Pascarella and Terenzini, 1991; Pascarella, Duby,
and Iverson, 1983).
Although the social and intellectual development of students may be the prin-
cipal goal of institutions of higher education (Tinto, 1975, 1987), retention re-
mains an important related outcome. As an outcome, retention reflects upon the
genuine concern of the institution for the social and intellectual development of
students (Tinto, 1975, 1987). The student integration model of Tinto (1975,
1987, 1993) has been extensively researched in various settings and many insti-
tutions. Some reformulations of the student integration model of Tinto (Cabrera
et al., 1993) have arisen, providing a wealth of empirical data and conceptual
clarification (Fox, 1986; Pascarella et al., 1983; Pascarella and Terenzini, 1991;
Braxton et al., 1997).

The Bean Model—Student Attrition Model
As a model for nontraditional student socialization, the integration concept of
Tinto remains a difficult task to achieve based on the social factors assumed. The
social experiences of nontraditional students include important external factors,
such as the influence of family, friends and employers. According to Bean and
Metzner (1985), Tinto does not address these external factors in his theory of depar-
ture. Bean explores an alternative model to explain persistence or attrition using an
organizational process model of turnover (Bean, 1980; Cabrera et al., 1993) and a
model of attitude-behavior relations and interactions (Bentler and Speckart, 1979),
often called the theory of reasoned action (Fishbein and Ajzen, 1975).
Behavioral intentions associated with persistence or attrition act as predictors
of persistence behavior or outcomes (Cabrera et al., 1993). Behavioral intentions
are formed by attitudes based on beliefs. With respect to attrition from an aca-
demic institution, the attitudes toward the academic experience at the institution


are presumed to affect the intent to continue in school, which in turn results in
persistence or attrition, the actual staying in, or leaving school by students. This
theoretical overlay of reasoned action links the sets of variables in the model
(Bean and Metzner, 1985). Through person-environment interaction, beliefs are
presumably influenced by the experiences of students with different characteris-
tics of the institution, including institutional quality, courses, and friends (Ca-
brera et al., 1993). In addition, factors external to the institution affect both
attitudes and decisions in the student attrition model and are active while the
student is attending the college (Bean and Vesper, 1990; Bean and Metzner,
1985). Environmental factors, excluded in the more traditional student analysis
of Tinto (1975, 1987, 1993), play a compensatory role in Bean’s student attrition
model (Cabrera et al., 1993). “Thus, for nontraditional students, environmental
support compensates for weak academic support, but academic support will not
compensate for weak environmental support (Bean and Metzner, 1985).
With subsequent variations and reformulations of the student attrition model,
Bean and Vesper (1990) acquired empirical support for the involvement of di-
rect and indirect effects of nonintellective or external factors, such as family
approval, in the student attrition model. These factors were found among six
environmental, personal, and organizational variables that served to explain
most of the variance in the student attrition process (Bean and Vesper, 1990).

The Cabrera Model—Integrated Model of Student Retention
The integrated model of student retention celebrates an amalgam of the con-
structs of Tinto and Bean where both models converge (Cabrera et al., 1993).
It includes the divergent advantages and discriminant emphases of the student
integration model (Tinto, 1975, 1987) and the student attrition model (Bean
and Metzner, 1985). A synthesis or baseline model of structural relations was
formulated that included all the structural paths substantiated from the student
attrition model (Cabrera, Castaneda, et al., 1992). In turn, the emphasis of the
student attrition model on the role of factors external to the institution is utilized
as a critical construct important to the attitudes and decisions of students (Bean,
1982a, b; Cabrera et al., 1993). Beyond the constructs of social and academic
integration, a relationship between goal identification and persistence has been
cited (Astin, 1975; Beal and Noel, 1980; Peterson, 1993a), and this line of
investigation is preserved in the integrated model of student retention with re-
spect to goal commitment.
Since grade point average (cumulative GPA) loaded poorly as an indicator
of academic integration in the student integration model, cumulative GPA and
academic integration were incorporated as interdependent constructs, each de-
marcating independent positions in the synthesis of Cabrera et al. (1993). Ac-
cording to propositions of the student attrition model (Bean and Metzner, 1985;

544 SANDLER

Bean and Vesper, 1990), family encouragement and financial attitudes can influ-
ence and channel their effects upon academic integration, commitments to the
institution, and produce persistence solutions as outcomes (Cabrera et al., 1993).
In the structural model employed in this research, the incorporation of cumula-
tive GPA and external variables that include financial attitudes and outside en-
couragement is similar.
Cabrera et al. (1993) also compare the student integration model and the
student attrition model based on the number of hypotheses validated in the em-
pirical literature and conclude that the student integration model remains
stronger than the student attrition model (“70 percent versus 40 percent”). In
the student attrition model, more of the observable variance was explained as
far as the persistence criterion is concerned (“44 percent versus 38 percent”;
Cabrera et al., 1993).

The Theory of Planned Behavior—
A Conceptual Bridge for Model Integration
Career decision-making self-efficacy, perceived stress, and an integrated
model of student persistence, a new structural model of adult adjustment, incor-
porates conceptual variables that are theoretically related by means of the social
cognitive theories of self-efficacy and planned behavior (Ajzen, 1991; Bandura
and Cervone, 1983; Bandura, 1997). These two complementary attitudinal
frameworks act as a conceptual overlay, or bridge, for model integration. In
turn, the deployment of two related social cognitive psychological theories facil-
itated the linking of variables in a new path model.
Ajzen (1991) believed that the theory of reasoned action needed reformulation
to explain behaviors not entirely under the volitional control of a given subject.
In response, perceived behavior control, which proved to be critical to the exe-
cution of given target behaviors (Madden, Ellen, and Ajzen, 1992), was intro-
duced (Ajzen, 1991). Perceived behavioral control is defined as “the separate
measurements of beliefs regarding the requisite resources and opportunities that
people have, or perceive they have for performing the behavior” (Ajzen, 1991).
With the confirmation of the efficacy of the theory of planned behavior by
Ajzen, self-efficacy and perceived behavioral control are conceptually incorpo-
rated by the present author as a social cognitive overlay or network within
the integrated model of student persistence. Bean and Metzner (1985) similarly
engaged the theory of reasoned action (Fishbein and Ajzen, 1975) in the student
attrition model. A large degree of persistence behavior borne by subsequent
empirical investigations of the student attrition model (Bean and Metzner, 1985)
suggest that the intent to persist variable explained a large degree of the vari-
ance, confirming the contribution of the theory of reasoned action to model
building (McCaffrey, 1991; Chartrand, 1992).


In a similar manner, by incorporating the theory of planned behavior as a
conceptual bridge (Ajzen and Madden, 1986; Ajzen; 1991, Madden et al., 1992),
the construct of career decision-making self-efficacy (Betz and Hackett, 1981,
1986) can be substituted for the analogous construct of perceived behavioral
control. As employed in the theory of planned behavior, “perceived behavioral
control is in turn determined by control beliefs, which are beliefs about the
likelihood that one possesses the resources and opportunities thought necessary
to execute the behavior or attain the goal” (Eagly and Chaiken, 1993). In this
examination of adult students’ persistence, it is hypothesized that a higher level
of explained variance might be achieved for the constructs intent to persist and
persistence with the inclusion of CDMSE in a new model.

