Implications of losing a need- and merit-based scholarship on the educational trajectory: a curricular analytics approach

Abstract

Understanding how students with low socioeconomic status finance their tuition over time can help us comprehend the impact of students’ decisions on their subsequent curricular progress, graduation, or dropout. This work presents a curricular analytics approach using process mining techniques to study educational funding trajectories as processes. Specifically, the SCHOLARSHIP-LOAN-SELF-FUNDED model is designed to reveal educational funding trajectories and obtain aggregate information. Academic and tuition records of 2484 undergraduate students from a private Chilean university who started their programs with a government need- and merit-based tuition aid were analyzed. Students who lost their scholarships were more likely to drop out, whereas students who maintained this aid were more likely to graduate on time. Curricular progress per semester was slower after scholarships ended or after the students lost them and stayed. Financial aid was associated with students’ curricular progress and linked to their permanence and graduation time. Higher education institutions should consider the eligibility criteria and maintenance requirements of financial assistance when designing their curricula.

Full text

Vol.:(0123456789)
Higher Education (2025) 89:441–464
https://doi.org/10.1007/s10734-024-01230-0
1 3
Implications oflosing aneed‑ andmerit‑based scholarship
ontheeducational trajectory: acurricular analytics approach
JuanPabloSalazar‑Fernandez1,2 · JorgeMunoz‑Gama1 · MarcosSepúlveda1
Accepted: 21 April 2024 / Published online: 21 May 2024
© The Author(s) 2024
Abstract
Understanding how students with low socioeconomic status finance their tuition over time
can help us comprehend the impact of students’ decisions on their subsequent curricular
progress, graduation, or dropout. This work presents a curricular analytics approach using
process mining techniques to study educational funding trajectories as processes. Specifi-
cally, the SCHOLARSHIP-LOAN-SELF-FUNDED model is designed to reveal educa-
tional funding trajectories and obtain aggregate information. Academic and tuition records
of 2484 undergraduate students from a private Chilean university who started their pro-
grams with a government need- and merit-based tuition aid were analyzed. Students who
lost their scholarships were more likely to drop out, whereas students who maintained this
aid were more likely to graduate on time. Curricular progress per semester was slower
after scholarships ended or after the students lost them and stayed. Financial aid was asso-
ciated with students’ curricular progress and linked to their permanence and graduation
time. Higher education institutions should consider the eligibility criteria and maintenance
requirements of financial assistance when designing their curricula.
Keywords Educational trajectories· Financial aid· Process mining· Educational data
mining· Curricular analytics
Introduction
In recent years, scholarships incorporating a need-based component have emerged as key
catalysts for expanding access to higher education, particularly among students with a
low socioeconomic status (SES) (Nguyen etal., 2019; Santelices etal., 2016). The grow-
ing demand for aid has placed pressure on governments to optimize resource utilization
and align incentives more effectively (Dynarski & Scott-Clayton, 2013). These pressures
have led to design adjustments in scholarship programs, such as introducing performance
requirements for scholarship maintenance. These adjustments have had significant impacts
* Juan Pablo Salazar-Fernandez
juansalazar@uach.cl
1 Department ofComputer Science, Pontificia Universidad Católica de Chile, Santiago, Chile
2 Instituto de Informática, Facultad de Ciencias de La Ingeniería, Universidad Austral de Chile,
Valdivia, Chile
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

442
Higher Education (2025) 89:441–464
1 3
on low-SES students. While the introduction of performance requirements may promote
timely graduation, it also introduces complexities that can adversely affect overall progres-
sion and graduation rates (Scott-Clayton & Schudde, 2020).
Researchers have investigated the impact of these aids on persistence, academic pro-
gress, and degree completion. Regarding eligibility criteria, research has shown that need-
based scholarships, compared to merit-based scholarships, have a more significant impact
on persistence and degree completion (Nguyen etal., 2019), although a combination of
need- and merit-based scholarships can also have positive effects (Nguyen etal., 2019). It
has also been found in previous research that students who receive scholarships and have
additional support mechanisms, such as mentoring and wraparound services, exhibit higher
levels of persistence and degree completion (Nguyen etal., 2019). With respect to loans,
researchers have identified factors leading low-SES students to make suboptimal decisions,
such as loan aversion (Lim etal., 2019) and the selection of institutions or programs that
are less demanding or have shorter durations (Nguyen etal., 2019), prioritizing immediate
benefits over long-term benefits. Moreover, part-time work (Thies, 2022), aversion to debt
(Lim etal., 2019), liquidity constraints associated with remaining needs (Dente & Piraino,
2011), and enrollment intensity (Goldrick-Rab etal., 2016) have been found to be factors
that could explain differences in dropout and on-time graduation rates between low-SES
students and their high-income peers.
Nevertheless, there remains limited understanding of student behavior following the
loss (Carruthers & Özek, 2016) or completion (Mabel, 2020) of their scholarships. Henry
et al. (2004) found that scholarship loss is associated with reduced curricular progress
and a decreased likelihood of graduation. Other researchers have shown that implement-
ing a performance standard of renewal for need-based scholarships might positively affect
academic progress (Scott-Clayton & Schudde, 2020). Mabel (2020) examined the impact
of implementing time limits on need-based scholarships. However, to our knowledge, no
previous research has elucidated students’ dynamic behavior after losing need- and merit-
based scholarships with a maintenance renewal standard, specifically in regard to curricu-
lar progress and degree completion.
In this article, we aim to research the educational funding trajectories of Chilean stu-
dents who have received the Bicentennial Scholarship and to analyze the relationship
between the loss of this benefit and dropout rates, curricular progress, and graduation. This
study uses a purposively selected case, employing a curricular analytics approach that lev-
erages the advantages of utilizing high-quality institutional administrative data, as well as
the ability to analyze population-level data (Figlio etal., 2016). To achieve this goal, we
propose a model to reveal educational funding trajectories and obtain aggregate informa-
tion following an adapted version of PM2, a process mining methodology (Maldonado-
Mahauad etal., 2018). By integrating curriculum and financial data, we aim to gain a com-
prehensive understanding of the dynamic behavior of students regarding tuition financing
and their progression through curricula. Specifically, we ask the following research ques-
tion: What is the relationship between the loss of a need- and merit-based scholarship and
the educational funding trajectories of students in terms of means of funding, curricular
progress, and graduation rates? We expect the obtained results to contribute to both the
related research field and the formulation of effective policies in this domain.
We conducted this study at a medium-sized Chilean university that embraces both edu-
cation and research. Although the Chilean context might have its own distinctive character-
istics, we believe that this analysis is attractive to a broader international audience. Govern-
ments are imposing accountability requirements on higher education institutions (HEIs)
that receive public funds (Lang, 2022), and the consequences of the successive changes in
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

