INTRODUCTION

Brazilian higher education is recurrently the object of formal and informal evaluations that use dropout as a major variable. Its participation is negative, that is, the higher the number of cases, the worse the perception of the institution's results. For public agencies that monitor the phenomenon, it can be quantified or measured from indicators (INEP, 2017; TCU; MEC, 2004) (The Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira, INEP, is an agency connected to the Brazilian Ministry of Education in charge of evaluating educational systems and the quality of education in Brazil.) that clearly point out the relationship between dropout rates and institutional success. In turn, the academic production (^{BAGGI; LOPES, 2011}) has produced reflections that, in some aspects, replicate the negative perception about dropout that is emanated from public agencies. However, part of the scientific reflections has focused on the criticism of the definitions in force (^{BUENO, 1993}; ^{COIMBRA; COSTA; SILVA, 2020}; LIMA JUNIOR et al., 2019; ^{RISTOFF, 1999}).

Definitions, both for public agencies and the academic universe, are very important because they establish the beacon from which the studied phenomenon will be quantified, analyzed, and evaluated. When a quantifiable phenomenon is defined, measurement tools are continuously produced. They can be simple public statistics (administrative records) or sophisticated indicators (JANNUZZZI, 2001), but they will always take the definition as the starting point for the construction of the measurer (^{FREITAS, 2016}). Therefore, once the indicator is used, it will generate important data for the diagnosis of the situation and will instrumentalize the decision of public management. Therefore, the definition of a phenomenon is strategic for the good result of a public policy.

The aim of this paper is precisely to problematize the definition of dropout officially in force and announced by INEP from its guiding document "Methodology for Calculating Indicators of Flow in Higher Education" (INEP, 2017), with emphasis on a special aspect: to find the shortcomings that make it difficult for academia and public management to tackle dropout in higher education. The question to be answered would be: is the current definition appropriate to the phenomenon of dropout adequately instrumentalizing public policies? To answer it, we opted for documental research with two methods: document analysis and data analysis. For the first case, the documentary corpus included the official texts of the federal government that define or use dropout as a variable in the organization of public policies. For the second case, microdata from the Census of Higher Education for the years 2010, 2012, and 2017 were used. The banks from 2010 and 2012 were merged, as well as those from 2010 and 2017, so that it was possible to observe the presence of students "unbound from the course" in 2010 in the other two years studied. To control the flow from one year to another, the variable "CO_ALUNO" was used, whose content is an individual and permanent identification code assigned by INEP.

The presentation of the results, in the section preceding the conclusions, will be accompanied by a previous debate on the place of dropout in higher education policies, as well as considerations on the definition of the phenomenon by the federal government. At the end, it will be possible to realize that, to a large extent, the current and official characterization of dropout in higher education is inadequate and, therefore, compromises the construction of indicators, diagnoses, and public policies. This observation induces the need for a revised definition that takes into consideration the specification of the public problem, discriminating what will be defined from the nature of the object, the motivation and the temporality (the re-entry), always observing the purposes of higher education expressed in the Law of Directives and Bases of Education.

THE PLACE OF DROPOUTS IN HIGHER EDUCATION POLICIES

Withdrawal plays a prominent role in Ministry of Education (MEC) policies for higher education. Although it does not appear textually in the Law of Directives and Bases of Education - LDB - (BRASIL, 1996), it is implied in the general principles when equality in the conditions of permanence is assured, leading to the reflection that the loss of the bond may depend on social inequalities; or when in the purposes of higher education, a section is dedicated to the graduation, making it clear that the conclusion with a diploma is a desired end.

Nevertheless, dropout can be seen in the National Education Plan - PNE - (BRASIL, 2014). The two occurrences of the expression are not allocated in the parts referring to higher education. However, in goals 12 and 13, specifically in strategies 12.3 and 13.8, the document makes clear the desire to gradually raise the average completion of undergraduate courses, which, by assumption, would mean reducing the percentage of dropout students. The Plan defines goals and strategies to guide the development of education for a decade. In retail, the National Education Council (CNE), the MEC and its policies are responsible for the regulation and operationalization of the actions. Consulting the resolutions of the CNE (MEC, 2018), it was not possible to find any that dealt with dropout in particular. In the case of recent MEC policies, however, the expression receives greater attention.

This would not yet be the case for SINAES, the National System for Evaluation of Higher Education. The reading of the law (BRASIL, 2004) that instituted it would not lead to the conclusion that dropout has some centrality in the evaluation process of higher education, since the term and its congeners (dropout) or antonyms (conclusion, graduation, etc.) do not appear there. One could use the expressions institutional effectiveness and academic effectiveness as possible synonyms for mere high graduation. However, a more careful reading leads to another perception, for one must consider that these expressions were written in clear alignment with

the promotion of the deepening of social commitments and responsibilities of institutions of higher education, through the valorization of its public mission, the promotion of democratic values, respect for difference and diversity, affirmation of autonomy and institutional identity (BRASIL, 2004).

