1 Assessing Dropout Is Not One Size Fits All: Policy Highlights

1.1 The International Perspective

In the context of the Bologna Process policy framework, the social dimension was an issue highlighted from the beginning of the process. Formally adopted by ministers responsible for higher education in 2007, it provided the well-known definition that the “student body entering, participating and completing higher education at all levels, should reflect the diversity of our population” (Bologna Process 2007). The work has been continued by the Bologna Process Follow-Up Group (BFUG) through the Working Groups on social dimension, and in 2020, in Rome, the ministers adopted the “Principles and guidelines to strengthen the social dimension of higher education in the EHEA” (Bologna Process 2020). The guidelines refer to concrete measures for EHEA member states such as setting clear targets for widening access to higher education, participation and graduation, making studies more flexible, collecting dropout data, improving student counselling services and adequate funding to support equity.

Although the mentioned definition is perfectible, it emphasises the challenge to remove social and economic barriers, barriers that prevent access and/or completion of university studies. Previous studies (Eurydice 2010; 2012) disclose that only some states focus on these issues in their higher education policies. A broad participation of individuals belonging to vulnerable/underrepresented groups (based on low socioeconomic status, gender, disability or with a minority status, linked to their ethnic, linguistic, religious, cultural, or residence characteristics) represents one of the core elements in order to achieve this goal (Bologna Process 2007).

In the Commissions’ Communication “Supporting growth and jobs” (2011), reducing higher education dropout is mentioned as a key issue for several member states, especially in countries with demographic decline (such as Romania).

Dropout has been studied as a research topic since the 80s worldwide. At the international level, there are a multitude of terms used in this regard, such as “dropout”, “non-persistence”, “academic performance/success versus academic failure”, “withdrawal”, “retention versus attrition”, “disengagement”, “desertion” (Jones 2008).

In Romania, very little research has been done on this subject, including addressing it in national public policy documents. Given the lack of common practice, both at the international and national level, there is a variety of approaches to the phenomenon of university dropout.

From a policy perspective, participation in higher education has been periodically analysed at the international level by the OECD through completion rates and Eurydice structural indicators for monitoring education systems that are also linked with the European Commission participation benchmark and policies. The definition of the indicator focuses on the percentage of students who complete the higher education programme they have started (Table 1).

Table 1 International definitions of dropout

After researching international approaches on defining the dropout phenomenon, two reports emerged: the NESET report from 2013 and the CHEPS report made in 2015 for the European Commission. Both explore the diversity of the national data collection systems and map different definitions given across countries. The NESET report shows how countries mostly use completion rates, to a large extent due to the commitment to report data to the OECD’ (Quinn 2013).OECD defines completion rates as “the number of degrees awarded per 100 students enrolled/registered in a given year”, while NESET report notes differences in national data collection methodologies and timeline.

The largest study on dropout and completion in higher education at the European level made for the European Commission (CHEPS 2015) managed to explore the indicators used across 36 European countries and note the top three most used: completion rate, retention rate and time-to-degree. While completion rate looks at the proportion of graduates among a cohort, retention rate represents the proportion of a cohort of beginners that continue their studies measured per semester or year, sometimes seen as the complement of the dropout rate. Time-to-degree represents the average number of years from registration to graduation.

Nevertheless, there is a multitude of variables that can be taken into account when calculating the indicators mentioned above at a system level, thus making the international comparability even harder (Table 2).

Table 2 International definitions of dropout

Besides the variations on the population that is being analysed, it is also important to take into account the various definitions and regulations that describe different dynamics of student life, involving temporary interruption of studiestransfer between or within universities, pursuing two study programs at the same time or delayed graduation.

1.2 The Case of Romania

The same conceptual diversity can be found in Romania, where there is no nationally agreed definition of university dropouts. The Ministry of Education publishes an annual report on the state of higher education. It also includes an indicator that presents the “school situation of students” referring to dropouts calculated as the percentage difference between the number of students (all students, regardless of the cohort they belong to) from the beginning and those from the end of an academic year (including students with unfinished academic status). The values of the mentioned indicator vary between 8.5% in the academic year 2014–2015 and 9.6% in the academic year 2018–2019 (Ministry of Education 2019).