443
Higher Education (2025) 89:441–464
1 3
how students finance their tuition in higher education in Chile have aroused international
interest (Espinoza etal., 2022; Larraín & Zurita, 2008; Santelices etal., 2016).
Literature review
Effects offinancial aid onprogress anddegree completion
The rational choice model of educational decisions (Breen & Goldthorpe, 1997) defines
three factors that influence commitment at a given transition point: the cost of continuing
to study, the perceived probability of success, and the perceived returns of each educa-
tional outcome. Low-SES students often underestimate the returns of higher education and
overestimate enrollment costs, leading them to make biased decisions (Geven & Herbaut,
2019). To address this issue, governments offer various forms of financial aid, including
scholarships and loans, to low-SES students. Scholarships and loans offer students different
incentives to pursue their studies (Cho etal., 2015). Scholarships do not require repayment,
reducing the cost of obtaining a degree from a student’s perspective. Loans may provide
fewer incentives for degree completion, especially when students face liquidity or borrow-
ing constraints (Johnson, 2013).
Governments offer different types of scholarships based on eligibility criteria. Students
receive need-based scholarships based on their financial needs, such as having a low family
income. Conversely, students receive merit-based scholarships based on their high school
GPA or university entrance exams. Scholarships that have both need-based and merit-based
components are called need- and merit-based scholarships (Nguyen etal., 2019). There is a
rapidly expanding academic literature focused on understanding the effects of scholarships
on enrollment, student persistence, and degree completion (Nguyen etal., 2019). Most
related research has shown that scholarships with a need-based component have greater
effects on persistence, and a combination of need and merit has proven to have even greater
positive effects (Nguyen etal., 2019).
There is an expanding body of literature examining how students behave after the loss
(Carruthers & Özek, 2016; LaSota etal., 2021) or completion (Mabel, 2020) of their schol-
arships. Several researchers have found that scholarship loss is associated with reduced
curricular progress and a lower likelihood of graduation (Carruthers & Özek, 2016; Mabel,
2020). Carruthers and Özek (2016) found, specifically for Georgia’s HOPE scholarship,
that a possible explanation for this lies in students exhibiting reduced engagement with col-
lege and increased engagement with work after experiencing scholarship loss. Research has
shown that implementing a performance standard of renewal on need-based scholarships
might improve academic progress (Mabel, 2020). However, if the standard is overly strict,
it may have a negative impact on enrollment, progression, and graduation, particularly for
the most disadvantaged students (Scott-Clayton & Schudde, 2020). Mabel (2020) found
that implementing time limits on need-based scholarships could produce both positive and
negative effects. On the positive side, these limits encourage students to complete their
studies sooner. On the negative side, they also contribute to an increase in dropout rates.
The availability of well-designed loans could minimize a negative impact on students
who lose their scholarships. For instance, loans that incentivize repayment could help pre-
vent excessive debt among academically underperforming students or in higher education
programs with a low economic return (Lochner & Monge-Naranjo, 2016). However, it is
also important to consider the characteristics of students that influence their willingness to
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

444
Higher Education (2025) 89:441–464
1 3
take on loans. Among these are debt aversion, the educational level of their parents, and the
income level of their family (De Gayardon etal., 2019).
Recent research has shown that interventions that combine outreach and financial aid
can have positive effects on graduation rates (Geven & Herbaut, 2019). These activities
include counseling and guidance, tutoring, transportation, and school supplies.
Financial aid intheChilean higher education system
In Chile, both public universities and private institutions with a longer history have been
receiving government funding since their inception. These universities are members of
the Council of Chancellors of Chilean Universities (Consejo de Rectores de Universi-
dades Chilenas, hereinafter CRUCH) (Santelices etal., 2016). Only in recent years (after
the period covered by this study) has the Chilean government started opening up funding
opportunities to include other private universities.
The Bicentennial Scholarship (BS) is a need- and merit-based scholarship provided
by the Chilean government (Santelices etal., 2016). During the period of our research,
this scholarship was restricted to students from universities affiliated with CRUCH who
exhibited high-level academic performance in national admission exams (70th percentile
or higher) and came from the two lowest income quintiles (Schmidt etal., 2019). Both uni-
versities and the government actively encouraged all eligible students to apply for the BS.
The BS and government loans cover tuition fees only up to a reference amount determined
by the Ministry of Education (Santelices etal., 2016) for each degree program. Typically,
there is a difference ranging from 10 to 20% between reference tuition and full tuition
(Dooner & Mena, 2006), which amounts to approximately US$500–1000 per year. Low-
SES students who have the BS usually cover the remaining difference with a loan, although
some of them pay for it by themselves. Students who lose the BS and use a government
loan must cover the difference between the reference tuition and full tuition by themselves.
In addition to tuition, students with the BS receive two additional scholarships. The Schol-
arship for Higher Education Living Expenses (BMES) provides approximately US$30 per
month for discretionary use. The Food Scholarship for Higher Education (BAES) offers
approximately US$50 per month exclusively designated for meal-related expenses. When
students lose the BS, they also lose the BMES but might retain the BAES if they use a
government loan. Although the amount of the BMES is only US$30 per month, students
may face financial challenges in covering transportation and other living expenses. To meet
the maintenance requirements and retain the BS, a student must pass 70% of their enrolled
courses each year (60% in the first year). Students know these requirements when apply-
ing for the scholarship, but universities also reinforce this message through their student
services offices. There are other public scholarships that can be used by students who are
enrolled at the university where this study was conducted, but all of them only partially
cover tuition (Cáceres-Delpiano etal., 2018), and students use them in combination with
government loans as the main funding mechanism.
CRUCH university students mainly have access to two loan types: The Solidary Fund of
University Loans (Fondo Solidario de Crédito Universitario, hereinafter FSCU) and Gov-
ernment Guaranteed Loans (Crédito con Aval del Estado, hereinafter CAE) (Cáceres-Del-
piano etal., 2018). The FSCU was created in 1994 and is managed by the government; its
loan interest rate is 2%, and loan payments depend on income. However, this type of loan
is available only to students at CRUCH universities. The CAE was created in 2005 and is
managed by private financial entities; its loan interest rate was 6% during the time period
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

445
Higher Education (2025) 89:441–464
1 3
in which this study was conducted, and it is available for the entire Chilean higher educa-
tion system, not only for CRUCH universities (Larraín & Zurita, 2008). Like the BS, these
loans cover tuition fees only up to the reference amount, and students are responsible for
financing the remaining difference themselves.
According to the government, these tuition aids have two main effects on students: they
increase the enrollment rate and improve the retention rate in higher education (Meneses &
Blanco, 2010).
Method
Data
We conducted this study at a medium-sized traditional private university in Chile that is
affiliated with CRUCH. The study analyzed a sample of 2484 students, representing 40%
of the population with lower socioeconomic status (SES), who received the BS when they
were admitted to their academic programs. These students belonged to cohorts 2004 to
2009, and the observation period ranged from the beginning of 2004 to the end of 2018.
A total of 46.6% of the students were women, and the mean age at first enrollment was
19.4years (SD = 1.7). The students were enrolled in a broad range of International Stand-
ard Classification of Education (ISCED) fields of education (Schneider Silke, 2013): 32.5%
in health and welfare; 24.8% in engineering; 17.8% in social science, business, and law;
10% in agriculture; 7.3% in science; 6.6% in education; and 1% in humanities and arts.
We analyzed anonymized student data provided by the institution, including enrollment
records, government loans and scholarship assignments, records of expelled students, and
grades obtained in each course. This information also included the curriculum structure
and nominal duration of each program, with undergraduate programs ranging from 8 to 14
semesters (median = 10).
Methodological approach
Learning analytics (LA) is a growing research area that focuses on measuring, collecting,
analyzing, and reporting data about students and their learning processes (Viberg etal.,
2018). Curricular analytics (CA) has emerged as a subfield of LA that aims to use evidence
to drive curriculum decision-making and program improvement (Pinnell etal., 2017). The
ability to analyze data at the population level (Figlio etal., 2016) is a characteristic of CA
that enhances the internal validity of the findings.
In this study, we are interested in understanding the educational funding trajectories
of students who begin their programs with the BS using a curricular analytics approach
through process mining (PM). PM aims to extract knowledge from event logs obtained
from information systems to discover process models, verify conformance, and suggest
improvements, acting as a bridge between data science and process science (van der Aalst,
2016). Through PM, we can see educational funding trajectories as a set of events that
occur while students remain in a given program. Statistical methods complement PM anal-
ysis, strengthening our findings (Ramaswami, 2019). We expect that this novel approach
will contribute to gaining a comprehensive understanding of the outcomes associated with
the loss of need- and merit-based scholarship in terms of funding, curricular progress, and
graduation rate.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

