Roma Population in the Spanish Education System: Identifying Explanatory Frameworks and Research Gaps

This chapter makes a literature review based on the Grant and Booth qualitative systematic methodology of the studies about the educational situation of the Roma in Spain, with an wider, extended scope that allows to compare the findings with those conducted on other countries’ Roma populations. Studies on the Roma educational situation in Spain are hindered by the lack of official, periodical statistics, having to rely on sample-based surveys and ethnographic studies. In spite of the inaccuracy of the studies all of them show, as a general picture, a staggering educational gap between the Roma and the rest of society which is common to all Western countries. Most of the studies on Roma education have concentrated in this negative aspect. Numerous theoretical frameworks have been developed to explain this staggering education gap. All them acknowledge the phenomenon as a multidimensional one but for heuristic purposes they can be ordered along an endogenous/exogenous factors continuum depending on how much they stress the weight of factors stemming from characteristics of the Roma ethnic group itself or, on the contrary, of the majority non-Roma society. The literature review has also identified an emergent critical current that sees this studies focused on educational underachievement as a sharing a common essentialist bias that helps to reinforce the stigmatization of Roma and have turned to focus, instead, on the study of academic success among the Roma. Although this emerging field is very promising, our review has identify several significant research gaps in this regard: a lack of longitudinal studies, a lack of studies on the Roma upper and middle classes and a lack of studies on Roma students in post-compulsory education, particularly the university level. This article encourages researchers to fill this gaps with the conviction that the knowledge obtained can help combat the negative stereotypes and the self-fulfilling prophecy effect that approaches focused on Roma underachievement may have.