In many European countries, the completion rate is an indicator often used in higher education funding or quality assurance policies. In Romania, in addition to the basic university funding (based on student numbers), universities receive additional funds based on quality indicators. In the methodology for university public funding published in 2018, the indicator “graduation rate of bachelor programs” is proposed and defined as: “the ratio between the total number of students who have obtained a bachelor’s degree in the last four completed academic years and the total number of students enrolled in the first year of the bachelor’s degree, in the year the study program at the bachelor cycle started” (Ministry of Education 2018). In the following yearly methodologies, the proposal to introduce such an indicator was no longer found.

The assessment of the successful completion of studies is also missing from the external quality assurance process carried on by the Romanian Quality Assurance Agency (ARACIS). Although the structure and design of study programmes is a quality standard found in the quality assurance and accreditation methodology, there is no reference to the dropout, participation, or completion rates.

The monitoring of university dropouts is also largely missing from the main policy documents. The ongoing Educated Romania project developed by the Presidential Administration (2021) addresses equity in the education system in a cross-cutting approach and includes reducing the dropout at higher education level as a policy objective. Thus, a whole series of specific measures are proposed, such as better data collection for the development of equity policies, removal of financial barriers for access to higher education, especially for students from disadvantaged groups, rewarding inclusive universities, providing service packages in support for students at high risk of dropping out, introducing flexible access routes in higher education.

We analysed the dropout phenomenon at Romanian universities, using formal documents adopted in 39 public universities, documents that include a definition or monitoring procedures for dropout.Footnote 1 As a result, we can conclude that there is no strategic approach to this phenomenon in Romanian public universities. Apart from one case, no clear definition and monitoring procedures could be identified. In most of the analysed cases, the phenomenon of dropouts is briefly addressed within general regulations of students’ academic activities or specific projects implemented by the university.

One of the few definitions found in regulations explains: “school dropout means the complete absence of the student from teaching activities for at least two consecutive months, without the approval of the dean of the faculty for it” (George Emil Palade University of Medicine, Pharmacy, Science and Technology in Târgu Mureş, 2020)

Among the university documents identified and related to our research topic, we noted a series of good practice examples:

  1. 1.

    The adoption of a strategy to reduce the risk of university dropout (Example of Babeş-Bolyai University);

  2. 2.

    The existence of formal monitoring procedures and calculating dropout risk based on data from student registers (Example of Babeş-Bolyai University);

  3. 3.

    Procedures regarding the prevention of dropout (Examples: University of Bucharest, “Aurel Vlaicu” University);

  4. 4.

    Analyses, articles and studies on dropout at the institutional level (Examples: Western University, Ovidius University);

  5. 5.

    Implementation of projects from public funding (Institutional Development Fund, ROSE Project or European funds) whose objectives include reducing dropout (Example of the Politehnica University of Bucharest).

As these initiatives target the dropout phenomenon explicitly, the focus of these strategic documents is still on using the same tools to reduce the risk of dropout, many of them being the mirror of the policies developed at the national level: social scholarships, student counselling, covering the student accommodation and food costs, and support for studies (tutoring program, distance study programs, access to libraries).

All these institutional practices show that there is a need for comparable data on the human capital that is lost during studies. It can be a good reflection of the quality of education, or the lack of equity, or the integration of specific groups of people, or it can show a mismatch between expectations and reality. All these can be addressed as long as the phenomenon is monitored and analysed.

2 Methods

2.1 Data and Variables

The data have been extracted from the National Student Enrolment Registry (Registrul Matricol Unic al Universităţilor din România, RMUR). RMUR is a digital platform that ensures the integrated management of data on students of both public and private higher education institutions (HEIs) in Romania for all academic years and all study cycles. Personal data, student school data, student scholarships and accommodation services, respectively data on pre-university and previous university studies are recorded in RMU.

  1. 1.