On the other hand, the permanence policy developed by the federal government from the 2000s on has kept a prominent place in the fight against dropout in higher education. In the National Plan for Student Assistance (FONAPRACE/ANDIFES, 2007), a preparatory document for the creation of PNAES (National Program for Student Assistance), one can see the evident concern with permanence and dropout. This expression appears nine times, usually associated with the public costs involved, the representation of waste of resources, the links with retention, and the compromising of the right to education. Two specific objectives cite it textually, when it commits to:

contribute to increasing the efficiency and effectiveness of the university system, preventing and eradicating retention and dropout; [and] [... ] define a system for evaluating student assistance programs and projects by adopting quantitative and qualitative indicators for analyzing the relationship between assistance and dropout, assistance and academic performance (FONAPRACE/ANDIFES, 2007).

From this plan was born the PNAES, created in 2007 (BRASIL, 2007b), constituting itself as the main action of student assistance in the country, given its scope (all IFES) and its budget (volume close to R$ 1 billion/year in 2016). Until the birth of PNAES, the entire history of student assistance in Brazil had been portrayed as an irregular process, with advances and setbacks, not necessarily incremental and that finds its apex with the implementation of this program (^{CROSARA; BARBOSA E SILVA, 2020}), first as an ordinance (BRASIL, 2007b) and then as a decree (BRASIL, 2010b). Currently, besides its importance for the democratization of permanence, it also seems to serve the purpose of tackling dropout and the achievement of "academic success" represented by graduation.

In the same year of 2007, another government program aimed at higher education was implemented in direct dialogue with the dropout. It is the REUNI (BRASIL, 2007a), responsible for the expansion of the federal system, expanding vacancies, courses, and institutions. Moreover, it promoted an important process of interiorization and democratization of access (^{MARQUES; CEPÊDA, 2012}). Its regulatory framework, Decree No. 6096/2007, established the Program of Support for Restructuring and Expansion Plans of Federal Universities and brought as one of its guidelines the "reduction of dropout rates, occupation of vacancies and increase of entry vacancies, especially in the evening period" (BRASIL, 2007a). Thus, it is evident that there was a concern of the public administration to ensure that the expansion process was accompanied by the extension of the conditions of permanence, combating dropout and increasing completion rates.

There is another very important dimension of higher education policies in which dropout is highlighted and that, as a rule, is overlooked by the academic production. It is about financing, currently done through three sources of resources: those coming from the Federal Treasury, those collected directly from service provision, and those from contracts and agreements with public and private agencies. Of the three sources, only the funds from the Federal Treasury are directly related to the subject of this article, as they are in some way influenced by dropout³.

However, it is important to remember that these resources can be divided into two very distinct slices: the one responsible for payroll, pensions, and retirement, and the one destined to the funding and capital expenses of the IFES. Both have their total volume defined according to the inertial or incremental logic, that is, the previous year's budget is used as a benchmark, with increments associated if necessary. However, the distribution of resources of the second fraction, of funding and capital, follows the logic of formula financing (^{VELLOSO, 2000}), i.e., the budget cake is shared among the different institutions from a matrix.

Called Andifes Matrix or OCC Matrix (Other Costs and Capital), its name derives from the fact that its preparation resulted from a joint action between the Ministry of Education and the Association of Directors of Federal Institutions of Higher Education (Andifes), conferred by Decree 7.233/2010 (BRASIL, 2010a). This decree already signaled, in its article 4, paragraph 2, the parameters to be considered in the composition of the formula and that would guide the distribution of budget resources. They were

I - the number of enrollments and the number of students entering and completing undergraduate and graduate programs in each period; II - the offer of undergraduate and graduate courses in different areas of knowledge; III - the institutionalized production of scientific, technological, cultural and artistic knowledge, recognized nationally or internationally; IV - the number of patent registration and commercialization; V - the ratio between the number of students and the number of teachers in undergraduate and graduate programs; VI - the results of the evaluation by the National System for Evaluation of Higher Education - SINAES, established by Law no. 10. 861, of April 14, 2004; VII - the existence of master's and doctoral programs, as well as the respective results of the assessment by the Higher-Level Personnel Improvement Foundation - CAPES; and VIII - the existence of institutionalized extension programs, with monitoring indicators (BRASIL, 2010a).

Such parameters were not fully materialized in the OCC Matrix, as can be seen in the MEC Ordinance 651/2013 (BRASIL; MEC, 2013), which defines its current contours. Aspects such as patent registration or the institutionalization of extension programs were ignored, while others related to the production of knowledge, the evaluation of courses by Sinaes and post-graduation by Capes are so under-represented that they are not decisive in the budget allocation. The result was a reduced matrix, less sensitive to the diversity of purposes of higher education and the diversity of actions that mark its existence. Descriptively, the matrix takes as parameters the indicators of: (a) graduated students ; (b) entering students; (c) standard duration of courses (minimum time to complete credits established in the Pedagogical Project of the Course); (d) weight of the group of courses (constant established by ANDIFES assigning different weights to courses); e) retention factor (constant established by ANDIFES assigning retention weights to courses); f) off-site bonuses (bonuses assigned to courses at institutions located in municipalities away from headquarters); and g) night shift bonuses (bonuses assigned to courses whose activities occur during night shifts).