Career Decision-Making Self-Efficacy
Career decision-making self-efficacy (Betz and Hackett, 1981; Taylor and
Betz, 1983) identifies the degree of confidence students express about their com-
petency or ability—self-efficacy—to embark on educational and occupational
information-gathering and goal-planning activities (Peterson, 1993a). CDMSE
employs social cognitive theory as a career development measure, and explores
the role of self-referent thinking in guiding human motivation and behavior. It
encompasses the person-environment situation or “mutually interacting influ-
ences between persons and their behavior and environments” (Lent, Brown, and
Hackett, 1994). As originally conceived, “self-efficacy percepts are postulated
as helping to determine one’s choice of activities and environments, as well as
one’s efforts expenditure, persistence, thought patterns, and emotional reactions
when confronted by obstacles” (Lent, Brown, and Hackett, 1994; Luzzo, 1996).
Self-efficacy must be related to specific target behavior, whereby beliefs about
the ability of one to perform certain tasks successfully remain characteristic
of certain behaviors (Peterson, 1993a). By definition, self-efficacy represents a
perception and judgment of an individual of his or her abilities and competen-
cies to organize and execute causes of action required to fulfill specific types of
behavior or performances (Bandura, 1986).
Self-efficacy beliefs are central to the notion of personal agency. They operate
in a pervasive fashion (Lent, Brown, and Hackett, 1994; Multon, Brown, and
Lent, 1991; Sadri and Robertson, 1993), and interact with other motivational
systems of human endeavor that include personal capabilities and performance
accomplishments (Peterson, 1993a; Bandura and Cervone, 1983). As suggested
previously, self-efficacy beliefs may also be employed as a conceptual analog
for perceived behavioral control incorporated in the theory of planned behavior
(Ajzen, 1988, 1991; Ajzen and Madden, 1986). Elements of the theory of
planned behavior and self-efficacy beliefs are conceptually utilized in the inte-
grated model of student persistence by this author, in particular where attitude-

546 SANDLER

behavior interactions, reasoned action, environmental resources, and volitional
control are concerned (Fishbein and Ajzen, 1975; Bentler and Speckart, 1979;
Ajzen, 1991; Eagly and Chaiken, 1993). CDMSE is comprised of a measure of
perceived self-efficacy or behavioral competence in relation to a complex do-
main of behavior. It summarily describes a form of “agency in educational and
career pursuits” (Betz and Hackett, 1987).

Perceived Stress
Perceived stress examines the stress experienced by adult students. As a psy-
chological/attitudinal variable, it measures the amount of stress adult students
perceive due to the energy they expend and the amount of work that college
requires (Cabrera, 1988). The stress experienced by adult students may have a
positive or negative impact, as the terms “eustress” and “distress” imply respec-
tively (Bean and Metzner, 1985; Munoz, 1987; Selye, 1974; Lazarus and Folk-
man, 1984; Bandura, 1997; Dill and Henley, 1998). The cognitive, or attitudinal,
component of perceived stress was introduced due to evidence of Chartrand
(1992) with regard to the mediating effects of the “absence of psychological
distress” variable among the constructs of the student attrition model of Bean
and Metzner (1985). Perceived stress fits within the expanded context of theoret-
ical inquiry explored with this investigation. It couples conceptually with the
theory of planned behavior (Ajzen, 1991), the parallel notion of self-efficacy
(Bandura, 1997), and the financial attitudes of difficulty and satisfaction utilized
in model building.

Financial Attitudes/Difficulty
Contrary to career decision-making self-efficacy, the environmental construct
of financial attitudes/difficulty (Mallette and Cabrera, 1991) is also added to
explore a new integrated model of student persistence. Financial attitudes/diffi-
culty examines the experiences of financial hardship by adult students, and their
perceived limits of available resources. Although the model of Cabrera et al.
(1993) used financial attitudes/satisfaction as a single endogenous variable for
finances in the integrated model of student retention, the present study incorpo-
rates two attitudinal measures regarding student finance: financial satisfaction
and financial difficulty. In addition, unlike the model of student finances of
Cabrera, Castaneda, et al. (1992), this researcher used a continuous measure for
financial aid, an actual disbursed monetary figure as an objective component for
student finances. As described, financial aid is explored as an enriched exoge-
nous or background variable within a new integrated model—along with two
orthogonal attitudinal constructs, financial attitudes/satisfaction (Cabrera et al.,
1993) and financial attitudes/difficulty (Cabrera, Nora, and Castaneda, 1992;
Mallette and Cabrera, 1991).


Career Decision-Making Self-Efficacy, Perceived Stress, and an
Integrated Model of Student Persistence: A Hypothetical Model
By linking the integrated model of student retention of Cabrera et al. (1993)
with the constructs of career decision-making self-efficacy, perceived stress, and
financial attitudes/difficulty, an attempt is made to better explain the variance
observed regarding the intent to persist and persistence of adult students. In
addition, an effort is made to recognize the perceived stress experienced by
adult students about the parallel processes of educational development and ca-
reer development, in particular with respect to the concepts of agency and career
decision making within a student/learner context. The following variables bear
antecedent reference to the student integration theory of Tinto (1975, 1987,
1993): academic integration, social integration, institutional commitment, and
goal commitment, whereas family encouragement, perceived stress, financial
attitudes (satisfaction and difficulty), academic performance, and intent to per-
sist originated conceptually with the student attrition model of Bean and Met-
zner (1985).
As Braxton et al. (1997) and Lent et al. (1994) have clearly stated, CDMSE
has shown strong promise for persistence/attrition research. Within a student/
learner context, it explores the confidence students have in making decisions
about their career futures in the workplace. Although Peterson (1993a) incorpo-
rated CDMSE with elements of the student integration model of Tinto (1975,
1987, 1993) related to the constructs of integration and institutional commit-
ment, empirical evidence was not explored with respect to adult student persis-
tence (Peterson, 1993a, b). Peterson and delMas (1996) subsequently attempted
to structurally evaluate CDMSE in a persistence model. Their path model fell
short of the scope explored with the integrated model of student persistence that
is adapted here with significant conceptual modifications to the integrated model
of student retention of Cabrera et al. (1993). Through the dynamic interplay of
CDMSE with other constructs therein, the new model explains persistence deci-
sions or behavior that may be constrained by limited resources (financial diffi-
culty and family encouragement) not entirely under the volitional control of
adult students (Ajzen, 1991).

RESEARCH DESIGN AND METHODS
Population and Sample
The population for this research investigation was composed of adult/nontra-
ditional undergraduate students 24 years of age or older studying on a part-time
or full-time basis in a two-year and four-year degree bearing program for adult
students at a private urban research university. The sample was chosen from a
sample frame of 1,634 matriculated adult degree studies students who met the
criteria for advisement and registration within one division of a large private

548 SANDLER

research university in the northeast United States. A division of adult undergrad-
uates was chosen because its programming, curriculum, and advisement were
designed in large part for adult/nontraditional learners who were employed and
seeking an undergraduate education with flexible schedules that included week-
day, evening, and weekend study.
A survey questionnaire was distributed to a randomly selected sample of 937
adult students enrolled during the Fall 1995 semester. After a 63 percent survey
response and listwise deletion, the sample for data analysis comprised 469 adult
students. The sample included students who were 24 years of age or older,
whose cumulative credit hours were four or more, and whose curriculum hours
at the institution were two or more credits during the term of the investigation.
These criteria described a sample of continuing full-and part-time, nontradi-
tional, adult students. Over 95 percent of the students were commuters. In addi-
tion, the sample included students who began their studies at the institution as
freshman (41.2%) and transfer students (58.8%). Approximately half the stu-
dents in the sample were studying for an associate’s degree (50.3%); the remain-
ing students were studying for a bachelor’s degree (49.7%). Approximately one
third of the students were majors in the liberal arts and humanities combined,
whereas the remaining students were majors in business, the health professions,
and the social sciences combined. White students were in the majority (51.0%);
nonwhite students were in the minority by a small margin (49.0%). Female
students were predominant (71.2%); male students were smaller in number
(28.8%). The persistence rate for the sample population was 79.5 percent be-
tween the fall and spring semesters. Prospective graduates in attendance during
the term of investigation were excluded from the sample so persistence data
could be accurately assessed from enrollment records after the survey adminis-
tration.
Admission was based largely on an admissions test and not on previous
grades or SAT scores. A high school diploma or General Equivalency Diploma
(GED) verification was required. Other background characteristics were ob-
tained from university and school/divisional records. From these data sources, a
total of 23 variables were included in this research study in an effort to ascertain
their relationship to persistence for the sample and population of this inquiry.