446
Higher Education (2025) 89:441–464
1 3
In this study, we introduce the SCHOLARSHIP-LOAN-SELF-FUNDED model with
the aim of systematizing the analysis of educational funding trajectories using PM tech-
niques. This model represents students’ educational funding trajectories as a sequence of
records that show the main means of funding that each student used to finance each semes-
ter. Table1 shows the three means of funding that were included.
Analytical strategy
We conducted this research following an adapted version of the PM2 methodology (Mal-
donado-Mahauad etal., 2018). This methodology defines the following four stages: data
extraction, event log generation, discovery, and analysis. The process models that were
obtained provide graphic representations of the educational funding trajectories of the stu-
dents; nodes represent events (milestones) relevant to their educational funding trajecto-
ries, while arrows represent transitions between events. For this research, we labeled nodes
based on the number of students, as well as the percentage of progress or the duration of
funding for students. To label arrows, we use the number and percentage of students who
had a transition between each pair of events.
Data extraction
First, we filtered anonymized records provided by the institution, considering only students
who started their programs supported by the BS. Next, we categorized the final state events
based on the data extracted from the academic records. Table2 describes those events. We
then examined a range of student data, including enrollment records, government loan and
scholarship assignment records, records of expelled students, and grades obtained in each
course.
Event log generation
In this stage, to facilitate subsequent discovery and analysis through PM, the data should
be modeled as an event log (Van der Aalst, 2019). Formally, an event log is defined as a
set of executions of the process (namely, cases), where each one is an ordered sequence of
actions that occurred during that execution (namely, events) (Van der Aalst, 2019). There-
fore, to define an event log, (1) how to identify a case and (2) how to specify a sequence of
events must be defined.
As mentioned in the methodological approach subsection, the SCHOLARSHIP-
LOAN-SELF-FUNDED model proposes analyzing educational funding trajectories
Table 1 Main means of funding used by students
Means of funding Criteria
SCHOLARSHIP The student has primarily financed the tuition for this semester with the Bicentennial
Scholarship
LOAN The student has primarily financed the tuition for this semester mainly with a govern-
ment loan, specifically the FSCU or CAE
SELF-FUNDED The student has not financed the tuition for this semester with either the BS or a gov-
ernment loan
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

447
Higher Education (2025) 89:441–464
1 3
through PM. The event log consists of educational funding trajectories, where each case
represents a student enrolled in a specific undergraduate program, and each event repre-
sents a record in the educational funding trajectory of that student. Table3 describes the
attributes included for each event.
To construct the educational funding trajectories, we followed these steps. First, we
recorded the main means of funding tuition for each semester during which a student
had academic activity, as defined in Table1. Next, we added an initial state for each
student, representing enrollment (ENR) in an undergraduate program. Then, we built
funding events, grouping the semesters in which each student had maintained the same
means of funding tuition. Additionally, for the case of students who lost their BS, we
added a SCHOLARSHIP-LOST event after the end of the semester when the BS was
lost. For students who had the BS and met the maintenance requisites but who stayed
longer than the nominal duration of the curriculum, we added a SCHOLARSHIP-END
Table 2 Criteria used to define the final state of a student
Final state event Criteria
EARLY-DROPOUT-EXPELLED The student stayed at most 2years and then were expelled from the
program due to poor academic performance
EARLY-DROPOUT-VOLUNTARY The student stayed at most 2years and then voluntarily dropped out
of the program
LATE-DROPOUT-EXPELLED The student was expelled from the program due to poor performance
and had more than 2years of academic records
LATE-DROPOUT-VOLUNTARY The student neither enrolled in 2019 nor graduated before that date
but had more than 2years of academic records
LATE-GRAD The student graduated in a period longer than the nominal duration
of the undergraduate program
ON-TIME-GRAD The student graduated in a period equal to or lower than the nominal
duration of the undergraduate program
Table 3 Description of the attributes of all events contained in the event log
Event log attributes Description
Student-ID This serves as a label that denotes a sequence of events (van der Aalst, 2016).
In this context, it corresponds to the educational funding trajectory of a given
student. It is composed of the student identifier and the identifier of the under-
graduate program to which the student belongs
Event We defined three types of events: enrollment, funding, and final state. For each
educational funding trajectory, the enrollment event (ENR) corresponds to the
entry into the academic program, and the last state event corresponds to one of
the six final states, as defined in Table2
Percentage-of-progress This quantifies a student’s advancement in the curriculum of their program,
calculated as the approved credits relative to the total program credits
Gender This represents a student’s gender
ISCED This represents the ISCED field to which an undergraduate program belongs
Start-date These timestamps correspond, in this context, to the start and end dates of each
semester, respectively
End-date
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

448
Higher Education (2025) 89:441–464
1 3
event after the end of the semester when the BS ended. Finally, we added a final state at
the end of each trajectory. Table2 describes the criteria used to define each final state.
Table 4 shows, as an example, the educational funding trajectory for one student,
namely, 11. In this example, student 11 stayed in his or her academic program for three
semesters, using the Bicentennial Scholarship for the first two semesters, making 8% pro-
gress in the study plan. This student lost the Bicentennial Scholarship at the end of the
second semester and continued with a LOAN for another semester, making 4% progress in
the study plan and then dropping out voluntarily.
In this research, we employed a Directly-Follows Graph (DFG) to represent the edu-
cational funding trajectories. A DFG consists of nodes representing events in the funding
trajectory and directed edges (transitions) that represent direct relationships among these
events (Van der Aalst, 2019). DFGs are among the most popular and widespread process
modeling notations (Van der Aalst, 2019). Figure1 shows the DFG for student 11 accord-
ing to the information provided in Table4.
Model discovery
The event log was loaded in R using bupaR, an integrated collection of R packages that
creates a framework for the reproducible analysis of processes in R, which includes both
graphical and analytical tools (Janssenswillen etal., 2019). We created models that group
educational funding trajectories by final state event.
Data analysis
To address the research question, we developed models using the event log, considering
different perspectives. These models enabled us to perform three distinct analyses based
on the final state achieved. First, we explored the trajectories involving the maintenance or
loss of the BS. For students who were unable to maintain the BS, we analyzed the tuition
funding methods they opted for upon completion or loss of the BS. Second, we examined
the curricular progress before and after losing the BS. Third, we explored the curricular
progress before and after ending the BS. The variables considered were the number and
percentage of students who reached different states and were involved in different transi-
tions, the number of semesters that each student stayed in a particular state, and the per-
centage of progress that the students had when using each tuition funding method. We
incorporated these variables into the models using nodes or transitions based on the spe-
cific focus of the analysis.
Table 4 Example of educational funding trajectory
Student ID Event % of progress Start date End date
11 ENR - 01 January 2015 01 January 2015
11 SCHOLARSHIP 8% 01 January 2015 31 December 2015
11 SCHOLARSHIP-LOST - 31 December 2015 31 December 2015
11 LOAN 4% 01 January 2016 30 June 2016
11 EARLY-DROPOUT-VOLUNTARY - 30 June 2016 30 June 2016
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