That said, it seems clear that the financing of federal higher education in Brazil, especially for cost and capital, depends on structural and institutional conditions that are not very variable, on the one hand, and on university graduation results, on the other. If one takes into consideration that universities replicate, with varying degrees of fidelity, the OCC Matrix for their local matrices, establishing budget apportionment mechanisms among academic units, one could intuit that graduation is equally important for the allocation of resources to each course, determining its capacity to reproduce and grow.

Mirroring the OCC Matrix, the National Student Assistance Program Matrix (PNAES) was also developed, a formula responsible for sharing the PNAES budget among the various IFES. The only variable that deviates from the origin of the others is the Human Development Index of the Municipality (IDHM) in which the course is located (^{MACHADO, 2017}; ^{MARTINS; ARAÚJO JÚNIOR; RODRIGUES, 2019}). Note that there is no specific indicator of students' socioeconomic vulnerability condition to guide the budget allocation of the student assistance program. This stems from the fact that a funding matrix was used whose intention was to print a budgetary logic that induces the expected academic results.

However, student assistance requires funding to combat the social inequalities that interfere with access to the right to education. Therefore, the PNAES Matrix should not replicate content from the OCC Matrix, since they are allocation mechanisms with very different objectives. Even if any emphasis is given to the HDI, it must be recognized that this indicator registers income, longevity, and education conditions of the municipality, equalizing the municipalities, ignoring social and regional origins, besides disregarding the characteristics of our higher education system, especially with the creation and implementation of the Unified Selection System (SiSU).

Therefore, we must conclude, as we did for the OCC Matrix, that the PNAES Matrix is also sensitive to the structural and institutional conditions of the IFES, as well as to the number of students who graduate or drop out, recognizing the influence of the IDHM of the municipality where the course is located. It would not be too much to reinforce, therefore, that dropout is a key factor in determining the allocation of funding and capital resources between IFES and their internal academic units, as well as for their student assistance program. Thus, it is evident that the volume of students who drop out has a relevant role in defining the budgets of the IFES and their student assistance programs, clearly signaling a governmental stimulus to increase graduation rates.

THE DEFINITION OF DROPOUT FOR INEP/MEC

The bibliography has been repetitive in using Vicent Tinto's studies as the founders of the scientific field that deals with dropout in higher education. Drawing on the introduction to his 1975 book Dropout from higher education: A theoretical synthesis of recent research, it is worth remembering that among the motivators for the text was the observation that the theoretical production, up to that date, was accompanied by two deficiencies: inadequate attention to definition and the absence of a longitudinal theoretical model. Regarding the first deficiency, Tinto said that it was very common to find works that did not differentiate "academic failures" from voluntary desertions, or definitive desertions from temporary ones; finally, they did not even discriminate transfers between institutions from abandonment (^{TINTO, 1975}, p. 89-90). The result of this undifferentiation deceived researchers by leading them to contradictory findings and incapacitated the public manager by leading him to overestimate the extent of the phenomenon, not allowing him to adequately define the target public and making him unable to formulate the proper public policy to deal with it. More than describing dropout occurrences, or associating them with "academic failures" and social conditions, Tinto wanted to build a longitudinal model capable of separating what should and should not be called dropout, as well as explaining how the intervening factors produce their effects.

The academic debate on the subject has produced articles, dissertations, and theses that give dropout reasonable importance. Generally speaking,

In Brazil there is a significant diversity of conceptions of dropout. For didactic purposes, and in order to simplify the analysis of the various existing conceptions of dropout, it is possible to group the various concepts as derived from three major matrices, which can be summarized as follows: a) those derived from the concept of dropout elaborated by the Special Commission constituted by the MEC, which analyzes dropout from three dimensions (dropout of the course, the institution and the system); b) those derived from the theory and concept established by ^{Tinto (1975}), which start from the analysis of the trajectory of the student to verify dropout; c) those that envisage dropout, as a public problem, only when there are excluding factors that do not depend on the will of the student, and that imply in the total departure of the student from higher education, disregarding mobility, for example, as dropout, as ^{Ristoff (1999}) specifies (^{SILVA et al. , 2019}).

In turn, for the institutional, administrative universe, composed of the routine of policy makers⁴, dropout has also been a very present administrative category - dear, above all, to the attentions of the Ministry of Education and its research bodies (INEP - Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira), from which the official understanding of the phenomenon starts. From a 2017 document, entitled Methodology for Calculating Higher Education Flow Indicators (INEP, 2017), one can extract the definition that interests this study. In truth, as acknowledged by its authors, the definition that we will see below originates from the federal government's analyses on basic education, adapted for higher education (INEP, 2017, p. 11). INEP understands that dropout is the

early exit, before the conclusion of the year, series or cycle, due to dropout (regardless of the reason), representing, therefore, a final condition of failure in relation to the objective of promoting the student to a condition higher than that of entry, with respect to the expansion of knowledge, cognitive development, skills and competencies desired for the respective level of education. Obviously, the interruption of the program due to the death of the student cannot be attributed as failure, given that, in general, it is a fortuitous case and it cannot be assumed an intentionality of the individual to interrupt the course, cease it, or an inability of the individual to remain in the educational program (INEP, 2017, my emphasis).