Instrumentation
Twelve endogenous variables are included in the integrated model of student
persistence. In addition, eleven exogenous variables are included as background
or demographic variables. Degree program, one exogenous variable, differenti-
ates between two-year and four-year degree programs and serves as a control.
Two instruments were integrated and adapted for the sample being examined:
the Career Decision-Making Self-Efficacy—Short Form (CDMSE-SF) scale


(Betz, Klein, and Taylor, 1996; Betz and Taylor, 1994; Taylor and Betz, 1983)
and the Student Experiences Survey (Cabrera, 1988) employed in the “inte-
grated model of student retention” (Cabrera et al., 1993; instruments used by
permission). As an amalgam of these two instruments, a single survey question-
naire, the Adult Student Experiences Survey (ASES) was administered to collect
attitudinal data and self-reported background characteristics. Other background
characteristics were obtained from university and school/divisional records. As
tested, the reliability of the scales employed in large part matched or surpassed
the levels reported by their respective developers.

Conceptual Summary
A path of longitudinal interactions hypothetically occurs between students
and an institution in a given term of study. With a conceptual overlay of the
theory of planned behavior (Ajzen, 1991) and input from environmental vari-
ables (family encouragement, financial attitudes/difficulty), and background
characteristics, career decision-making self-efficacy percepts, perceived stress,
social integration, and academic integration take place within related subsystems
of the college community to culminate in two fundamental commitments—a
commitment to the institution and a commitment to the personal goals of adult
students (Tinto, 1975, 1987, 1993). Through these commitments and perfor-
mance outcomes measured by cumulative GPA, students either express an intent
to persist or leave the institution, which results in a behavioral outcome of per-
sistence or attrition, respectively (Cabrera et al., 1993; Bean and Metzner, 1985).
In summary, the Career Decision-Making Self-Efficacy, Perceived Stress, and
an Integrated Model of Student Persistence: A Hypothetical Model is presented
in Figure 1.

Variables in the Study
The predictor endogenous (independent) variables included: career decision-
making self-efficacy (CDMSE), family encouragement, perceived stress, finan-
cial attitudes/satisfaction, financial attitudes/difficulty, academic integration, so-
cial integration, cumulative GPA, institutional commitment, goal commitment,
and intent to persist. The exogenous variables of the model included eleven
variables that pertained to student background: gender, race/ethnic affiliation,
household income, relatives/dependents, financial aid, parents’ educational
level, academic degree aspirations, student type, degree program, curriculum
hours, and hours employed. The criterion (a dependent endogenous variable)
was persistence, a dichotomous outcome. Definitions of the variable constructs
and related information are provided below.
Academic Integration was comprised of three survey items. The notion of the

550 SANDLER

FIG. 1. Career decision-making self-efficacy, perceived stress, and an integrated model
of student persistence: A hypothetical modelb
b
The conventional syntax used in path diagrams may be deviated from in order to simplify represen-
tation.

“anticipation of academic performance” and “satisfaction with course curricu-
lum” by students was explored by Cabrera et al. (1993). As it is explored in the
present investigation, academic integration concerns the feelings students ex-
press about being a part of the academic life of the institution. By examining
the perceptions of adult students regarding their (1) academic performance, (2)
their satisfaction with the curriculum, and (3) their feelings of being a part of
the academic institution, the research examined the degree to which students
become involved in the academic system and intellectual life of the university
as the student integration model of Tinto suggests (1975, 1987, 1993). A five-
category Likert scale—strongly agree to strongly disagree—was employed. The
Cronbach alpha reliability coefficient of the three-item scale for academic inte-
gration was .57.
Background Variables are data or specific information collected about stu-
dents prior to their enrollment at the institution, and originate from institutional
records and by means of the ASES. Some background variables investigated


are proximal to the term of inquiry reflecting data collected during a survey
administration. The background variables investigated include eleven exogenous
variables: (1) gender, (2) race/ethnic affiliation, (3) household income, (4) rela-
tives/dependents, (5) financial aid, (6) parents’ educational level, (7) academic
degree aspirations, (8) student type, (9) degree program, (10) curriculum hours,
and (11) hours employed. Race and gender were included to ascertain the soci-
ostructural impact of these critical background characteristics on the adult
learner. Four exogenous variables (numbers 3, 4, 5, and 6) served as controls of
the socioeconomic background of the adult population examined. Student type,
variable number 8, controlled for the presence of both freshman students, first
time undergraduates at the institution in which the research was conducted, and
transfer students. Degree program, variable number 9, controlled for the pres-
ence of adult students studying for two-year and four-year degrees respectively.
Curriculum hours served as an institutional measure representing adult students’
investment or involvement in their program of study. Hours employed, a self-
reported variable, examined the impact that employment had on adult learners.
Career Decision-Making Self-Efficacy (CDMSE-SF), the CDMSE-SF is com-
prised of 25 items and identified the extent to which students are confident
(have self-efficacy) about their ability to engage in educational and occupational
information gathering and goal planning activities (Betz, Klein, and Taylor,
1996; Hackett and Betz, 1981; Taylor and Betz, 1983; Peterson, 1993a). CD-
MSE was included to examine the impact that the vocational futures of adult
students have on educational development. Critical new theoretical insight from
the social-cognitive psychological frameworks of social learning theory, self-
efficacy, and the theory of planned behavior are considered (Ajzen, 1991; Eagly
and Chaikin, 1993; Bandura, 1997). A single total score (interval level) was
calculated by summing the score of each of the 25 tasks for each subject. The
maximum score for all 25 items is 225 on the CDMSE-SF scale. Each task was
measured on a 10-point scale that ranged from no confidence (0) to complete
confidence (9). The Cronbach alpha reliability coefficient of the 25-item scale
for CDMSE was .96.
Cumulative GPA (Grade Point Average) means academic performance as a
continuous measure (0.000–4.000). Cumulative GPA was obtained from institu-
tional records to examine the academic performance of adult students within an
undergraduate student/learner context, as conceptualized in the student attrition
model of Bean and Metzner (1985).
Family Encouragement, an environmental component, conceptually origi-
nated with the student attrition model of Bean and Metzner (1985). The scale is
comprised of two items and explores the construct of encouragement from fam-
ily of Cabrera, Nora, et al. (1992) and Cabrera et al. (1993) within the specifica-
tion of a new model introduced here. A five-category Likert scale, strongly