449
Higher Education (2025) 89:441–464
1 3
Fig. 1 Example of an educational
funding trajectory according to
the SCHOLARSHIP-LOAN-
SELF-FUNDED model. The
DFG shows the funding trajec-
tory for the student presented in
Table4. A darker node represents
greater curricular progress per
semester
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

450
Higher Education (2025) 89:441–464
1 3
Table 5 Filters and properties applied to the event logs to perform each analysis to answer the research question
Analysis Node type Transition type General filters Figure or table Additional filters Finding
Maintenance or loss of the
BS
Number of students - Final states: EARLY-
DROPOUT-EXPELLED,
EARLY-DROPOUT-
VOLUNTARY, LATE-
DROPOUT-EXPELLED,
LATE-DROPOUT-VOL-
UNTARY, LATE-GRAD,
ON-TIME-GRAD
Figure2
Table6
Include transition SCHOL-
ARSHIP > SCHOLAR-
SHIP-LOST
F1
Include transition SCHOL-
ARSHIP > SCHOLAR-
SHIP-END
Do not include transi-
tions SCHOLAR-
SHIP > SCHOL-
ARSHIP-LOST
or SCHOLAR-
SHIP > SCHOLARSHIP-
END
Curricular progress of those
who lost the BS
Average progress per
semester
Average number of semes-
ters
Percentage of students
Number of students
Final states: EARLY-DROP-
OUT, LATE-DROPOUT,
LATE-GRAD, ON-TIME-
GRAD
Includes transitions SCHOL-
ARSHIP > SCHOLAR-
SHIP-LOST
Do not include transi-
tions SCHOLARSHIP-
LOST > {EARLY-DROP-
OUT, LATE-DROPOUT,
LATE-GRAD, ON-TIME-
GRAD}
Figure3F2
Figure4Includes transitions SCHOL-
ARSHIP-LOST > LOAN
Figure5Includes transitions SCHOL-
ARSHIP-LOST > SELF-
FUNDED
Figure6
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

451
Higher Education (2025) 89:441–464
1 3
Table 5 (continued)
Analysis Node type Transition type General filters Figure or table Additional filters Finding
Curricular progress of those
for whom the BS ended
Average progress per
semester
Average number of semes-
ters
Percentage of students
Number of students
Final states: EARLY-DROP-
OUT, LATE-DROPOUT,
LATE-GRAD, ON-TIME-
GRAD
Include transitions SCHOL-
ARSHIP > SCHOLAR-
SHIP-END
Do not include transi-
tions SCHOLARSHIP-
END > {EARLY-DROP-
OUT, LATE-DROPOUT,
LATE-GRAD, ON-TIME-
GRAD}
Figure7F3
Figure8
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

452
Higher Education (2025) 89:441–464
1 3
Table5 describes the different analyses performed to address the research question,
including the identification of the educational funding trajectory models created: pri-
mary and secondary node types, primary and secondary transition types, and filters
applied. The filters that were applied to the model helped us analyze the behavior of
students who met specific criteria. We used general filters for an overall analysis of each
finding and additional filters to select specific transitions. The combination of these fil-
ters facilitated the construction of figures and tables that supported the findings.
For the first analysis, regarding the maintenance or loss of the BS, general filters
correspond to the final state events, as defined in Table2, whereas additional filters cor-
respond to the conditions associated with the maintenance, loss or ending of the BS. For
the second analysis, which focused on students who lost the BS, general filters enabled
us to identify students who lost the BS but remained in their programs for at least one
semester afterward. Additional filters allowed us to identify those who chose LOAN or
SELF-FUNDED after losing the BS. For the third analysis, which focused on students
for whom the BS had ended, we employed general filters to identify students whose BS
had ended but who remained in their programs for at least one semester afterward.
We utilized four distinct tools to analyze educational funding trajectories: We ana-
lyzed the maintenance or loss of the BS with a Sankey diagram (Soundararajan etal.,
2014). In this kind of diagram, flows represent the relationship between different nodes
in a network; the thickness of the flow is proportional to the magnitude of the rela-
tionship. In our case, nodes represent events in the educational funding trajectory of
students, and flows represent the transitions between those events. We analyzed the cur-
ricular progress of the students who changed their funding method using DFGs (Van der
Aalst, 2019) and classification and regression trees (CART). To evaluate the statistical
significance of the results, we applied the Fisher test.
Results
In this section, we present the three main results obtained in the discovery and analysis
stages of the PM2 methodology to answer the research question.
(F1) The maintenance orloss oftheBS relates tothefinal state reached
Figure2, depicted through a Sankey diagram, illustrates that a high proportion of the
flow passing through SCHOLARSHIP-LOST resulted in dropout. Figure2 also shows
that the proportions of flows from SCHOLARSHIP-LOST to each state that repre-
sents dropout are remarkably similar (all of the values are presented in Table 6). In
contrast, most of the flow originating from SCHOLARSHIP-END leads to LATE-
GRAD. This indicates that a significant number of students who fulfilled the mainte-
nance requirements but stayed in the curriculum longer than the nominal duration ended
up graduating late (80.5%, as shown in Table6). Finally, Fig.2 illustrates the main
flow originating from Maintain SCHOLARSHIP leading to either on-time graduation
(ON-TIME-GRAD) or voluntary early dropout (EARLY-DROPOUT-VOLUNTARY).
Table6 shows the specific percentages for these states: 22.5% for EARLY-DROPOUT-
VOLUNTARY and 68.8% for ON-TIME-GRAD.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