The above excerpt is rich in detail and requires a detailed reading. There are four important aspects to be analyzed: independence in relation to the reasons, the terminative condition, the failure condition, and the exclusion of death cases. First of all, dropout is defined here as an early departure, indicating the understanding that there would be a suitable time for this, coinciding with graduation. Such an exit is registered regardless of the reason, that is, without any attention to the causation that produces it. This means that, in the eyes of INEP and the MEC, the reasons for the loss of bond are not important to characterize the phenomenon, because, whatever the causes, the interpretation will be limited to understanding it as a simple loss of bond.

However, the option for this cutout raises a necessary problematization. The social reality that one wishes to inform is the loss of the link to the higher education system; however, the disregard of the reasons allows making identical situations immiscible. To illustrate the point, one can take as a reference the study conducted by the MEC and by ANDIFES in 1996 on dropouts, when the Special Commission for Studies on Dropouts in Brazilian Public Universities was formed (SESU/MEC; ANDIFES; ABRUEM, 1996). Without conducting field research, the study presents as hypothetical reasons for dropping out more than 40 different situations, aggregated into three major groups: a) factors related to individual characteristics of the student; b) factors internal to the institutions; and c) factors external to the institutions.

The first are related to

study skills; related to personality; arising from previous schooling; linked to the early choice of profession; related to personal difficulties in adapting to university life; arising from incompatibility between academic life and the demands of the world of work; arising from the disenchantment or demotivation of students with courses chosen as a second or third option; arising from difficulties in the teaching-learning relationship, translated into constant failures or low class attendance; arising from misinformation about the nature of the courses; arising from the discovery of new interests that lead to a new vestibular (SESU/MEC; ANDIFES; ABRUEM, 1996).

On the other hand, those factors internal to the institutions are

peculiar to academic issues; outdated curricula, elongated; rigid chain of prerequisites, in addition to the lack of clarity about the pedagogical project itself of the course; related to didactic-pedagogical issues: for example, inappropriate criteria for evaluation of student performance; related to the lack of pedagogical training or lack of interest of the faculty; linked to the absence or small number of institutional programs for the student, such as Scientific Initiation, Monitoring, PET programs (Special Training Program), etc.. The lack of a national public system that enables the rationalization of the use of vacancies, ruling out the possibility of enrollment in two universities (SESU/MEC; ANDIFES; ABRUEM, 1996).

And finally, factors external to the institutions include those aspects

related to the labor market; related to social recognition of the chosen career; related to the quality of primary and secondary schools; related to specific economic conjunctures; related to the devaluation of the profession, for example, the "case" of licentiate degrees; related to the student's financial difficulties; related to the difficulties of updating the university in face of the technological, economic, and social advances of contemporaneity; related to the absence of consistent and continued governmental policies directed at undergraduate education (SESU/MEC; ANDIFES; ABRUEM, 1996).

Without wishing to promote an analysis on the degree of correctness of the typology or of the aggregation of cases in each modality, it seems very fruitful to collect the recognition of the referred Commission for the diversity of causes. It is also reasonable to think that different situations may lead to equally different diagnoses, leading public management to produce specific public policies. When cases are undifferentiated, policy makers are left with few alternatives. In fact, INEP's option of defining dropout "regardless of the reason" produces an inconclusive diagnosis of the phenomenon and prevents assertive, precise, effective, and efficient public action.

On the other hand, and surely more importantly, by discriminating the cases according to three types, the Commission makes it clear that two of them are factors external to the institutions or of individual nature. The externality that leads to the loss of attachment does not allow us to associate the dropout with failures of the HEIs. In this way, the causal connections of the dropout phenomenon are taken to the wide range of personal, institutional, and social relationships, highlighting the complexity of this object. It is, therefore, an incorrect simplification to associate it exclusively with the institutional dimension, especially because the participation or weight of each group of factors in determining the volume of dropped out students is unknown.

Therefore, the invisibility of the reasons for dropping out leaves the phenomenon unenlightened and the public agency without action. What the administrative category "dropout" communicates, as defined by INEP, is of no use in indicating a public problem, since it does not allow us to know, first of all, if it even is one and whose problem it belongs to. Indeed, when a social problem becomes a public problem (^{SECCHI, 2016}), the public agency mobilizes its efforts to point out a solution. It is clear that truancy has been addressed as a problem, but it is worth questioning whether all the possible reasons for its existence lead to this condition.

Thus, we are heading towards an unsolvable situation, that is: we have chosen to ignore the reasons for the dropout and, at the same time, the MEC is required to take actions to deal with it without being given any information on what to attack. It is impossible to know if reinforcement of student assistance or curriculum revision would be appropriate; if the volumes found in distance education courses at private institutions can be treated equally to the cases of on-site courses at federal institutions; or, still, if the actions fall under this ministerial portfolio or another one, for cases involving, for example, the low attractiveness of the professional career or labor market problems.