552 SANDLER

agree to strongly disagree was employed. The Cronbach alpha reliability coeffi-
cient of the two-item scale for family encouragement was .84.
Financial Attitudes/Difficulty was comprised of two items and involved the
“experience of financial difficulty” while at the institution (Cabrera, 1988; Mal-
lette and Cabrera, 1991) and the difficulty in financing a college education (Ca-
brera, 1988; Mallette and Cabrera, 1991). As an environmental/attitudinal mea-
sure that conceptually originated with the student attrition model of Bean and
Metzner (1985), financial difficulty was explored to ascertain the impact finan-
cial hardship has on the lives of adult learners. A five-category Likert scale,
strongly agree to strongly disagree was employed. The Cronbach alpha reliabil-
ity coefficient of the two-item scale for financial attitudes/difficulty was .69.
Financial Attitudes/Satisfaction, another environmental/attitudinal measure
introduced in the student attrition model (Bean and Metzner, 1985) was em-
ployed in this investigation to examine the degree to which the institution is
able to satisfy the financial needs of adult students. Financial Attitudes/Satisfac-
tion was comprised of two items that involved the satisfaction with the amount
of financial support (grants, loans, family and jobs) received while attending the
institution (Cabrera, Nora, et al., 1992; Cabrera et al., 1993) and the satisfaction
with financial aid programs at the institution as expressed by students (Mallette
and Cabrera, 1991). A five-category Likert scale, strongly agree to strongly
disagree was employed. The Cronbach alpha reliability coefficient of the two-
item scale for financial attitudes/satisfaction was .84.
Goal Commitment was comprised of two items introduced in the student inte-
gration model. It concerned the importance students ascribe to a college degree
and the “importance of completing program of study” (Pascarella and Terenzini,
1979, 1980; Cabrera et al., 1993). An examination of adult student goals and
commitment to the college degree was critical in light of the orthogonal con-
struct of career decision-making self-efficacy that was explored in this study. A
five-category Likert scale, strongly agree to strongly disagree was employed.
The Cronbach alpha reliability coefficient of the two-item scale for goal com-
mitment was .66.
Institutional Commitment was comprised of four items regarding the student
integration model and involved the confidence students have in their institu-
tional choice, and their perceptions of “institutional fit and quality” (Pascarella
and Terenzini, 1979, 1980; Cabrera et al., 1993). Institutional commitment con-
cerned the feelings of attachment or belonging that students establish with the
institution. The degree to which adult students form an institutional bond was
considered important due to the environmental and career development compo-
nents that were incorporated with this investigation. A five-category Likert scale,
strongly agree to strongly disagree was employed. The Cronbach alpha reliability
coefficient of the four-item scale for institutional commitment was .78.


Intent to Persist was conceptually introduced in the student attrition model of
Bean and Metzner (1985) and was deployed here to examine its impact on adult
students in light of new theoretical considerations regarding the theory of
planned behavior (Ajzen, 1991) and related variables introduced within a new
structural model. It was comprised of four items and involved the likelihood in
re-enrolling at the institution as expressed by students (Cabrera et al., 1993). A
five-category Likert scale, strongly agree to strongly disagree was employed.
The Cronbach alpha reliability coefficient of the four-item scale for intent to
persist was .69.
Perceived Stress examines the stress experienced by adult students and was
originally conceptualized among other psychological variables by Bean and
Metzner (1985). As a mediating variable of intent to persist, Chartrand (1992)
provides evidence that suggest its inclusion here. Unlike Chartrand, it mea-
sures the amount of stress adult students perceive due to the energy they expend
and the amount of work that college requires. A two-item scale was included
that originated with a survey developed by Cabrera (1988). A five-category
Likert scale, strongly agree to strongly disagree was employed. The Cron-
bach alpha reliability coefficient of the two-item scale for perceived stress
was .85.
Persistence, the principal variable of this inquiry, is determined by the actual
re-enrollment at the institution for the following term of study.
Social Integration was comprised of two items that conceptually originated
with the student integration model. The scale examined the experience adult
students have in making “close personal friendships” and their “ease of meeting
and making friends” in college (Cabrera et al., 1993). Social integration con-
cerns the feelings of being a part of the social life of the institution, as expressed
by students, and was examined here due to the introduction of a new career
development construct that was deployed in tandem with constructs from the
student integration and student attrition models. A five-category Likert scale,
strongly agree to strongly disagree was employed. The Cronbach alpha reliabil-
ity coefficient of the two-item scale for social integration was .73.

Tests for Univariate and Multivariate Normality
Tests of univariate normality and multivariate normality were conducted by
PRELIS 2 (Joreskog and Sorbom, 1993). At prescreening, PRELIS 2 revealed
¨ ¨
a moderate level of kurtosis and skewness among the variables to be investi-
gated. That is, the omnibus test employed by PRELIS 2 revealed a moderate
level of multivariate non-normal data (Joreskog and Sorbom, 1993). Indications
¨ ¨
of non-normality were suspected because of the ordinal and dichotomous vari-

554 SANDLER

ables that were investigated. PRELIS 2 was engaged to prepare the data, that is,
to compute various transformations of the variables so that structural modeling
could proceed using LISREL 8.14 (Joreskog and Sorbom, 1993). An alternative
¨ ¨
estimator was utilized by employing the weighted least squares (WLS) method
to serve as an adjustment for the multivariate non-normal conditions encoun-
tered (Joreskog and Sorbom, 1993). That is, the WLS estimator employed,
¨ ¨
“allows for non-normality and is asymptotically efficient” (Bollen, 1989). The
order of data analysis procedures that include PRELIS 2 is addressed in the next
section.

Data Analysis Procedures
A two-step data analysis was conducted that included measurement and struc-
tural stages. The measurement stage was performed separately with SPSS 6.13
(Norusis, 1994). Cronbach alpha reliability coefficients are provided for nine
scales listed in the variables in the study section above. After a reliability analy-
sis was completed on the respective endogenous variable scales, data reduction
was performed by means of a principal components procedure on the items
of these same variable scales with SPSS 6.13 (Norusis, 1994). The principal
components procedure incorporated a factor extraction with varimax rotation
and computed factor scores for each respondent, single numerical values that
were equivalent to the respective scale data of the endogenous variables. The
principal components procedure and computation of factor scores served as a
measurement stage for the structural equation path model that followed. A sum-
mary table of descriptive statistics that include means and standard deviations
of all the variables employed is presented in Table 1 below.
Once reliable scale data for the endogenous variables were computed and
subsequently reduced to single numerical values (factor scores) for each respon-
dent by means of the data reduction procedure described above, PRELIS 2 was
employed. PRELIS 2 produced data transformations among ordinal and continu-
ous variables and provided appropriate covariance matrices that included the
asymptotic covariance matrix (Joreskog and Sorbom, 1993). Polychoric, polys-
¨ ¨
erial, and product moment correlations were included in the computation of
these matrices to enable an advanced estimation of the asymptotic covariance
matrix for arbitrary non-normal distributions (Browne, 1982, 1984). A correla-
tion matrix among all the variables of this investigation is provided in the Ap-
pendix. The mixture of the aforementioned correlations did not cause problems
that pertain to a positive definite matrix. That is, the measurement procedure
employed above permitted the data transformation by PRELIS 2 to continue
without a computational failure on the listwise deleted research sample of 469.


TABLE 1. Summary Descriptive Statistics of the Endogenous and
Exogenous Variables (N = 469 [Listwise Deletion])

Meanc Std. Dev.
Endogenous Variable
Career Decision-Making Self-Efficacy .00 33.40
Family Encouragementa .00 1.00
Perceived Stressa .00 1.00
Financial Attitudes/Satisfactiona .00 1.00
Financial Attitudes/Difficultya .00 1.00
Academic Integrationa .00 1.00
Social Integrationa .00 1.00
Institutional Commitmenta .00 1.00
Cumulative GPA .00 .81
Goal Commitmenta .00 1.00
Intent to Persista .00 1.00
Persistence .00 .40
Exogenous Variable
Gender .00 .45
Race/Ethnicity .00 .50
Household Income .00 3.21
Relatives/Dependents .00 1.13
Financial Aid .00 2492.33
Parents’ Educational Level .00 2.24
Academic Aspirations .00 .65
Student Type .00 .49
Degree Program .00 .50
Curriculum Hours .00 3.31
Hours Employed .00 14.91
a
Principal components factor extraction has been performed as indicated in related text section.
c
All variables have been centered.