    The selection of relevant data for the present analysis followed a 5-level approach: (a) public HEIs, (b) undergraduate (bachelor) study programmes, (c) 3-year study programmes, (d) the cohort of students enrolled in the analysed study programme, in the 2015–2016 academic year for the first time (n = 60,510 students), (e) the timeframe covered (2015–2016 through 2019–2020 academic years) creates an information line (one case) for each academic year, resulting 177,256 number of cases (students) in the database. International students were kept in the dataset used by current research.

  2. 2.

    We define for this paper the dropout rate as “the percentage of the student population who failed to graduate within two years of the theoretical completion date for the study program”. We consider as graduates the students having as status “own graduate with a diploma – code 15” or “completing diploma date – code 75” for the variable sitScolaraId_final_an.

  3. 3.

    We developed the dropout dependent variable (abnd_non_absolvent) as below:

    1. a.

      Recoding the variable sitScolaraId_final_an: codes 15 and 75 become sequential 0 for the 2017–2018 year (31,197 graduates), 2018–2019 (2,392 graduates) and 2019–2020 (638 graduates)Footnote 2: total graduates in all three academic years: 34,227.

    2. b.

      A new database (graduates.sav) with a list of graduates identified in previous stage (3.a) could be used to populate the entire database (stage 1.e.) with information relative to graduates. Thus, duplicates (185 cases) were eliminated, as students who completed two study programmes within the time horizon analysed. The personId identifier was used, and a sample volume without duplicates was used: 34,042.

    3. c.

      Using the stage 3.b stageinformation in the database, the abnd_non_absolvent variable has been entered into the database in stage s 1.e.

    4. d.

      The dropout rate was calculated comparing the abnd_non_absolvent variable to the students registered in the initial cohort (in the 2015–2016 academic year). The university dropout rate of the 3-year undergraduate programs (cohort of individuals registered in the 2015–2016 academic year for their first study programme and academic year ever) is 43.8% (statistical details for each field of study are available in Annex).

2.2 Model Specifications

Considering the dichotomous nature of the dependent variable (dropout), with its associated probability (\(p_i\)), a Logit model has been built using the following factors: students’ personal motivation, previous educational outcomes, and factors related to the integration in the students’ life.

$$\begin{aligned} \begin{aligned} {\mathrm {ln} \left( \frac{P_i}{1-P_i}\right) \ }= \,&\alpha _{0} + \alpha _{1}BAC\_ATTEMPTS_{i} + \alpha _{2}SAME\_TOWN_{i} + \\&+ \alpha _{3}BAC\_AVERAGE_{i} + \alpha _{4}FIRST\_YEAR_{i} + \\&+ \alpha _{5}TUITION\_PAYER_{i} + \varepsilon _{i} \end{aligned} \end{aligned}$$

As a proxy measure for motivation, two variables were kept. The first one is BAC_ATTEMPTS (number of participations at the Baccalaureate exam before passing the exam) since a higher number of attempts can indicate a more powerful personal desire to be enrolled in higher education. The second one is SAME_TOWN (a dummy variable to highlight if the student’s home is located in the same town as the university). This variable was introduced to test the hypothesis that students being located outside big cities (where usually the most important universities are located) tend to be better motivated to finish their educational program in time to benefit from specific support measures (e.g., student dormitories, scholarships) and, as an ultimate goal, to use their degree as a social elevator.

The second type of factors describing previous educational outcomes consists of the one variable BAC_AVERAGE (the average grade obtained at the Baccalaureate exam).

The third category of factors measuring the integration in the students’ life includes two variables: FIRST_YEAR (if the student is registered as a freshman in the first year of a study programme) and TUITION_PAYER (if the student was supporting the tuition fees within the first semester/year of study).