A second aspect to be highlighted in the definition concerns the indication of its terminative condition. Early departure, before conclusion, is understood as the definitive end of the link with the course, with the institution or with the higher education system. It sounds like an academic death. On the one hand, this seems to be a vice of origin, redundant from the importation of categories applied to basic education. On the other hand, INEP's option in attributing this nature to the phenomenon is understandable. After all, if the condition of leaving were recognized as possibly transitory, the measurement of dropout would depend on the monitoring of the flow of ingression and egression in courses and institutions during a certain historical series or throughout history. If this were done, the flow would bring the need to recognize as disengaged only the person who had never rejoined the higher education system, requiring the use of indicators of gross and net dropout (discounting the rejoining). Thus, treating it as a terminative phenomenon makes it easier to measure.

At the outset, it would be important, at least, to bring up the considerations of the already mentioned study by the Special Commission of Studies on Dropout in Brazilian Public Universities. There one can read that there is an expectation on the part of the authors of the report that students who drop out will return to higher education, especially in cases where factors related to the individual characteristics of the student predominate (SESU/MEC; ANDIFES; ABRUEM, 1996). The text argues that students entering at a young age, still unsure about the professional career to follow and with little knowledge about the course they are entering, tend to promote the change as soon as their intentions and notions become clearer. Thus, dropping out and re-entering are taken there as expected components in the student maturation process, and not as a final, irrevocable process identified with failure.

This impression is reinforced by the very architecture in force until 2018 of the Unified Selection System (SISU). Before the recent change, the system responsible for the admission of an important fraction of students in the public higher education system allowed the option for two courses, and approval in the second option could be made while waiting for the definition of the vacancy in the waiting list of the first option. As soon as the vacancy of the first option was guaranteed, the student would migrate to the second option course. This possibility was responsible for guaranteeing intense mobility between courses, respecting the priority will of the candidate for the vacancy. Without making any judgment about MEC's decision to change this procedure, the undeniable fact is that mobility among courses, dropout and re-entry were part of the routine of the selection of vacancies via SISU. Therefore, dropout would not be a terminating condition for what was established by the main mechanism for access to higher education in the country.

Besides the fact that the very mechanics of entering higher education enables mobility (accounted as dropouts), the reality in the HEIs also seems to indicate that dropouts are commonly succeeded by re-entries, sometimes immediate, sometimes postponed. This is the basis that supports the text by Lima Junior et al. (2019). The article makes a fertile criticism of the validity of the Undergraduate Success Rate (TSG) and Undergraduate Completion Rate (TCG) indicators - the first used by the Federal Audit Court (TCU), and the second, created by REUNI (BRASIL, 2007a). The weaknesses pointed out can be read below:

Another problem that is present in the calculation of TSG and TCG concerns the way students are counted: usually vacancies, entries and enrollments are considered, but not the individuals themselves. With the expansion and diversification of higher education, it is increasingly common for students to re-enter to change courses or institutions. Increasingly, course dropout is not dropout per se, but mobility. One way to account for the increasingly complex trajectories undertaken by students is to track them by their physical person registration number (CPF), in a longitudinal approach (LIMA JUNIOR et al., 2019, p. 165).

This finding about the strong presence of mobility among the so-called dropouts can be reinforced when one observes the list of findings of the master's thesis of ^{Viana (2020}). The author studied the flow of the students with quota who dropped out in the year 2010, according to the Census of Higher Education. She found 1,315 cases in this condition; of these, 289 (21.97%) did not return to any higher education institution until 2016. However, 1,026 (78.02%) reentered, and, of the total number of those who had dropped out (1,315), 227 (17.26%) had dropped out again, 317 (24.10%) were studying, and 482 (36.61%) had returned and graduated. This study leads to the conclusion that the majority of this select group of students has dropped out as a postponement or a temporary suspension of the link, and not as a final condition for the loss of this link. In order to confirm what was perceived for such a small sample, a longitudinal study that covers the flow (dropout and re-entry) in the whole of Brazilian higher education is still needed.

Before moving on to the new topic of the definition of dropout in the document under analysis, it is worth mentioning another aspect of the quote above. The fact that INEP accounts for dropouts, in general, and loss of ties, in particular, based on the registration of vacancies, and not of individuals, calls attention. When we count locked enrollments, for example, we do not quantify the number of students who locked them, but how many vacancies in courses had their status suspended. This way, as an illustration, a student enrolled in three courses at the same time (two distance learning and one on-site), can cancel two distance learning courses and still attend a face-to-face college. At the end of the year, INEP will count two cancelled enrollments, even if only one CPF is the author of both.

This obvious fact should be questioned, because the same thing happens with dropout. When INEP announces the absolute and relative dropout rates, it is indicating the quantity of vacancies whose bond with a student was lost, and not the quantity of dropped out students. This finding instigated Lima Junior et al. (2019) to propose new evaluation indicators for higher education.