Parameter estimation followed and was computed with LISREL 8.14 (Joreskog
¨
and Sorbom, 1993), which employed structural equation modeling procedures.
¨
The structural parameter estimation procedures and path analytic protocol of
LISREL 8.14 followed using a WLS method that adjusted for non-normal condi-
tions when one or more of the observed variables are ordinal (Joreskog and
¨
Sorbom, 1993). In short, LISREL 8.14 incorporated and processed the measure-
¨
ment stage data and simultaneously computed specified structural equations and
a path model (Joreskog and Sorbom, 1993).
¨ ¨

556 SANDLER

RESULTS
The total effects among the endogenous variables in the integrated model of
student persistence, above an effect size criterion of .10, are summarily shown
in Figure 2. Standardized effects .10 or greater have a “meaningfulness” that
can influence policy analysis. This cutoff criterion approximates those com-
monly found in the literature (Pedhazur, 1996; Stevens, 1996; Hoyle, 1995;
Loehlin, 1992; Cohen and Cohen, 1983). The standardized total effect size crite-
rion of .10 suggests that a unit change in the total effect of a given endogenous
or exogenous variable is associated with at least a 10 percent change (or more)
on a given endogenous variable (or dependent outcome) examined. The trimmed
display of the model among the endogenous variables (see Figure 2) includes a
total of 42 endogenous path relationships greater than the effect size criterion
of .10. The trimmed display of the model is also presented as a final structural
model for the Career Decision-Making Self-Efficacy, Perceived Stress, and an
Integrated Model of Persistence. In addition, 32 exogenous variable path rela-
tionships of the endogenous variables were included and controlled.

Goodness-of-Fit Statistics
As a structural model, Career Decision-Making Self-Efficacy, Perceived
Stress, and an Integrated Model of Persistence has close to a “perfect fit” (chi-
square = 136.719 with 173 degrees of freedom; p = .981) (Joreskog and Sorbom,
¨ ¨
1993). Other indicators of goodness-of-fit are provided below to more compre-
hensively assess overall goodness-of-fit for the model (Joreskog and Sorbom,
¨ ¨
1993). These indicators included: chi-square/degrees of freedom ratio (.790),
goodness-of-fit index (GFI = .994), adjusted goodness-of-fit index (AGFI =
.990), and the root mean square residual (RMR = .0596).

The Squared Multiple Correlation (R2) for Each
Endogenous Variable
The explained variance determined by the Squared Multiple Correlation (R2)
for each endogenous variable of the integrated model of student persistence is
as follows: CDMSE (8%), family encouragement (27%), perceived stress (8%),
financial attitudes/satisfaction (16%), financial attitudes/difficulty (19%), aca-
demic integration (18%), social integration (27%), institutional commitment
(39%), cumulative GPA (11%), goal commitment (12%), intent to persist (65%),
and persistence (43%).

FINANCES, ATTITUDES, BEHAVIOR, AND CAREER DEVELOPMENT
FIG. 2. Career decision-making self-efficacy, perceived stress, and an integrated model of student persistence: total effects among the
endogenous variables within a trimmed display of the modelb

557
b
The conventional syntax used in path diagrams may be deviated from in order to simplify representation.

558 SANDLER

The Total, Direct, and Indirect Effects on the
Endogenous Variables of CDMSE
Most importantly, within a nonrecursive structural path model, CDMSE has
significant total effects on all twelve endogenous variables within the model.
Six of these relationships are above .10 in magnitude, within a trimmed display
of structural effects, and are listed in order of descending magnitude; four path
relationships arise largely from direct effects. The total, direct, and indirect ef-
fects of CDMSE on the following endogenous variables are: academic integra-
tion (total effect = −.230, p < .001, direct effect = −.217, p < .001, indirect effect
= −.013, p < .001), social integration (total effect = −.197, p < .001, direct effect
= −.142, p < .001, indirect effect = −.055, p < .001), intent to persist (total effect
= .197, p < .001, direct effect = .130, p < .001, indirect effect = .067, p < .001),
perceived stress (total effect = .139, p < .001, direct effect = .157, p < .001, indi-
rect effect = −.018, p < .001), institutional commitment (total effect = −.110, p <
.001, direct effect = .000, indirect effect = −.110, p < .001), and persistence (to-
tal effect = .107, p < .001, direct effect = .000, indirect effect = .107, p < .001),
the ultimate dependent variable within the integrated model (see Figure 2). The
total effects of CDMSE on institutional commitment and persistence are small
and are composed largely of indirect effects.
An examination of the total effects by CDMSE on endogenous variables
within the integrated model suggests that cognitive-initiated agentive behavior
related to the career development of adult students negatively affects the feeling
nontraditional learners express about being a part of the academic life of the
institution (academic integration) at a moderate level. CDMSE also negatively
affects the feelings expressed by adult students about being a part of the social
life of the institution (social integration) at a moderate level. In addition, the
career decision-making self-efficacy of adult students has a relatively small total
effect on the perceived stress of nontraditional learners. CDMSE also has a
small negative effect on the degree of affiliation experienced by adult students
regarding the institution in which they are enrolled (institutional commitment).
Two remaining effects of intent to persist and persistence on CDMSE are posi-
tive path relationships. Both paths are mentioned in the following sections.

The Total, Direct, and Indirect Effects Explaining Intent to Persist
The Squared Multiple Correlation (R2) explaining the variance in intent to
persist was high at 65 percent. The four highest ranked total effects on intent to
persist of the endogenous variables, within the trimmed display of the integrated
model of student persistence (effects > .10), are listed as follows in descending
order of magnitude and arise largely from direct effects; the indirect effects are


very small or marginal in size (see Table 2 and Figure 2): (1) institutional
commitment (total effect = −.273, p < .001), (2) academic integration (total ef-
fect = −.253, p < .001), (3) CDMSE (total effect = .197, p < .001), and (4) finan-
cial attitudes/difficulty (total effect = .173, p < .001). The degree of affiliation
or identification adult students experience with the institution (institutional com-
mitment) has a negative total effect on the intent to persist of nontraditional
learners at a moderate level. In addition, the degree of feeling of being a part
of the academic life of the institution (academic integration) of adult students
has a negative moderate total effect on the intent to persist of nontraditional
learners at a relatively moderate level. Also, the cognitive-initiated agentive
behavior about career tasks (CDMSE) of nontraditional learners and their atti-
tudes about financial difficulty (financial attitudes/difficulty) have small total
effects on the intent to persist of adult students respectively.
In addition, the five highest ranked total effects on intent to persist of the

TABLE 2. Total, Direct, and Indirect Effects on Intent to Persist of the Endogenous
Variables with the Exogenous Variables Controlled (N = 469)

Rank of
Indirect Direct Total Total
Effectb Effecta Effectb Effect@
Career Decision-Making Self-Efficacy .067*** .130*** .197*** 3
Family Encouragement −.035*** .000c −.035*** 8
Perceived Stress .065*** .000c .065*** 7
Financial Attitudes/Satisfaction −.075*** .000c −.075*** 6
Financial Attitudes/Difficulty .036*** .137*** .173*** 4
Academic Integration −.110*** −.143*** −.253*** 2
Social Integration .000 .000c .000
Institutional Commitment −.017*** −.256*** −.273*** 1
Cumulative GPA −.004 .000c −.004
Goal Commitment −.087*** .000c −.087*** 5
Intent to Persist .015*** .000c .015*** 9
Persistence — — — —
R2 = .646.
***p < .001.
a
Standardized structural coefficient.
b
computed from standardized structural coefficient.
c
Fixed or constrained relationship determined by hypotheses, or with subsequent theoretical reassess-
ment adjusted by the researcher, as a constrained relationship, to facilitate structural equation model
fitting.
@
Total effect values at the p < .001 level of significance are ranked in order of descending magni-
tude.