We estimate the Eq. (1) above as the general model. Furthermore, the following control socio-demographic characteristics are included in the model (M2—Eq. 2) AGE, GENDER and SOCIO-ECON_INDEX (an index developed by Pană, 2020 to measure the locality’s level of development from socio-economic point of view) for the locality where the student’s home is located:

$$\begin{aligned} \begin{aligned} {\mathrm {ln} \left( \frac{P_i}{1-P_i}\right) \ }=\beta _{0} +&\beta _{1}AGE_{i}+\beta _{2}GENDER_{i} + \beta _{3}SOCIO - \\&- ECON\_INDEX_{i} + \varepsilon _{i} \end{aligned} \end{aligned}$$

In a third stage, the control variables are inserted in the model (M3):

$$\begin{aligned} \begin{aligned} {\mathrm {ln} \left( \frac{P_i}{1-P_i}\right) \ }=&\,\gamma _{0} + \gamma _{1}BAC\_ATTEMPTS_{i} + \gamma _{2}SAME\_TOWN_{i} + \\&+ \gamma _{3}BAC\_AVERAGE_{i} + \gamma _{4}FIRST\_YEAR_{i} + \\&+ \gamma _{5}TUITION\_PAYER_{i} + \gamma _{6}AGE_{i} + \gamma _{7}GENDER_{i} + \\&+ \gamma _{8}SOCIO-ECON\_INDEX_{i} + \varepsilon _{i} \end{aligned} \end{aligned}$$

In all models (M1), (M2) and (M3), \(P_i\) is the probability to dropout the university (DROPOUT), and \(\varepsilon _i\) is the residual variable. The regression parameters (\(\alpha _i\), \(\beta _i\) and \(\gamma _i\)) are estimated via SPSS 16.0. The outcomes of regression coefficients, as well as their pseudo-\(R^2\), are presented in the Annex. The significance for each regression coefficient is tested with the Wald test, and the level of statistical p value is included. To keep a balanced outcome for predicted Yes and No points for the dependent variable, the cut value was set to 0.37.

3 Results

43.8% of students enrolled in the first year of a 3-year bachelor programme dropped out from the university within 5 years. The chi-square test was applied to examine the relationship between the dropout and the year of study variables: \(X^2\left( n=177256\right) =3.17 \cdot 10^3,p<0.001.\)

3.1 Dropout by Students’ Field of Study

To analyse the dropout rate in Romanian universities, calculations were made filtering those in their first year of study during the 2015–2016 academic. The dropout rates computed for the various fields of studies (with 3-year bachelor programs) revealed sensitive domains such as Philosophy, European studies and International relations, Cultural studies, Political sciences, and Geology, where all the rates are above the national average (43.7%). On the opposite side, significantly lower rates were found in Military sciences, intelligence, public order (4.6%), and in Health and Healthcare or Arts.

Table 3 Dropout rate per field of study (selection)

Regarding the hierarchy presented in Table 3, a difference that stands out between the fields of study with the highest dropout rates and those with the lowest rates can be explained by the typology of admission processes. In Romania, admission within specific fields of study is decided at the faculty/university level based on national general regulations. In this sense, universities can choose to base their admission system mainly on the baccalaureate exam or to organise other institutional admission exams. As the data shows, most of the study programs from the study fields rated as having the highest dropout rate also have among least selective admission processes, while the study programs with low dropout rates have institutional admission exams. This conclusion is also supported by Orr et al. (2017), showing that last selective admission systems are linked with lower graduation rates, while double selection systems are more efficient in terms of graduation rates.

3.2 Dropout by Student Home

Looking at the student’s origin domicile by urban-rural distribution, the analysis shows that students from urban areas are more likely to drop out (see Fig. 1). A chi-square test of independence was performed to examine the relationship between the student dwelling place by urban-rural distribution and university dropout variables. The relation was significant: \(X^2\left( n=53653\right) =35.98,\ p<0.001\).