Resuming the problematization of the terminative condition of the dropout in INEP's definition and drawing on microdata from the Higher Education Census, in 2010 (INEP, 2011), there were 8,337,219 enrollments in Brazilian HEIs - of these, 1,138,298 (13.65%) registered in the academic situation of "unbound from the course." Besides this characterization, INEP also uses the conditions "studying", "locked registration", "transfer to another course at the same HEI", "graduated" and "deceased". Thus, under the heading "unattached to the course" are the most diverse kinds of dropouts, such as withdrawals, expulsions, graduation, transfer to other HEIs, etc.

When broken down by the HEI's administrative category, those who had left the course represented 15.08% among students of private non-profit institutions; 13.85% for private for-profit institutions; 12.46% for public municipal institutions; 11.76% for public federal institutions; and 8.79% for public state institutions. As for the academic degree, undergraduate represented 13.64% against 12.89% for bachelors. Regarding the modality, for distance learning courses, they comprised 16.01% against 13.35% for face-to-face courses. Among male students, the unattached ones were 14.88%, a higher percentage than the same picture for females (12.68%). For the racial criterion, the difference did not surpass one percentage point, but when one takes the look at the mode of admission, among students with quota, the untied were 7.07%, and among students without quota, they totaled 13.76%. Among those covered by student assistance, the unconnected corresponded to 10.55%, while those without coverage reached 13.88%. Observing the phenomenon among those who participated in academic activities (scientific initiation, monitoring, extension etc.), the volume reached the level of 6.60%, while among those who did not participate, the figure rose to 14.56%. As for the shift, the values were higher among night shift students (14.20%), followed by morning (12.5%), afternoon (11.04%), and full-time (10.03%) students. Finally, when the age bracket is observed, the unattached represented 15.2% among students 25 years old or older and 12.08% among students between 18 and 24 years old.

To verify the terminative nature of the dropout, it would be important to note that the volume of students who lost their link did not return to the higher education system. However, part of the volume disconnected in 2010 was registered as an enrolled student in later years. By way of illustration, 53.96% of the students registered as "disconnected from the course" in 2010 were enrolled in 2012 (INEP, 2013), two years later. Not all were registered as re-entrants, as some did not even exit the system. What seems strange, in fact, results from the way dropout is counted: by vacancy, and not by student. It is already known that there were 8,337,219 enrollments in 2010, but only 7,610,696 students, of which 6,938,739 (91.17%) had one enrollment and 671,957 (8.83%) had more than one enrollment (from two to nine).

In this way, a student with more than one enrollment can have been accounted as an escapee (terminating condition), when he or she has only closed one of the links, keeping the others. At this point, it is worth pondering that, if one of the purposes of higher education is graduation, the loss of one of several links with the system should not be treated as a problem in itself, since the conclusion may occur in the course in which the enrollment was kept - or, on the other hand, it could also not be considered as a loss of link similar to that of those who have one single and exclusive link.

Following the same reasoning, it is worth saying that, if in 2012, 53.96% of the so-called 2010 dropouts were still enrolled, in 2017 (INEP, 2018) this value was 28.21%. Thus, one out of every four, classified as evaders in 2010, was not disengaged from the higher education system seven years later. In this way, the so-called dropout is not final for a good part of the students who leave the HEIs, and they make re-entry or the reduction of multiple ties a usual stage of the academic trajectory.

Table 1 indicates that students who left the HEI in 2010 and were in the bound condition in 2012 are mostly in the "studying" situation, followed by the situations "unbound from the course", "locked enrollment", "graduated", "transferred to another course at the same HEI", and "deceased". Note that the profile of the academic situation of the re-entering students is very similar to their peers, that is, if we compare the percentage composition expressed in Table 1, we will find numbers similar to those that express the academic situation of all students from the HEIs in the respective censuses. By way of illustration, in the year 2012, of the total number of students in the HEIs in Brazil, 62.7% were studying; 10.2% had locked enrollment; 15.1% were unattached to the course; 1.0% had transferred to another course in the same HEI; 11.0% had graduated; and 0.0% had died (INEP, 2013).

In any case, we can already point out that the reality of Brazilian higher education is permeated by the loss of ties without this necessarily meaning a final condition. It will be up to the official bodies to promote the adaptation of their diagnostic mechanisms so that they are able to read the reality as it is, that is, perceiving the flow as an inherent part of the higher education formation process. As long as the current definition is maintained, it will be doomed to produce a skewed portrait, a simulacrum of dropout, a constructed image of a phenomenon alien to reality. The findings should return our attention to the way INEP accounts for dropout and identifies it with failure, instigating diagnoses, evaluations, and public policies. Reducing to observe the gross (without discounting re-entries or multiple ties) and static (restricted to the contours of a year) loss of attachment, its net and dynamic dimension (observing the trajectories) is lost.

In this way, we can now devote some attention to the third aspect of the definition of dropout in the INEP document. When the text indicates the terminative condition of dropout, it does not do so in a vacuum, but rather as a representation of the failure to reach a higher level (of knowledge, cognitive capacity, competencies, and skills) in relation to the entry level (INEP, 2017). Here, two considerations are worth reflecting on.