560 SANDLER

exogenous variables, within the trimmed display of the integrated model of stu-
dent persistence (effects > .10), are listed as follows in descending order of mag-
nitude and arise largely from direct effects; the indirect effects are very small
in magnitude (see Table 3): (1) household income (total effect = .720, p < .001),
(2) financial aid (total effect = .257, p < .001), (3) degree program (total effect
= .155, p < .001), (4) relatives/dependents (total effect = −.142, p < .001), and
(5) gender (total effect = .122, p < .001). Household income is a sociostructural
determinant of the intent to persist of nontraditional students at a high level, as
the standardized total effect suggests. This path represents the largest effect size
obtained in this investigation. Financial aid has a moderate total effect on the
intent to persist of adult students, whereas degree program, relatives/dependents,
and gender have small total effects.

TABLE 3. Total, Direct, and Indirect Effects on Intent to Persist of the
Exogenous Variables (N = 469)

Rank of
Gender −.008 .130*** .122*** 5
Ethnicity/Race .009 .000c .009
Household Income .050*** .670*** .720*** 1
Relatives/Dependents −.001 −.141*** −.142*** 4
Financial Aid −.041*** .298*** .257*** 2
Parents’ Educational Level .021** .000c .021** 7
Academic Aspirations −.011 .000c −.011
Student Type −.035*** .000c −.035*** 6
Degree Program .011 .144*** .155*** 3
Curriculum Hours — .000c —
Hours Employed –.004** .000c −.004** 8
R2 = .646.
**p < .01.
***p < .001.
a
b
c
fitting.
@
tude.


The Total, Direct, and Indirect Effects Explaining Persistence

The Squared Multiple Correlation (R2) explaining the variance in persistence
was moderate at 43 percent. The seven highest ranked total effects on persis-
tence of the endogenous variables, within the trimmed display of the integrated
model of student persistence (effects > .10), are listed as follows in descending
order of magnitude (see Table 4 and Figure 2): (1) intent to persist (total effect
= .660, p < .001), (2) institutional commitment (total effect = −.203, p < .001),
(3) academic integration (total effect = −.172, p < .001), (4) social integration
(total effect = .147, p < .001), (5) family encouragement (total effect = −.140, p
< .001), (6) financial attitudes/difficulty (total effect = .119, p < .01), and (7)
CDMSE (total effect = .107, p < .001). The total effects on persistence regarding
intent to persist, social integration and family encouragement arise from direct
effects, whereas the total effects on persistence of institutional commitment,
academic integration, financial attitudes/difficulty, and CDMSE are composed
entirely of indirect effects.
In summary, the intent to persist of nontraditional learners has a strong total
effect on persistence or the re-enrollment for a subsequent term of study by
adult students; it is entirely composed of a direct effect. In addition, this path
represents the second largest effect size obtained in this investigation. The feel-
ings of affiliation or identification with the institution (institutional commit-
ment) experienced by adult students has a negative effect on persistence at a
moderate level; it is entirely an indirect effect. Also, feelings of being a part of
the academic life of the institution (academic integration) and social support
provided by family members of adult learners (family encouragement) have
negative total effects on persistence at low levels respectively. The effects of
academic integration on persistence are largely indirect, whereas the effects of
family encouragement on persistence are largely direct. Perhaps for the first
time reported, the feeling(s) of being a part of the social life of the institution
(social integration) is a positive total effect of persistence at a low level for
adult learners; this effect is largely a direct effect. Finally, financial difficulty
and CDMSE have low total effects on the persistence of adult students; each
effect is indirect.
In addition, the five highest ranked total effects of persistence on the exoge-
nous variables, within the trimmed display of the integrated model of student
persistence (effects > .10), are listed as follows in descending order of magni-
tude and arise largely from direct effects in the structural model with the excep-
tion of financial aid that is entirely an indirect effect (see Table 5): (1) degree
program (total effect = −.270, p < .001), (2) curriculum hours (total effect =
.191, p < .001), (3) hours employed (total effect = .140, p < .001), (4) financial
aid (total effect = .137, p < .001), and (5) student type (total effect = .100, p <

562 SANDLER

TABLE 4. Total, Direct, and Indirect Effects on Persistence of the
Endogenous Variables with the Exogenous Variables Controlled (N = 469)

Rank of
Career Decision-Making Self-Efficacy .107*** .000c .107*** 7
Family Encouragement −.019*** −.121*** −.140*** 5
Perceived Stress .030*** .000c .030*** 11
Financial Attitudes/Satisfaction −.031** .000c −.031** 10
Financial Attitudes/Difficulty .119*** .000c .119*** 6
Academic Integration −.172*** .000c −.172*** 3
Social Integration −.010** .157*** .147*** 4
Institutional Commitment −.203*** .000c −.203*** 2
Cumulative GPA −.006 .097*** .091*** 8
Goal Commitment −.059*** .000c −.059*** 9
Intent to Persist .010*** .650*** .660*** 1
Persistence — — — —
R2 = .427.
**p < .01.
***p < .001.
a
b
c
fitting.
@
tude.

.001). Degree program has a moderate total effect on the persistence of adult
students. Unlike associate degree program students, Bachelor of Science (BS)
and Bachelor of Arts (BA) adult degree students have a moderate level of
within-year persistence (from one semester to the next) at the institution. Curric-
ulum hours, an exogenous variable, has a relatively moderate total effect on the
persistence of adult students. The total effect of curriculum hours provides a
quantitative indication of the involvement of adult students, which is entirely
composed of the direct effect on persistence of the number of enrolled course
hours. Finally, three remaining exogenous variables—hours employed, financial
aid, and student type—have small total effects on the persistence of adult stu-
dents. Hours employed and student type are largely direct effects, whereas the
effect of financial aid is indirect.


TABLE 5. Total, Direct, and Indirect Effects on Persistence of the
Exogenous Variables (N = 469)

Rank of
Gender .099*** .000c .099*** 6
Ethnicity/Race .020** .000c .020** 9
Household Income .468*** −.538*** −.070
Relatives/Dependents −.093*** .000c −.093*** 7
Financial Aid .137*** .000c .137*** 4
Parents’ Educational Level .073*** −.145** −.072 8
Academic Aspirations −.011 .000c −.011
Student Type −.021*** .121*** .100*** 5
Degree Program .100*** −.371*** −.270*** 1
Curriculum Hours — .191*** .191*** 2
Hours Employed −.015** .156*** .140*** 3
R2 = .427.
**p < .01.
***p < .001.
a
b
c
fitting.
@
tude.

DISCUSSION AND CONCLUSION
Adult student persistence decisions are more comprehensively explained by
the new integrated model explored in this investigation, through the inclusion
and dynamic interplay of the theory of planned behavior (Ajzen, 1991) sub-
sumed therein. An examination of the effects of the endogenous variables on
CDMSE (see Figure 2), the effects of intent to persist and persistence (see Ta-
bles 2 , 3, 4, and 5), and the explained variance computed for all twelve endoge-
nous variables support this conclusion.