Fig. 1
figure 1

Dropout rates by students’ domicile (n \(=\) 53653)

This can be explained, as Haj and Tuca (2021) present, by the fact that the selection process towards higher education does not favour students with a rural background. Also, students who manage to enter higher education are already the most resilient students coming from rural areas. This conclusion is also complemented by the analysis regarding the relationship between the dropout rate and the before to be student home (if the university is in the city where they live). Figure 2 shows 5% more students living close to the university (residents) susceptible to quit the first year than commuting students, benefiting from student housing or renting spaces in a different place than home, town or home village (non-residents). As numerous studies during the last 50 years have shown, living on campus or close to the university leads to better academic and social integration and, finally, to a better chance for graduation. From the chi-test it resulted: \(X^{\mathrm {2}}\left( n=60510\right) \mathrm {=1.215}*{\mathrm {10}}^{\mathrm {2}}\mathrm {,\ }p\mathrm {<0.001.}\)

Fig. 2
figure 2

Dropout rates by students’ domicile—living or not in the same city with the university (n \(=\) 60510)

In terms of geographical accessibility, perceiving university as being too familiar (living in the same community with the university one attends) increases the probability to dropout. Those who are supposed to be commuters or those who had to move out homeplace experience or have a perception of extra costs they must pay living in the city. It is also known that most of them have a very strong perception about what quitting education (and returning to their origin community) supposes in the short or medium term.

In terms of mobility the distance students have to cover, and the real costs for commuting, accommodation and other living expenses may help to understand the relevance of such variables to discuss dropping out higher education or to enforce the motivation to succeed.

Dropout by Student Home Area Local Development

Figure 3 shows that related to community development (Pana 2020), only “very low” development level of the origin community is a relevant factor for a student to drop education. This is relevant as low levels of development are mostly correlated with rural areas. Chi-test was used for the relationship between students’ home community local development level and students’ performance (in terms of graduation or dropout). This case is a relation that involves the number of all programme graduates and a 5 levels scale: \(X^{\mathrm {2}}\left( n=144017\right) \mathrm {=1.521}*{\mathrm {10}}^{\mathrm {2}}\mathrm {,\ }p\mathrm {<0.001.}\)

Fig. 3
figure 3

Dropout rates by student home area local development (quintiles) (n \(=\) 144017)

3.3 Dropout by Student’s Gender

In terms of equal opportunities, gender has always been an issue and a challenge. Our data prove that male students have a much higher propensity to drop courses than females. A chi-square test of independence was performed to examine the relationship between the gender and university dropout variables. The relation is \(X^2\left( n=58600\right) =9.28,\ p<0.001,\) as can be seen in the figure (Fig. 4).

Fig. 4
figure 4

Dropout rates by the gender of the students (n \(=\) 58600)

These results can be explained by the fact that dropout rates were calculated for 3-year bachelor programs that include female-dominated fields of study. When looking at our general data from the 2015–2016 student cohort (all programs, no matter the duration), the gender distribution of the total student population is rather balanced. If we break it down by type of study programmes, one can see the over-representation of males in 4-year programs (the vast majority being engineering programs), while we notice a greater women presence in 3 and 6-year programmes (the vast majority being humanities and medical science) (Fig. 5).

With regards to the analysis on the 3-year programs, men are less represented in the student cohort (33.8%) and have dropout rates 13% higher than women (52.5% vs. 39.3%). At the end, the percentage of men in the “graduates” population decreases from 33.8% to 28.6%.

Fig. 5
figure 5

Number of students by gender and length of study programs

3.4 Dropout by Baccalaureate Exams

Prior to entering higher education, students have to pass the high school final exam (Baccalaureate). Inequities in higher education are a continuation of the unresolved inequities in secondary education, and the results of the baccalaureate exam fully reflect these equity issues. For instance, success rates are significantly higher in urban areas for students from theoretical high schools (62.9%) than they are in rural areas for students from technological fields (37.1%) (Ministry of Education 2021). In many cases, high school students from technological fields come from low-income families or families with low educational background.