First, the understanding of dropout as failure or failure - extremely common in specialized literature (^{BAGGI; LOPES, 2011}) - is built with the intention of producing an evaluation, a measurement of results, on higher education. It should be clear that, strictly speaking, considering dropout as an evaluative mechanism requires that this is done within a general exercise of institutional evaluation (^{RISTOFF, 1999}), thinking about its relationship with the various dimensions expressed in the educational activity for this level of education. However, once the survey of the situation of loss of attachment is done without the perception of motivation, it becomes impossible to evaluate whether the exit represents a failure or not. And, if it was used and showed failure, the motivation would also allow the student, the course, the institution, the Ministry of Education or society to attribute it. Since it is not perceived, this attribution cannot be made.

Secondly, the notion of failure presumes to be able to evaluate as null the level of accumulation of knowledge, cognitive capacity, skills and abilities of an evaded student. Taking to the ultimate consequences the expectations that the definition itself throws upon itself, it would understand as fully successful a case of graduation, just as it would interpret as indifferently unsuccessful the dropouts in the first or last period of the course. On the one hand, one should not conclude that the access to the diploma has conferred the desired accumulation, just as one cannot assume, without further evaluation, that the early departure has not allowed to fulfill the proposed purposes. The definition present in the INEP document elevates the student's graduation to the condition of an absolute criterion for the attribution of success.

If the results extracted from the application of the National Exams of Higher Education (ENADE) are observed, it can be seen that the construction of the evaluative indicators contemplates the accumulation of knowledge during the formative period. This is the case of the Indicator of Difference between Observed and Expected Performance (IDD). The IDD "is a quality indicator that seeks to measure the value added by the course to the development of the concluding students" (INEP, 2019). The very low IDD of some students must indicate that the conclusion of the course did not allow a substantive aggregation of cultural capital during their passage through higher education. Now, the combination of the presence of the IDD and the possibility of gauging low results is proof that graduation cannot be taken as an absolute criterion of success.

On the other hand, since ENADE is not applied in series, but only at the end of the course, it would not be possible to measure the value added during each stage of the educational process. Thus, and using the IDD as the only national measure of the accumulation of cultural capital for higher education, in all cases of anticipated departure before graduation, it would not be possible to state that this implies absolute failure.

In fact, attention should be drawn to the absence of a closer link between the notion of failure (and success) present in INEP's definition of dropout and the purposes of higher education established by the LDB (BRASIL, 1996). The law is clear in indicating as purposes:

I - stimulating cultural creation and the development of the scientific spirit and reflective thinking; II - forming graduates in the different areas of knowledge, who are apt for insertion into professional sectors and for participation in the development of Brazilian society, and collaborating in their continuous formation; III - to encourage research work and scientific investigation, aiming at the development of science and technology and the creation and dissemination of culture, and thus develop understanding of man and the environment in which he lives; IV - to promote the dissemination of cultural, scientific and technical knowledge that constitutes the patrimony of humanity and to communicate knowledge through teaching, publications or other forms of communication; V - fostering a permanent desire for cultural and professional improvement and making it possible to achieve it by integrating the knowledge being acquired into an intellectual structure that systematizes the knowledge of each generation; VI - stimulating knowledge of the problems of the present world, particularly the national and regional ones, providing specialized services to the community and establishing a reciprocal relationship with it; VII - promoting extension, open to the population's participation, aimed at disseminating the achievements and benefits resulting from cultural creation and scientific and technological research generated in the institution. VIII - to act in favor of the universalization and improvement of basic education, through the formation and training of professionals, the performance of pedagogical research and the development of extension activities that bring the two school levels closer together (BRASIL, 1996).

Thus, the notions of success or failure, understood as the measurement of positive and negative results, should be observed in the light of what is understood as the purposes of higher education. Law 9.394/96 does not list the purposes of higher education in a hierarchical or conditioned way, leading one to believe that the legislator's interest was to confer equality to them. This being true, it is worth highlighting the fact that, of the eight purposes, only one is strictly related to graduation for work, the second one. In all the others, it is possible for the student to develop them fully or almost fully during the course without necessarily obtaining a diploma. This is not due to the fact that the LDB is negligent about the diploma, after all, each pedagogical project of the course seeks to materialize the accumulation of skills and competencies in the credits distributed throughout the curriculum. However, the LDB does not seem to reduce university activity to the manufacturing of graduates or the formation of a workforce. In this sense, the referred law is faithful to the constitutional content when the magna carta asserts in its article 205 that education aims at "the full development of the person, its preparation for the exercise of citizenship and its qualification for work" (BRASIL, 1988). It is clear, therefore, that it would be reductionist to identify the fullness of the person's preparation, of his citizenship and his capacities for work to the acquisition of a diploma.