Discussion About the Findings of Career Decision-Making
Self-Efficacy Among the Endogenous Variables
Of the twelve endogenous variables within the present study, career decision-
making self-efficacy has the widest range of influence as an endogenous vari-
able. It affects all twelve endogenous variables at significant levels. Further, six

564 SANDLER

of the twelve endogenous variables have effects on the perceived CDMSE of
adult students above a total effect size of .10 within the trimmed display of the
integrated model of student persistence (see Figure 2). Three of these endoge-
nous variables have moderate total effects of career decision-making self-effi-
cacy that are largely composed of direct effects: academic integration, social
integration, and intent to persist. The remaining three endogenous variables:
perceived stress, institutional commitment, and persistence, the central variable
within the integrated model of student persistence, have lower total effects of
CDMSE. The effect on persistence of CDMSE is indirect, whereas the effects
on perceived stress and institutional commitment of CDMSE are direct.
The career decision-making self-efficacy of adult students negatively influ-
ences the feelings they express about being a part of the academic life and social
life of the institution, and positively influences their interest and attitudes in re-
enrolling at the institution for a subsequent term of study. All three effects are
recorded at moderate levels. In addition, the CDMSE of adult students affects
their perceptions of stress engendered due to the amount of work that college
requires, negatively impacts their feelings of institutional affiliation and identifi-
cation, and positively influences their actual re-enrollment at the institution, or
persistence, at a low level.
From these findings, clear policy implications arise for higher education insti-
tutions that provide undergraduate degree programs for adult/nontraditional stu-
dents. The academic and social systems of the adult undergraduate experience
must be more attuned to adult students’ percepts of confidence about their voca-
tional futures. In order to be effective and efficient, institutions need to help
adult students achieve their goals, and assist nontraditional learners with the
critical developmental task of career decision making and planning that remains
at odds with the academic and social integration of adult students and their
feelings of belonging. The liberal arts and professional curriculum need to be
made relevant as it relates to the larger environment that affects adult lives by
directly linking the curriculum with the world of work.
Since CDMSE has negative effects on adult students’ feeling of being a part
of the academic life and social life of the institution at moderate levels, the
faculty must develop strategies for excellence in teaching and cocurricular pro-
gramming that incorporate collaborative ties with the environment beyond the
institution walls. Beyond the calls for cultural literacy and diversity that are
important aims in and of themselves, the curriculum needs to address workplace
reality by establishing more direct linkages with emerging professions, new
technologies, cooperative education, internships, and continuing professional ed-
ucation for adult undergraduates. Most importantly and assuredly, the confi-
dence adult students express about their vocational futures affects their attitudes
about re-enrolling at the institution, as a direct effect, and to a lesser extent their
actual persistence at the institution, as an indirect effect. Albeit, the perceived


confidence about the vocational futures of adult students increases their percep-
tion of stress about the work college requires to a small degree, as the effect
that is largely direct indicates.

Discussion About the Effects Explaining Intent to Persist
The current study includes vital new constructs central to the assessment of
adult student attitudes about persistence. In addition, elements of traditional and
nontraditional theories are brought into question with new evidence explored
here regarding adult students. The synthesis of Cabrera et al. (1993) is adapted
and expanded upon by incorporating the theory of planned behavior (an update
of the theory of reasoned action), the constructs of perceived behavioral control
about career tasks (CDMSE), and nonvolitional control (financial attitudes/diffi-
culty) (Ajzen, 1991; Eagly and Chaiken, 1993).
The total effects on intent to persist include the effects of four endogenous
variables (CDMSE, financial attitudes/difficulty, academic integration, and insti-
tutional commitment) and the effects of five exogenous variables (gender,
household income, relatives/dependents, financial aid, and degree program). In-
tent to persist concerns the attitudes about persistence that adults express for the
subsequent term of study. As suggested by the data and modification indices of
LISREL 8.14 (Joreskog and Sorbom, 1993), the endogenous relationships of
¨ ¨
intent to persist are explored and empirically verified here within a nonrecursive
system of path linkages.
First, the perceived career decision-making self-efficacy of adult students has
a moderate total effect on their intent to persist; this path is largely composed
of a direct effect. In other words, an increase in the exercise of perceived control
over the vocational futures and career-related tasks of adult students (Ajzen,
1991; Eagly and Chaiken, 1993) results in an increase of the total effect on the
intent to persist of nontraditional learners. Second, the experience of financial
difficulty or hardship expressed by adult students results in a low-level increase
of the total effect on intent to persist of nontraditional learners; this path is
largely composed of a direct effect. Third, the feelings adult students express
about being a part of the academic life of the institution (academic integration),
negatively influence their attitudes about persistence and re-enrollment for a
subsequent term of study at a moderate level, as the total effect that is largely
a direct one in composition portends.
Academic integration originated with the student integration model of Tinto
(1975, 1987, 1993). Little empirical evidence published in the literature to date
(Braxton et al., 1997) has confirmed the effect of academic integration (a con-
struct that originated with Tinto) on intent to persist albeit in a negative manner
(a construct purported by Bean and Metzner, 1985), as the research conducted
here empirically substantiates. The findings that pertain to the classroom and

566 SANDLER

the academic integration of adult students should alert the faculty and adminis-
trators to take stock of the negative feelings of academic affiliation that nontra-
ditional students experience at a moderate level. Careful attention needs to be
paid to the lives of adult learners, to the complex multiplicity of roles they
encounter as nontraditional learners with respect to the student/learner context
of undergraduate education. In light of the negative effect of academic integra-
tion on the intent to persist of adult students, academic tasks, curriculum en-
hancements and cocurricular activities need to be carefully assessed and devel-
oped for this population by the faculty and administrators. These traditional
interventions may risk having a negative impact on the attitudes of persistence
of adult students.
Finally, the affiliation or identification with the institution (institutional com-
mitment) of adult students has a negative total effect on their intent to persist at
a moderate level; this path is largely composed of a direct effect. As the findings
elucidate, the involvement of adult learners at the institution decreases moder-
ately their intent to persist, that is, their attitudes about re-enrollment for a
subsequent term of study. Findings of persistence discussed below modify this
conclusion. A simple brush stroke assessment cannot be prematurely drawn re-
garding the effects of intent to persist alone. No definitive or final judgment can
be made here without looking at the effects on persistence that follow.
Three out of five of the exogenous variables that have total effects on intent to
persist directly concern the socioeconomic status of the adult students seeking an
undergraduate education: household income, relatives/dependents, and financial
aid. As a sociostructural determinant of adult students’ attitudes about persistence,
household income has a high total effect on intent to persist. It is largely composed
of a direct effect that is responsible for the largest total effect reported within the
integrated model of student persistence. As the standardized total effect of the
structural equation model suggests, a unit increase of household income is associ-
ated with approximately a 72 percent increase in the intent to persist of adult
students. That is, the higher the level of household income, the higher the level
of intent to persist expressed by adult students at the institution. Baker and Velez
(1996), in their review of the research literature on access and persistence in
higher education, emphasize the “declining importance of socioeconomic advan-
tage.” However, this investigation provides evidence that household income is a
critical sociostructural determinant of intent to persist. Although the research of
Peterson (1993a, b) involving CDMSE and institutional integration supports the
observations of Baker and Velez, the empirical data reported here involving
household income and intent to persist clearly suggests otherwise.
Relatives/dependents has a small total effect; it is largely composed of a direct
effect indicating the burden experienced by adult students with dependents. Nev-
ertheless, as an equilibriator, financial aid assists nontraditional students in need
and those who elect to borrow; this path is largely composed of a direct effect.