Individual Score at the Baccalaureate Exam

The best predictor of academic success is the result of the baccalaureate exam. As can also be seen in Fig. 6, the higher the Baccalaureate score, the lower the number of dropouts. We used the chi-test for the relationship between the individual score of the Baccalaureate exam (arithmetical average of grades for each Subject, from 6.00 to 10.00) and the ratio of students who drop HE studies within the first year: \(X^{\mathrm {2}}\left( n=54507\right) \mathrm {=2.906}*{\mathrm {10}}^{\mathrm {3}}\mathrm {,\ }p\mathrm {<0.001.}\)

Fig. 6
figure 6

Dropout rates by baccalaureate grades interval (n \(=\) 54507). Note: the dropout rates are calculated on a sub-population of students as data for students that passed the baccalaureate exam prior to 2004 was not available

The results can be explained by the fact that most 3-year programs use the baccalaureate exam results as the main criteria in the admission process. This means that the most resilient students in upper secondary education (who have good grades) will have more chances of accessing study programs that are their first choice, that are subsidised by the state, with scholarships.

Number of Attempts to Pass the Baccalaureate Exam

The dropout rate is much higher for students that did not pass the baccalaureate exam from the first time. This can be correlated with the previous finding as the average grade for students who do not pass from the first attempts is usually much lower. Nevertheless, an important particularity is that the statistical analysis shows a lower dropout rate for students who tried three or more times to pass the final exam, 25% more than those who tried twice and closer to those who passed the exam from the first try. This can be explained by the high level of motivation a student needs to participate in the baccalaureate exam after more than two attempts.Footnote 3 The relation \(X^2\left( n=60510\right) =6.339*{\mathrm {10}}^{\mathrm {3}},\ p<0.001,\) shows a higher probability for someone to pursue the studies (Fig. 7).

Fig. 7
figure 7

Dropout rates by number of attempts to pass the Baccalaureate exam (n \(=\) 60510)

3.5 Dropout by the Type of Study place in the First Year of Study

The Romanian Higher Education system includes within public universities subsidised study places (no tuition fees, public funded) and paid study places (with tuition). The allocation of these two categories is mainlyFootnote 4 based on merit (admission results for the first year and academic results for the rest of the program). Paid study places can be a strong challenge more than an incentive to successfully finish the studies. As Fig. 8 shows, students who pay tuition fees (themselves or the family) for various reasons at the beginning of their first year of studies have much higher dropout rates than students benefiting of public funded study places (whether they maintained this status until the end of programme or not).

The chi-test applied to those students who pay for their education (themselves or by family) shows the relation with the unsuccessful ending for the first year of university studies: \(X^2\left( n=60063\right) =1.421*{\mathrm {10}}^{\mathrm {3}},\ p<0.001.\)

Fig. 8
figure 8

Dropout rates nonpaid Study places and paid Study places students (n \(=\) 60063)

Table 4 Logit regression coefficients to measure the dropout propensity (DROPOUT)

This conclusion can be correlated with the previous finding regarding the grade at the baccalaureate exam. As the allocation on the two types of study places is based on the admission process results (which in turn, for the 3-year programs are mostly based on the final baccalaureate grade), we can also conclude that students with lower baccalaureate results have also low access to public funding and higher chances for dropout.

3.6 Logistic Regression Analysis of the Factors Influencing University Dropout

In the following, the result of the logistic regression analysis of the factors influencing university dropout is presented. As detailed in Table 4, we considered the below-mentioned variables to explain and describe the student motivation, previous educational outcomes, and integration into the students’ life. All coefficients are significant at 0.001.

Based on the pseudo \(R^{2}\) values, the explained variation in the dependent variable in our three models ranges from 9.8% to 13.37% in M1, from 1.9% to 2.6% in M2, and from 11.7% to 15.9% in M3, differences being induced by the methods of measurement (Cox & Snell R\({}^{2\ }\)or Nagelkerke \(R^2\)). The effect of tuition payments across M and M2 specifications is positive and significant (p\(\mathrm {<}\)0.001). Similarly, the effect of the first-year variable was found to be positive and significant across M1 and M2 (p\(\mathrm {<}\)0.001).

4 Conclusions

As part of the European Higher Education Area, Romania committed itself to implementing policies towards improving the social dimension of education. Even though the social dimension is part of the public debate and subsequently of the policy documents, the issue surrounding the effectiveness of the higher education system is still not high on the agenda. As at the international level, no clear definition is given, and the practices vary between states, no clear model has been proposed to measure the dropout. Moreover, until recently, the needed data was lacking in order to develop cohort analysis.