453
Higher Education (2025) 89:441–464
1 3
(F2) Scholarship loss relates tochanges incurricular progress andgraduation rates
Figure3 shows that for students who did not drop out immediately after losing their
BS, the average curricular progress per semester was slower than before. When stu-
dents had the BS, the average curricular progress per semester was 6.9%. Then, 90.6%
of the students transitioned from SCHOLARSHIP-LOST to LOAN, experiencing an
average curricular progress of 4.9% per semester, while for those who transitioned to
SELF-FUNDED, this figure was only 2.9%. The differences were statistically significant
(p < 0.01).
The graduation rate was lower for students who chose to finance tuition themselves
after losing the BS. Figure 4 shows that 49.6% of the students who had SCHOLAR-
SHIP-LOST > LOAN transitions graduated (3.1% finished ON-TIME-GRAD and 46.5%
LATE-GRAD), while Fig.5 shows that 31.5% of the students who had SCHOLAR-
SHIP-LOST > SELF-FUNDED graduated (8.6% finished ON-TIME-GRAD and 22.9%
LATE-GRAD). The differences were statistically significant (p < 0.01). Even though
Figs. 4 and 5 include transitions from LOAN to SELF-FUNDED or from SELF-
FUNDED to LOAN, these transitions occur at a low frequency.
Figure 6 illustrates a CART that provides insights into the factors influencing the
final state of students who lost their BS but remained in their programs. Three key vari-
ables were considered in this CART model: cumulative progress at the time of BS loss,
ISCED field, and gender. At the initial level, cumulative progress less than 16% at the
time of scholarship loss indicates a higher likelihood of dropping out (42.2%), while
higher cumulative progress suggests a greater chance of late graduation (53.4%). Gen-
der becomes a distinguishing factor only at the third level, among those with accumu-
lated progress between 7 and 16%. Female students exhibited an early dropout rate of
40.9%, whereas male students experienced late dropout at a rate of 56.1%. For students
with accumulated progress greater than or equal to 26%, the ISCED field became influ-
ential. Among those studying agriculture, arts and humanities, education, and sciences,
51.9% experienced late dropout. In contrast, students from other ISCED fields exhibited
a late graduation rate of 63.2%.
Fig. 2 A Sankey diagram that represents the educational funding trajectories according to the SCHOLAR-
SHIP-LOAN-SELF-FUNDED model. It includes only the events associated with the scholarship, as well as
the final states reached by the students
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

454
Higher Education (2025) 89:441–464
1 3
(F3) The students forwhom theBS had ended subsequently experienced slower
curricular progress
Figure 7 shows that the average curricular progress per semester after the SCHOLAR-
SHIP-END event was slower than before. While the students maintained the BS, they had
on average 9% progress per semester. After SCHOLARSHIP-END, students who chose to
use a LOAN to continue had, on average, 3.9% progress per semester. The students who
opted to be SELF-FUNDED had, on average, only 2.1% progress per semester. The differ-
ences were statistically significant (p < 0.01).
Figure8 illustrates a CART analysis revealing factors influencing the funding method cho-
sen after the BS ended. Three key variables were identified in the model: cumulative progress
at the time of BS ending, gender and ISCED field. At the first level, cumulative progress less
than 98% at the time of BS ending indicates a higher likelihood of choosing a LOAN. Below
this cumulative progress, the ISCED field influences this decision. Students in the arts and
Fig. 3 Only students who had
a SCHOLARSHIP > SCHOL-
ARSHIP-LOST transition and
direct transitions from SCHOL-
ARSHIP-LOST to LOAN or
SELF-FUNDED were included.
A darker node represents greater
curricular progress per semester
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

455
Higher Education (2025) 89:441–464
1 3
humanities, education, engineering, industry and construction, and health and social services
chose LOAN in 88.1% of the cases, while students from the sciences, agriculture, social sci-
ences, commercial education, and law chose LOAN in 47.1% of the cases (SELF-FUNDED
in 52.9% of the cases). The model revealed no significant relationship between gender and
funding method.
Fig. 4 Only students who had a
SCHOLARSHIP-LOST > LOAN
transition were included. A
darker node represents greater
curricular progress per semester.
The thickness of the arrows rep-
resents the percentage of students
who had transitions between two
nodes
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

456
Higher Education (2025) 89:441–464
1 3
Discussion andConclusion
By analyzing the educational funding trajectories for students who began with the BS, this
study advances the academic understanding of how the different means of financing tuition
are related to dropout, academic progress, and graduation.
The main contribution of this work is to examine the relationship between the loss
of a need- and merit-based scholarship and students’ educational trajectories in terms
of funding, curricular progress, and graduation rates, addressing this through a curricu-
lar analytics approach and discussing the results obtained through recent research. This
Fig. 5 Only students who had a
SCHOLARSHIP-LOST > SELF-
FUNDED transition were
included. A darker node repre-
sents greater curricular progress
per semester. The thickness
of the arrows represents the
percentage of students who had
transitions between two nodes
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

457
Higher Education (2025) 89:441–464
1 3
approach can aid in understanding the educational funding trajectories of students from
a longitudinal perspective, helping to elucidate their decision processes.
Next, the findings of this research are discussed, considering the previous literature.
First, the maintenance or loss of the BS relates to the final state reached (F1). Accord-
ing to the rational choice model of educational decisions (Breen & Goldthorpe, 1997),
the cost of education and the perceived returns relate to commitment. For students who
lost the BS, government loans in Chile covered tuition only up to the reference amount,
leaving the students or their families responsible for the remaining difference. Conse-
quently, most students who lost their scholarships dropped out afterward. In addition to
the difference between full tuition and the reference amount, a fear of debt and a percep-
tion of low utility in taking on debt, particularly among low-SES students, may influ-
ence dropout decisions (Long, 2021). These cultural factors are often transmitted along
family lines (Almenberg etal., 2021). On the one hand, the majority of students who
maintained the scholarship successfully graduated. However, nearly half of the students
reached the time limit of the BS before graduation. Imposing lifetime aid limits, as sug-
gested by Mabel (2020), can enhance commitment to on-time graduation. For those
whose scholarship ends before graduation, government loans are an alternative. Based
on this, we can hypothesize that students who made significant progress while on the BS
are likely to exhibit greater commitment to completing their studies, driven by the desire
to minimize future study costs (Zainol etal., 2018). The significance of this finding is
that it links the maintenance or loss of a Chilean need- and merit-based scholarship to
dropout and graduation rates, despite these outcomes occurring several semesters after
Fig. 6 Classification and regression tree (CART) that includes students who had a SCHOLARSHIP-
LOST > LOAN or SCHOLARSHIP-LOST > SELF-FUNDED transition. The outcome is the final event in
the educational funding trajectory. The variables are cumulative progress at the time of BS loss, student’s
gender. and ISCED field
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

458
Higher Education (2025) 89:441–464
1 3
Fig. 7 Only students who had
a SCHOLARSHIP > SCHOL-
ARSHIP-END transition and
direct transitions from SCHOL-
ARSHIP-END to LOAN or
SELF-FUNDED were included.
A darker node represents higher
curricular progress per semester
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