This being said, it can be seen that the content of the aforementioned law (LDB) attributes to the university a prominent role in the civilizing process, in the direction of a more just, democratic, developed, and plural society. The evaluation of university success or failure should, strictly speaking, take into consideration the totality of purposes, without reduction to the dimension of the labor market or of the conclusion of educational credits. Similarly, any other university phenomena, such as dropout, should be subject to judgments taking into account the range of purposes for which the system is intended. This would require considering loss of attachment to higher education as a problem in specific situations, depending on the causes and on whether or not the purposes of higher education are being fulfilled.

To this end, the evaluations of academic success or failure should rather focus on ways of perceiving the student's participation in the process of cultural creation, development of the scientific spirit, of reflective thinking, of understanding the human being and his environment, of spreading knowledge with the desire to improve it, as well as to expand them on national problems and develop services to the community, extending achievements and benefits, as well as contributing to the improvement of basic education.

Carrying out the recommendation, we should ask: what percentage of the student body has participated, during its academic career, in scientific initiation and university extension programs? How big is the fraction of the same body that has taken an interest in national problems, reflected on the theme, produced some knowledge and put it in dialogue with the community? Who has created and spread culture? Who has participated in the process of improving Brazilian basic education?

Note that none of these questions can be answered if we only have access to the quantitative numbers of graduates, failures, failures and dropouts, as is usually done. Therefore, an evaluation process capable of capturing the forgotten aspects of the purposes of education is required.

Finally, the last aspect of the definition under analysis to be problematized has to do with the exclusion of the calculation of the cases of death from the universe of those who drop out. In this way, besides cases of internal transfer (within the same Higher Education Institution), deaths will not be recognized as dropouts either. Such recognition is due to the fact that INEP treats it as a fortuitous case, representing an unintentional withdrawal from the course, or one that does not reveal the individual's inability to remain in the educational program (INEP, 2017).

Apparently, the definition has plunged into a paradox in terms. As is known, INEP purposely declined the possibility of investigating motivation, and this fact has impelled this paper to charge that such a choice nullifies the possibility of observing dropout as a problem. However, it resorts to a motivational argument to exclude deaths from accounting for dropout. When appealing to the judgment on the fortuitous nature, unplanned or unlinked to intention, INEP is indicating that it is unable to associate failure to death, a situation that requires ignoring its computation. Now, but this is precisely the point. When it is not possible to do such an exercise on all the cases of loss of bond, it cannot be said that the failure, the failure, even exists.

If from the semantic point of view one can perceive the paradox, from the empirical point of view it becomes even more evident. After all, there would be several cases of dropouts (other than death) that could also not be perceived as unintentional, fortuitous, and not predictive of the student's disability. Social, family, physical conditions, among others, could lead to the loss of bonds contrary to the intentions of the dropouts. For these cases, to whom should failure be attributed? In the same way, none of them would be predictive of possible incapacities of the student and, perhaps, not even of the institution. In the same vein, if it is about an dropout originating from a belatedly perceived vocation, to whom should the failure be attributed? In fact, is there any failure?

That being said, the definition offered presents insufficiencies, as it does not diagnose the phenomenon well, is unable to identify a concrete public problem, does not adequately guide public policies for its confrontation, does not reflect the reality of bond losses, and is based on a paradoxical construction. What emerges from the measurement of the phenomenon thus defined would be a reality different from the existing one, a bias, a simulacrum. The problems raised by ^{Tinto (1975}) for most of the literature specialized in the dropout phenomenon can be found in INEP's document, namely: the inadequate attention given to the definition and the absence of a longitudinal theoretical model, both leading to undifferentiated cases, the inability to diagnose and the immobilization of public policies.

FINAL CONSIDERATIONS

By taking as object of analysis the definition of dropout that is present in the document "Methodology for Calculating the Flow Indicators of Higher Education," published by INEP in 2017, it was realized that four problems could be responsible for hindering diagnoses and prognoses of higher education in Brazil.

Initially, the option to identify dropout to any loss of link with higher education, ignoring causalities, generated a survey that places in the rubric of dropout very distinct phenomena. This indistinctness does not allow an accurate diagnosis and prevents tackling the cases that should be treated as a problem. In the same way, the current and official definition considers dropout as a definitive action, ignoring re-entries and multiple links. Here we can see the importance of using longitudinal studies to determine what should in fact be called dropout and recognized as a problem.

Another relevant aspect concerns the immediate association of dropout with academic failure, without the necessary mediations capable of perceiving that higher education has purposes that go beyond the completion of credits and graduation.

Finally, the definition plunges into a paradox by excluding from the dropout calculation the cases of death by virtue of its fortuitous nature, forgetting that such exclusion is, in fact, an exercise of identification of causality of the loss of the bond, precisely what the INEP document did not want to do.

This being said, it seems important to conclude that a revision of the content of the definition of dropout will be necessary, so that it can more adequately express the reality, better instrumentalize the diagnostic mechanisms, enable public management to identify what the real problems are and act on them, always in the light of what the Law of Directives and Bases of Education establishes regarding the purposes of higher education. The urgency of the revision is also due to the fact that, as we have seen, the volume of dropout/graduation has been used as an important variable to guide public policies and the very financing of federal education and, within it, of student assistance.