As much diversity can be found at the international level in measuring the “success” or “failure” of the student’s academic path, even more diversity can be found at the national level. In line with the university autonomy granted by the Romanian constitution and a general legal framework regarding student’s progress within a study program or university, the analysis revealed a heterogeneous system of student management that represents a challenge to any analysis of student’s progression. That is why the most accurate way to monitor students’ progression is by analysing and using administrative data at an individual level. We used the evolution of the number of students between enrolment in the first year of study and two years after the end of the program (normal duration).

In relation to international practices, one of the results of our article is the proposed definition of the dropout phenomenon and the calculation methodology. The proposed definition “percentage of the student population who failed to graduate within two years of the theoretical completion date for the study program” and the method of calculating student dropout can be used in the monitoring process both at the national and, more importantly, at the institutional level. As the data from the national student registry (RMU) starts from 2015/2016, by 2024, decision-makers will be able to monitor the dropout rate for all programs (3–6 years) in the entire Romanian higher education system. The model is in line with the international practice and takes into account Romanian specificities regarding the administrative and data collection processes as it can be replicated yearly. At the same time, there are limits generated by the data collection process that can be improved in order to better use the proposed indicator. This is the case of the socio-characteristics of the student population, as current data does not provide sufficient information regarding minority status, student medical conditions (i.e. disability), level of income or parent education. These issues can be mitigated by using representative samples from the EUROSTUDENT/Student Satisfaction Survey that include some of these characteristics.

What Do Dropout Rates Show?

In terms of what our analysis of dropout rate shows, there are several relevant findings for the debate on equity.

From a quantitative point of view, 43.8% of the students enrolled in the first year of a 3-year bachelor programme dropped out from the university (within 5 years). Even if this is more or less in line with many European states, the issue of who is failing is an important equity discussion. As some results can be perceived as counterintuitive, it is important to highlight that, as Haj and Tuca (2021) showed, many of the losses of human capital are happening before the enrolment in higher education (at the point of transition). With that in mind, looking at the dropout data, students coming outside the town where universities are located (including rural areas) are more resilient as they have lower dropout rates (\(+\)5.2% residents over non-residents and \(+\)3.2% urban over rural areas).

For the 3-years study programmes, there are relevant discrepancies regarding student dropout by study domain. The rates start from 4.6% in Military sciences, intelligence, public order and go to 64.1% for Philosophy and 59.2% for European studies and international relations. Given these results, the analysis of admission systems in higher education in correlation with graduation rates may be a further research topic, especially its impact on equity.

When looking at the student location, cultural and social student integration is a great resilient factor as living on campus brings better academic results.

Even though our analysis included dropout rates calculated for 3-year bachelor programs with female-dominated fields of study, the difference between female dropout and male dropout is significant.

We have also shown that inequalities in secondary education are also continued in higher education and reflect the failure of current equity policy instruments. Students with a low individual score at the baccalaureate exam also have high dropout risk. From previews studies, we know that they are likely to be part of disadvantaged groups and have one or multiple deprivation factors (income, health, or disability, living environment, gaps in prior education/skills, so on), but more in-depth analysis should be made on the social characteristics of dropout students.

Finally, we have demonstrated that (high) costs paid by students also bring a higher risk for dropping, this phenomenon being inflamed by the fact that access to resources is almost exclusively merit-based.

Policy Implications

The indicator on student dropout can represent a valuable indicator supporting equity policies as the provided information can support:

  • The adoption of institutional and national strategies to reduce the risk of university dropout as there is enough information in order for decision-makers to set clear targets for widening access to higher education, participation and graduation;

  • Institutional reforms, as when we take into account how the public higher education system is funded, high dropout rates can lead to financial losses for universities;

  • Improving the student support services for first-year students;

  • The implementation of equity agreements with universities in order to tackle this specific issue (including through the use of the Institutional Development Fund);

  • The monitoring of the Romanian Higher Education System through the yearly report developed by the Ministry of Education.