459
Higher Education (2025) 89:441–464
1 3
changes in funding. Policy-makers can benefit from understanding the long-term effects
of need- and merit-based scholarships on dropout, progress, and degree completion.
Second, curricular progress per semester was slower after students lost the BS (F2)
or after their BS ended (F3). Low-SES students may have financial needs beyond tuition
that are not met with scholarships (Dente & Piraino, 2011; Zembrodt, 2019). Low-SES
students may need to finance transportation, school supplies, or living expenses. In par-
ticular, the literature states that part-time work could reduce progress in their studies
(Thies, 2022) and the likelihood of degree completion (Almenberg etal., 2021) because
students may underestimate the academic costs of working while studying (Long, 2021).
Therefore, students may require assistance in effectively balancing their part-time work
commitments with their academic load to fulfill the maintenance requirements of the
BS.
Although more than 90% of students whose BS ended or was lost opted for a loan to
finance their studies, those whose BS ended exhibited a greater decline in semester pro-
gress than those who lost their BS. A possible explanation for this is that students whose
BS ended were closer to graduation, leading a higher proportion to seek part-time or full-
time employment (Carruthers & Özek, 2016). While more than 90% of students whose
BS ended eventually graduated, the restrictive duration of the BS, limited to the nominal
duration of the curriculum, proved detrimental, leading to delays in graduation. Figure7
illustrates this behavior, showing that students whose BS ended need an average of 2.8
semesters to complete the final 11% of the curriculum. This observation aligns with the
findings of Mabel (2020).
In addition to borrowing, there are two main differences between the BS and the Chil-
ean government loans. First, the students must finance the remaining difference between
full tuition and reference tuition by themselves. Second, they are not eligible for the BMES
scholarship. When students have to engage in part-time work or lack sufficient funds for
living expenses, they can compromise their dedication to their studies. This could have
implications not only for their curricular progress but also for their degree of attainment.
Fig. 8 Classification and regression tree (CART) that includes students who had a SCHOLARSHIP-
END > LOAN or SCHOLARSHIP-END > SELF-FUNDED transition. The outcome is the funding method
chosen after ending BS. Variables are cumulative progress at the time the BS ended and the ISCED field
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

460
Higher Education (2025) 89:441–464
1 3
Table 6 Number and percentage of students who started with the BS, broken down by final state and last event associated with the BS
Final state SCHOLARSHIP-LOST SCHOLARSHIP-END Maintain SCHOLARSHIP Total
# Students % Students # Students % Students # Students % Students # Students % Students
EARLY-DROPOUT-EXPELLED 184 16.5% 0 0.00% 11 1.7% 195 7.9%
EARLY-DROPOUT-VOLUNTARY 179 16.0% 0 0.00% 146 22.5% 325 13.1%
LATE-DROPOUT-EXPELLED 187 16.7% 17 2.4% 1 0.2% 205 8.3%
LATE-DROPOUT-VOLUNTARY 190 17.0% 38 5.3% 33 5.1% 261 10.5%
LATE-GRAD 333 29.8% 578 80.5% 11 1.7% 922 37.1%
ON-TIME-GRAD 45 4.0% 85 11.8% 446 68.8% 576 23.2%
Tot a l 1118 718 648 2484
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

461
Higher Education (2025) 89:441–464
1 3
Among students who lost the BS but remained in their programs, those with less cumu-
lative progress were more likely to drop out. Interestingly, within this group, female stu-
dents tended to drop out earlier than male students, possibly due to their greater aversion to
debt (Almenberg etal., 2021). Among students who lost the BS with cumulative progress
above 26% but continued in their programs, dropout rates varied across ISCED areas. Stu-
dents in agriculture, arts and humanities, education, and sciences had higher dropout rates.
Two hypotheses for this are differences in financial literacy and subjective evaluation of
loan costs and benefits (Cho etal., 2015).
The contribution of these findings is that they quantify, for this study, the average reduc-
tion in curricular progress after the loss or the end of this need- and merit-based scholar-
ship according to the main means of funding chosen by the students.
In conclusion, the findings presented in this study have both theoretical and practical
implications for dropout and late graduation associated with funding. HEIs should design
curricula considering the maintenance requisites of financial aid. In this case, students
must pass at least 70% of the enrolled courses each year to maintain the BS. Therefore,
these requisites should be considered in study plan design so that multiple high-failure rate
courses are not taken within the same semester. Universities should implement mentoring
and support programs to prevent the loss of the BS. According to Nguyen etal. (2019),
providing multidimensional support, such as proactive advising, academic tutoring, and
supplementary services, can help students maintain academic performance and meet main-
tenance requirements. It can also guide students in addressing their unmet financial needs
(Zembrodt, 2019). Carruthers and Özek (2016) suggest the idea of implementing “nudges”
(small economic incentives) as a cost-effective approach to improving students’ adherence
to requirements. Finally, promoting financial literacy among low-SES students can foster
a more informed approach to managing debt (Almenberg etal., 2021; Long, 2021). Gov-
ernments should balance the eligibility criteria and maintenance requisites to encourage
progress and on-time graduation without increasing early loss or dropout (Mabel, 2020).
PM supports a comprehensive understanding of students’ dynamic behavior over time,
including the identification of decision points and the events preceding and following
them. Moving forward, future research should aim to apply the SCHOLARSHIP-LOAN-
SELF-FUNDED model in diverse contexts, encompassing variations in curricula, stu-
dent profiles, and financial aid designs. Researchers can also utilize this technique as a
complement to causal studies and enrich their understanding by integrating qualitative
research. We believe that researchers can use this technique to identify decision points
and understand the events preceding and following those decision points, the frequencies
of transitions between states, and the variables that they could then use in a causal study.
Qualitative methods can also be integrated with PM (Koorn etal., 2021) and used to elu-
cidate how psychological characteristics, contextual information, and other factors, such as
debt aversion, part-time work, and liquidity constraints, influence academic progress and
persistence.
We identified three main limitations of this study that readers should consider. First,
our results are related to a specific study that included curricula and educational policies in
force at a Chilean university. This context-specific nature of the study may limit the exter-
nal validity of the findings. Nevertheless, we believe that researchers can use the SCHOL-
ARSHIP-LOAN-SELF-FUNDED model in diverse contexts. Furthermore, we believe that
the main findings of this research are likely to hold true for other need- and merit-based
scholarships, as supported by the current literature. In summary, this article contributes
to the understanding of how governments should participate in financing university edu-
cation, a topic of significant relevance not only in Chile but also in countries such as the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