The total effect of financial aid positively influences adult students’ attitudes
about persistence at a moderate level, thereby increasing their available re-
sources for participation and inclusion regarding the attainment of an undergrad-
uate education. Two remaining exogenous variable effects on intent to persist,
gender and degree program, are small, bearing evidence that favors the intent
to persist of men and the stronger intent to persist of two-year degree students.
With the additions of career decision-making self-efficacy, financial difficulty
and all the effects included in the structural equation, the explained variance of
intent to persist is evaluated at a moderate to high level (65 percent). It sur-
passed, by a respectable margin, the level of explained variance of intent to
persist examined in the integrated model of retention (“43 percent”) of Cabrera
et al. (1993), that was tested with a traditional population of students. In conclu-
sion, perceived behavioral control (Ajzen, 1991; Eagly and Chaiken, 1993), an
analog for CDMSE (Bandura, 1997, 1989, 1977; Betz et al., 1996) is integrated
within a new model to better explain nontraditional students’ intent to persist
and the processes subsumed therein.

Discussion About the Effects Explaining Persistence
First, the total effect on persistence of career decision-making self-efficacy is
small and is composed solely of indirect effects. This finding is new to the field
of higher education and the persistence research on adult undergraduate stu-
dents. The vocational futures and career expectations of adult students affect
their re-enrollment at the institution at a low level, representing one of six signif-
icant total effect findings of CDMSE among the endogenous variables in the
trimmed display of the integrated model (see Figure 2).
The integrated model of student persistence moves beyond the traditional
sociological schemata of Tinto (1975, 1987, 1993), the nontraditional conceptual
model of student attrition of Bean and Metzner (1985) and the synthesis of the
converged constructs and conceptual framework of the integrated model of stu-
dent retention (Cabrera et al., 1993). According to the findings of this research
investigation, adult students are choosing to persist by employing their per-
ceived career decision-making self-efficacy within an integrated structural
framework that addresses nontraditional learners. That is, this research suggests
that the vocational futures of adult students in the workplace include the larger
environment in which they participate in households and families. Comprehen-
sive nonrecursive feedback loops are deployed to reach the environment in new
inclusive ways, by employing a social interactionist perspective. The intercon-
nectedness of variable relationships that have not been looked at before are
empirically tested and explained.
Second, the intent to persist of adult students affects their persistence at the
institution at a moderate to high level. The total effect on persistence of intent

568 SANDLER

to persist is largely composed of a direct effect. As the standardized total effect
of the structural equation model suggests, a unit increase of intent to persist is
associated with approximately a 66 percent increase in the persistence of adult
students. This large effect size from intent to persist is counterindicated by the
negative effects on persistence of three other endogenous variables. Of these,
two effects have conceptual origins from the student integration model of Tinto
(1975, 1987) (institutional commitment and academic integration), whereas a
third stems from the student attrition model of Bean and Metzner (1985) (family
encouragement). The affiliation or identification of adult students with the institu-
tion (institutional commitment) and the feelings of being a part of the academic
life of the institution (academic integration) have negative total effects on persis-
tence at a moderate and low level, respectively. Both paths are composed entirely
of indirect effects. In addition, the social support that family members provide
(family encouragement) has a negative total effect on the persistence of adult
students at a low level; this path is largely composed of a direct effect.
Adult student bonds of affiliation with the institution (institutional commit-
ment) decreases their re-enrollment for a subsequent term of study (persistence)
at a low level. This relatively small indirect effect constitutes a new empirical
finding that the integrated model of student retention of Cabrera et al. (1993)
did not explore. Cabrera et al. (1993) and Bean and Metzner (1985) conceptually
assumed that the effects on the affiliation of adult students with the institution
were mediated through “intent to persist.” This investigation reveals that the
total effects on institutional commitment are indeed mediated through the atti-
tudes of persistence (intent to persist) of adult students. In addition, this investi-
gation empirically documents that the total effects on persistence of institutional
commitment are composed largely of negative effects, as the path in Figure 2
illustrates (also see Table 4). There are clear implications regarding this empiri-
cal evidence. Adult bonds of affiliation with the institution are an important
component that enrollment managers and student affairs professionals must pay
closer attention to when assisting nontraditional learners with advisement infor-
mation and registration. This is particularly important since electronic and tele-
phone registration techniques that are widely employed by a large number of
institutions risk being too impersonal or laissez-faire, and may contribute to the
weakening affiliation process.
The feelings of being a part of the academic life of the institution of adult
students also decrease moderately their persistence at the institution. The aca-
demic system of undergraduate student life is apparently not well calibrated for
nontraditional learners and may interfere with their perceptions of satisfaction.
The multiple roles adult students experience in the workplace and as family
members may contribute to this negative response. Certainly, as this investiga-
tion uncovers, support received from family members decreases the re-enroll-
ment of adult students at the institution at a low level.


Two other endogenous variable effects on persistence that are included in
the trimmed display of the model are explored and empirically verified. The
experiences of financial difficulty can positively affect the persistence of adult
students at the institution, albeit at a low level. This path is composed entirely
of indirect effects. As a reprieve, adult students nevertheless do find that their
social relations with other adult students and their social life at the institution
increase their persistence, a new finding that is empirically substantiated at a
low level. In summary, adult students are persisting against difficult odds in an
institutional system that is not well calibrated for the multiple roles of adults.
Adult students do receive positive support from their peers (social integra-
tion)—a new finding of persistence that is largely composed of a direct effect.
Assumptions that originate from traditional and nontraditional conceptual
models of student departure are complexly refuted here, particularly regarding
the negative effects described above. These negative effects add to a further
questioning of “appropriateness” concerning the usage of both traditional mod-
els and nontraditional models in the examination of adult undergraduate persis-
tence. These negative effects reach beyond the deployment of the construct of
academic performance alone, about which Kasworm and Pike (1994) expressed
caution. Academic performance (cumulative GPA) also proved to be of little
importance regarding adult student persistence as it is investigated here. More
importantly, new compensatory effects are present, as the final structural model
explicates. In particular, the positive compensatory effect of persistence on so-
cial integration makes up for the negative environmental effect of persistence
on family encouragement, albeit by a very small margin. Although elements of
the theories of Tinto and Bean are refuted, the findings bear important meaning
with regard to new compensatory effects that are empirically demonstrated, and
also with regard to the deconstruction of factors that are critical to adult student
persistence. Taken together, the positive total effects of persistence on social
integration and on career decision-making self-efficacy act to counterbalance
the negative effects of persistence on family encouragement and on institutional
commitment. That is, these compensatory interaction effects described above
attest to the positive capacities of self-efficacy and individuation throughout the
life span and the affirming role that lifelong learning can bring to social adapta-
tion (social integration) within the context of adult undergraduate education. As
for the examination of the deconstruction of factors mentioned above, hypotheti-
cal elements of precursor theories relate differently in this investigation to pro-
duce new or alternative outcomes. Nevertheless, the explained variance of the
constructs that originated with the student integration and the student attrition
models remain at a respectable size and cannot be deemed inappropriate or
discounted for academic performance, as Kasworm and Pike (1994) conclude.
As the findings indicate, the complexity of adult student persistence may require
a complete paradigm shift beyond Tinto, Bean, and the synthesis of Cabrera et

Martin E. Sandler, Ph.D., Research In Higher Education Article, October 2000

Martin E. Sandler, Ph.D., Research In Higher Education Article, October 2000

Recomendados

Recomendados

Mais conteúdo relacionado

Destaque

Destaque (7)

Semelhante a Martin E. Sandler, Ph.D., Research In Higher Education Article, October 2000

Semelhante a Martin E. Sandler, Ph.D., Research In Higher Education Article, October 2000 (20)

Martin E. Sandler, Ph.D., Research In Higher Education Article, October 2000