462
Higher Education (2025) 89:441–464
1 3
United States (Long, 2021), Canada (Lang, 2022), and the UK (Marginson, 2018). Second,
the information we had regarding the students and their educational trajectories was lim-
ited to the curriculum, the BS, and government loans. There could be additional variables
influencing educational trajectories apart from the funding, which is the focus of this study.
For example, we lacked contextual information about the students, including their physical
and mental health, parental support, and living conditions. Third, the conclusions obtained
through PM depend on the completeness of the information used (Bose etal., 2013). In this
particular case, there was no available information regarding transferred students, and it is
possible that the institution recorded transfer credits as approved within the new program.
However, less than 5% of the students were transfer students; therefore, we expect that the
effect on the progress analysis is negligible.
Acknowledgements Universidad Austral de Chile provided anonymized data.
Author contribution • Conceptualization: Juan Pablo Salazar-Fernandez and Marcos Sepúlveda.
• Funding acquisition: Juan Pablo Salazar-Fernandez and Marcos Sepúlveda.
• Data curation: Juan Pablo Salazar-Fernandez.
• Software: Juan Pablo Salazar-Fernández.
• Methodology: Juan Pablo Salazar-Fernandez, Jorge Munoz-Gama, and Marcos Sepúlveda.
• Writing—original draft: Juan Pablo Salazar-Fernandez.
• Writing—review and editing: Juan Pablo Salazar-Fernandez, Marcos Sepúlveda and Jorge
Munoz-Gama.
• Supervision: Marcos Sepúlveda and Jorge Munoz-Gama.
• Validation: Marcos Sepúlveda and Jorge Munoz-Gama.
Funding This study was funded by National Agency for Research and Development (ANID) – Scholarship
Program/Doctorado Nacional 2015—21150985 and supported by National Agency for Research and Devel-
opment (ANID)/FONDECYT—Chile Regular Project—1200206.
Data availability Due to institutional restrictions, the anonymized student data is not available. However, the
event log and the supporting R code for the findings of this study are accessible upon reasonable request to
the corresponding author, Juan Pablo Salazar-Fernandez.
Declarations
Conflict of interest • Juan Pablo Salazar-Fernandez has received research support from National Agency for
Research and Development (ANID) – Scholarship Program/Doctorado Nacional 2015 – 21150985.
• Juan Pablo Salazar-Fernandez receives a salary from Universidad Austral de Chile. He is an assistant
professor.
• Marcos Sepúlveda has received research support from National Agency for Research and Development
(ANID)/FONDECYT—Chile Regular Project—1200206.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Com-
mons licence, and indicate if changes were made. The images or other third party material in this article
are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly
from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
References
van der Aalst, W. (2016). Process Mining: The Missing Link. In W. van der Aalst (Ed.), Process mining: Data
science in action (pp. 25–52). Springer Berlin Heidelberg. https:// doi. org/ 10. 1007/ 978-3- 662- 49851-4_2
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

463
Higher Education (2025) 89:441–464
1 3
Almenberg, J., Lusardi, A., Säve-Söderbergh, J., & Vestman, R. (2021). Attitudes toward debt and debt
behavior. The Scandinavian Journal of Economics, 123(3), 780–809. https:// doi. org/ 10. 1111/ sjoe. 12419
Bose, R. P. J. C., Mans, R. S., & Aalst, W. M. P. v. d. (2013). Wanna improve process mining results?. 2013
IEEE Symposium on Computational Intelligence and Data Mining (CIDM). https:// doi. org/ 10. 1109/
CIDM. 2013. 65972 27
Breen, R., & Goldthorpe, J. H. (1997). Explaining educational differentials: Towards a formal rational
action theory. Rationality and Society, 9(3), 275–305. https:// doi. org/ 10. 1177/ 10434 63970 09003 002
Cáceres-Delpiano, J., Giolito, E., & Castillo, S. (2018). Early impacts of college aid. Economics of Educa-
tion Review, 63, 154–166. https:// doi. org/ 10. 1016/j. econe durev. 2018. 02. 003
Carruthers, C. K., & Özek, U. (2016). Losing HOPE: Financial aid and the line between college and work.
Economics of Education Review, 53, 1–15. https:// doi. org/ 10. 1016/j. econe durev. 2016. 03. 014
Cho, S. H., Xu, Y., & Kiss, D. E. (2015). Understanding student loan decisions: A literature review. Family
and Consumer Sciences Research Journal, 43(3), 229–243. https:// doi. org/ 10. 1111/ fcsr. 12099
De Gayardon, A., Callender, C., & Green, F. (2019). The determinants of student loan take-up in England.
Higher Education, 78, 965–983. https:// doi. org/ 10. 1007/ s10734- 019- 00381-9
Dente, B., & Piraino, N. (2011). Models for determining the efficiency of student loans policies. Journal of
Higher Education Policy and Management, 33(4), 375–386. https:// doi. org/ 10. 1080/ 13600 80X. 2011.
585737
Dooner, C., & Mena, P. (2006). Arancel de referencia v/s arancel real: diagnóstico e interrogantes iniciales.
Calidad en la Educación, (24), 287–318. https:// doi. org/ 10. 31619/ caledu. n24. 280
Dynarski, S., & Scott-Clayton, J. (2013). Financial aid policy: Lessons from research. National Bureau of
Economic Research. https:// doi. org/ 10. 3386/ w18710
Espinoza, O., González, L. E., Sandoval, L., etal. (2022). Reducing inequality in access to university in
Chile: The relative contribution of cultural capital and financial aid. Higher Education, 83, 1355–1370.
https:// doi. org/ 10. 1007/ s10734- 021- 00746-z
Figlio, D., Karbownik, K., & Salvanes, K. G. (2016). Education research and administrative data. In Hand-
book of the economics of education. Vol. 5, 75–138. Elsevier. https:// doi. org/ 10. 1016/ B978-0- 444-
63459-7. 00002-6
Geven, K., & Herbaut, E. (2019). What works to reduce inequality in higher education? International
Higher Education, 99, 10–11. https:// doi. org/ 10. 6017/ ihe. 2019. 99. 11649
Goldrick-Rab, S., Kelchen, R., Harris, D. N., & Benson, J. (2016). Reducing income inequality in educa-
tional attainment: Experimental evidence on the impact of financial aid on college completion. Ameri-
can Journal of Sociology, 121(6), 1762–1817. https:// doi. org/ 10. 1086/ 685442
Henry, G. T., Rubenstein, R., & Bugler, D. T. (2004). Is HOPE enough? Impacts of receiving and losing
merit-based financial aid. Educational Policy, 18(5), 686–709. https:// doi. org/ 10. 1177/ 08959 04804
26909
Janssenswillen, G., Depaire, B., Swennen, M., Jans, M., & Vanhoof, K. (2019). bupaR: Enabling repro-
ducible business process analysis. Knowledge-Based Systems, 163, 927–930. https:// doi. org/ 10. 1016/j.
knosys. 2018. 10. 018
Johnson, M. T. (2013). Borrowing constraints, college enrollment, and delayed entry. Journal of Labor Eco-
nomics, 31(4), 669–725. https:// doi. org/ 10. 1086/ 669964
Koorn, J. J., Beerepoot, I., Dani, V. S., Lu, X., van de Weerd, I., Leopold, H., & Reijers,H. A. (2021). Bring-
ing rigor to the qualitative evaluation of process mining findings: An analysis and a proposal. 2021
3rd international conference on process mining(ICPM). Eindhoven, Netherlands. 31 oct - 4 nov 2021.
120–127. https:// doi. org/ 10. 1109/ ICPM5 3251. 2021. 95768 77
Lang, W. (2022). Financing higher education in Canada: A study in fiscal federalism. Higher Education, 84,
177–194. https:// doi. org/ 10. 1007/ s10734- 021- 00761-0
Larraín, C., & Zurita, S. (2008). The new student loan system in Chile’s higher education. Higher Educa-
tion, 55(6), 683–702. https:// doi. org/ 10. 1007/ s10734- 007- 9083-3
LaSota, R. R., Polanin, J. R., Perna, L. W., Austin, M. J., Steingut, R. R., & Rodgers, M. A. (2021). The
effects of losing postsecondary student grant aid: Results from a systematic review. Educational
Researcher. https:// doi. org/ 10. 3102/ 00131 89X21 10568 68
Lim, H., Lee, J. M., & Kim, K. T. (2019). What Factors Are Important in Aversion to Education Debt? Fam-
ily and Consumer Sciences Research Journal, 48(1), 5–21. https:// doi. org/ 10. 1111/ fcsr. 12324
Lochner, L., & Monge-Naranjo, A. (2016). Student loans and repayment: Theory, evidence, and policy. In
Handbook of the Economics of Education (Vol. 5, pp. 397–478). Elsevier. https:// doi. org/ 10. 1016/
B978-0- 444- 63459-7. 00008-7
Long, M. G. (2021). The relationship between debt aversion and college enrollment by gender, race, and
ethnicity: A propensity scoring approach. Studies in Higher Education, 1–19. https:// doi. org/ 10. 1080/
03075 079. 2021. 19683 67
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

464
Higher Education (2025) 89:441–464
1 3
Mabel, Z. (2020). Aiding or dissuading? The effects of reducing lifetime eligibility limits for need-based
aid on bachelor’s degree attainment and time to completion. Research in Higher Education, 61(8),
966–1001. https:// doi. org/ 10. 1007/ s11162- 020- 09600-0
Maldonado-Mahauad, J., Pérez-Sanagustín, M., Kizilcec, R. F., Morales, N., & Munoz-Gama, J. (2018).
Mining theory-based patterns from Big data: Identifying self-regulated learning strategies in Massive
Open Online Courses. Computers in Human Behaviour, 80, 179–196. https:// doi. org/ 10. 1016/j. chb.
2017. 11. 011
Marginson, S. (2018). Global trends in higher education financing: The United Kingdom. International
Journal of Educational Development, 58, 26–36. https:// doi. org/ 10. 1016/j. ijedu dev. 2017. 03. 008
Meneses, F., & Blanco, C. (2010). Financial aid and higher education enrollment in Chile: A government
policy analysis. Munich Personal RePEc Archive Paper No. 23321. Retrieved January 05, 2023, from
https:// mpra. ub. uni- muenc hen. de/ id/ eprint/ 23321
Nguyen, T. D., Kramer, J. W., & Evans, B. J. (2019). The effects of grant aid on student persistence and
degree attainment: A systematic review and meta-analysis of the causal evidence. Review of Educa-
tional Research, 89(6), 831–874. https:// doi. org/ 10. 3102/ 00346 54319 877156
Pinnell, C., Paulmani, G., Kumar, V., & Kinshuk (2017). Curricular and learning analytics: A big data per-
spective. In: Kei Daniel B. (eds) Big Data and Learning Analytics in Higher Education. Springer,
Cham. 125–145. https:// doi. org/ 10. 1007/ 978-3- 319- 06520-5_9
Ramaswami, G. (2019). Using educational data mining techniques to increase the prediction accuracy of
student academic performance. Information and Learning Sciences, 120(7/8), 451–467. https:// doi.
org/ 10. 1108/ ILS- 03- 2019- 0017
Santelices, M. V., Catalán, X., Kruger, D., & Horn, C. (2016). Determinants of persistence and the role
of financial aid: Lessons from Chile. Higher Education, 71(3), 323–342. https:// doi. org/ 10. 1007/
s10734- 015- 9906-6
Schmidt, A., de Dios Ortúzar, J., & Paredes, R. D. (2019). Heterogeneity and college choice: Latent class
modelling for improved policy making. Journal of Choice Modelling, 33. https:// doi. org/ 10. 1016/j.
jocm. 2019. 100185
Schneider Silke, L. (2013). The International Standard Classification of Education 2011. In B. Gunn Elisa-
beth (Ed.), Class and Stratification Analysis. 30, 365–379. Emerald Group Publishing Limited. https://
doi. org/ 10. 1108/ S0195- 6310(2013) 00000 30017
Scott-Clayton, J., & Schudde, L. (2020). The consequences of performance standards in need-based aid
evidence from community colleges. Journal of Human Resources, 55(4), 1105–1136. https:// doi. org/
10. 3368/ jhr. 55.4. 0717- 8961R2
Soundararajan, K., Ho, H. K., & Su, B. (2014). Sankey diagram framework for energy and exergy flows.
Applied Energy, 136, 1035–1042. https:// doi. org/ 10. 1016/j. apene rgy. 2014. 08. 070
Thies, T.(2022). International students in higher education: The effect of student employment on academic
performance and study progress. Higher Education. https:// doi. org/ 10. 1007/ s10734- 022- 00950-5
van der Aalst, W. M. P. (2019). A practitioner’s guide to process mining: Limitations of the directly-follows
graph. Procedia Computer Science, 164, 321–328. https:// doi. org/ 10. 1016/j. procs. 2019. 12. 189
Viberg, O., Hatakka, M., Bälter, O., & Mavroufi, A. (2018). The current landscape of learning analytics in
higher education. Computers in Human Behavior, 89, 98–110. https:// doi. org/ 10. 1016/j. chb. 2018. 07.
027
Zainol, Z. B., Yahaya, R., & Osman, J. (2018). Application of relationship investment model in predicting
student engagement towards HEIs. Journal of Relationship Marketing, 17(1), 71–93. https:// doi. org/
10. 1080/ 15332 667. 2018. 14401 43
Zembrodt, I. (2019). Commitment: Predicting persistence for low-SES students. Journal of College Student
Retention: Research, Theory & Practice, 23(3), 580–606. https:// doi. org/ 10. 1177/ 15210 25119 858340
Geolocation information
-39.8148°S, -73.2441°W.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center
GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers
and authorised users (“Users”), for small-scale personal, non-commercial use provided that all
copyright, trade and service marks and other proprietary notices are maintained. By accessing,
sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of
use (“Terms”). For these purposes, Springer Nature considers academic use (by researchers and
students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and
conditions, a relevant site licence or a personal subscription. These Terms will prevail over any
conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription (to
the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of
the Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may
also use these personal data internally within ResearchGate and Springer Nature and as agreed share
it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not otherwise
disclose your personal data outside the ResearchGate or the Springer Nature group of companies
unless we have your permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial
use, it is important to note that Users may not:
use such content for the purpose of providing other users with access on a regular or large scale
basis or as a means to circumvent access control;
use such content where to do so would be considered a criminal or statutory offence in any
jurisdiction, or gives rise to civil liability, or is otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association
unless explicitly agreed to by Springer Nature in writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a
systematic database of Springer Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a
product or service that creates revenue, royalties, rent or income from our content or its inclusion as
part of a paid for service or for other commercial gain. Springer Nature journal content cannot be
used for inter-library loans and librarians may not upload Springer Nature journal content on a large
scale into their, or any other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not
obligated to publish any information or content on this website and may remove it or features or
functionality at our sole discretion, at any time with or without notice. Springer Nature may revoke
this licence to you at any time and remove access to any copies of the Springer Nature journal content
which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or
guarantees to Users, either express or implied with respect to the Springer nature journal content and
all parties disclaim and waive any implied warranties or warranties imposed by law, including
merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published
by Springer Nature that may be licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a
regular basis or in any other manner not expressly permitted by these Terms, please contact Springer
Nature at
onlineservice@springernature.com