Introduction: The Statistical "Invisibility" of Roma People in Spain and the Scarcity and Discontinuous Nature of Quantitative Nationwide Studies One of the problems that hinder an accurate knowledge of the educational characteristics of Roma in Spain is the lack of census data on any sociological dimension concerning this group. The misuse of ethnic censuses along the twentieth century as a political weapon to commit ethnic discrimination or cleansing led to the abhorring of this statistical tool in many countries and the passing of laws banning the collection of ethnic, racial or religious data at the individual's personal level. However, the elaboration of anonymous statistical ethnic censuses has never been banned by any international legislation. On the contrary, the current standards set by the United Nations, the Council of Europe or the European Union (Directive 95/46/ CE) recommend the recollection of data about the ethnic composition of the population through national censuses and other periodical statistical instruments, because the "visibilization" of ethnicity is seen as an effective mean of preventing discrimination (Simon 2007;Arp 2009;Estévez Hernández 2015). As has been highlighted by many researchers, accurate knowledge about the Roma population is a key prerequisite to designing well-targeted policies and monitoring their outcomes (McGarry and Tremlett 2013). Many countries with Roma populations have, indeed, ethnic censuses and statistics. This is the case of most states in the American continent and some in Europe (Kertzer and Arel 2002) but not many among them include the Roma as an specific ethnic group for the gathering of data, and the information about them is merged (and, thus, lost for analytical purposes) within a residual category of "other ethnic groups or races", as, for example, in the USA Census (US Census Bureau 2010). We dare say that the Roma are one of the most "invisible" ethnic groups in the Western world. Currently, only a few countries having Roma minorities collect specific census data about them: Romania (Institutul Naţional de Statistică 2011), Hungary (Hungarian Central Statistical Office 2011), Bulgaria (National Statistical Institute of Bulgaria 2011), the Czech Republic (Czech Statistical Office 2011) and Colombia (Sistema Estadístico Nacional de Colombia 2018). To that list, we will have to add the UK in its 2020 census (Office for National Statistics 2019) and maybe Brazil (Ministério Público Federal do Brasil 2018). It is not the case of Spain, though, where all administrations have resisted to collect census and other statistical data on the Roma. And that, in spite of recommendations from the European Commision and pressures from the UN Committee for the Elimination of Racial Discrimination to do so (Arp 2009). All Spanish governments have defended this position affirming that asking people about their ethnic origin would be a violation of the constitutional mandate that upholds the equality of all citizens (Arp 2009;Estévez Hernández 2015). However, nothing in the Spanish legislation forbids the collecting of census ethnic data. The two organic laws on data protection (1999 and its 2018 amendment, aimed at aligning the Spanish legislation with the European Union Data Protection Directive) allow and regulate the collection of statistic ethnic data provided it serves the public interest and some safeguards (anonymity and consent of respondents) are respected (BOE 1999(BOE , 2018. In fact, ethnic data about the populations of migrant origin have been gathered by the Spanish National Institute of Statistics (INE) since its 1991 Census (INE 1991) and regional governments where a language other than the common Spanish one is spoken, have being producing what can be considered as some sort of "ethnic" censuses-language being a proxy for ethnic identity-of Catalan (IBESTAT 1991;IDESCAT 2011, SIES 2015, Basque (EUSTAT 2016) and Galician (IGE 2018) speakers in their territories since that very same year. As a result, the Roma turn out to be the only ethnic group lacking official statistics in Spain. Because they don't have an ancestral territory where they mostly concentrate, unlike the rest of Spanish ethnic minorities, but are dispersed across the whole country, don't speak their original language anymore and lack an affluent and educated bourgeoisie to lobby for their interests, their influence in the Spanish post-franquist identity politics has been weaker than any other national ethnic group. They have never been considered eligible to be granted any kind of political autonomy and although they undertook a significant organizational process this was in many cases guided by external actors (the oldest and some of the major Roma organizations, like Fundación Secretariado Gitano or Secretariado Gitano de Barcelona were founded and are still administered by the Catholic Church) and rather fragmentary in nature, resulting in a plethora of local and regional associations that operate under the legal status of NGOs. To that we must add the deep cultural and social chasm that exists between the Spanish majority group (Gitanos) and the more recently arrived Portuguese (Ciganos) and Eastern European Roma. Had they had a strong and unified lobbying body, the Spanish Roma could have exerted a stronger political pressure possibly advocating for a reform of the state official statistical tools, as a way of coming out of the state of "invisibility". But for many years they never even expressly asked for it.
The situation has somehow changed since 2005 when the Spanish government fostered the creation of the State Council of the Roma People, that brings together most of the existing associations and acts as an advisory body on social programmes and policies aimed at the Roma population (Royal Decree 891/2005). In 2017 this Council backed the first initiative to make an ethnic census for the Roma, a proposal that came from the Andalusian regional government and was meant to be applied exclusively to Eastern European Roma living in Andalusia, a sub-group mainly consisting of nomadic or highly-mobile families for whom data, or even estimates, are almost nonexistent (Junta de Andalucía 2017). For the first time, the statistical "invisibility" of the Roma was acknowledged as being a hindrance for the provision of social services and affirmative actions by a Spanish public administration. In spite of this, the proposal was criticized by some sectors of the Spanish and Gitano society alike, a reminder of how persistent the negative image of ethnic censuses still is in Spain. Critics compared the Andalusian proposal with the one launched the same year by Italy's Ministry of Home Affairs, Matteo Salvini, and the State Council had to publicly defend its decision by pointing out the completely opposite purposes of both initiatives (FSG 2018b). In 2019, the Andalusian Autonomous Region changed administration and the proposed census has not been carried out yet as of the date of this writing.
The Roma population in Spain is, therefore, still statistically "hidden" inside other ethnic or national categories. Following a widely accepted consensus, we will use the term Gitano to refer to the native Spanish Roma, the term Cigano to talk about those recently arrived from Portugal while reserving the most generic terms Rom, or Romani for those of Eastern European origin. Statistically, the Gitano are "hidden" inside the native Spanish population group and so are, respectively, the Portuguese Ciganos and the Eastern European Rom (with many individuals from the latter group completely unrecorded by any official statistic due to their nomadic way of living (Gamella 2007;Piemontese 2017)).
As a consequence of this, quantitative nationwide data about Roma in Spain are based on sample surveys whose margin of error cannot be estimated accurately for several reasons: (1) the total number of the studied population is an estimate itself.
(2) all surveys, and even censuses, dealing with ethnic identity are confronted with very important methodological unresolved biases (Stavenhagen 1992;Messing 2014): Figures can change significantly depending on the identification criterion applied: self-identification (individuals are only classified as Roma if they themselves declare to be Roma), external categorization (the interviewer decides if the person is Roma) or some combination of these two. The surveys conducted in Spain have not been consistent in the application of a single standardized criterion. 1 With regard to the self-identification criterion, Roma, as other ethnic minorities historically subjected to stigmatization and acculturating pressures can be reluctant to acknowledge their identity or can have double or multiple identities which are not often reflected in the questionnaires. 2 On the other hand, using an external criterion may lead to the alteration of the results due to the subjectivity and racial/ethnic prejudices of the interviewer.

Methodology
This work shows the results of a qualitative systematic review (Grant and Booth 2009). This technique is a narrative method for integrating and comparing findings from qualitative studies. It looks for 'themes' or 'constructs' that lie in or across individual studies. It may employ selective or purposive sampling, using quality assessment (Grant and Booth 2009). The method was selected because most 1 As an illustration of this the 2007 and 2013 FOESSA surveys were implemented using external categorization ( FOESSA 2008( FOESSA , 2014 whereas the 2018 FOESSA survey applied the selfidentification criterion (FOESSA 2019). 2 As an illustration of the significantly potential impact of this bias, when the 2011 Hungarian census changed the question on ethnic identity to allow co-equal dual identifications, the number of Roma, as measured by the census, increased by 53% (Messing 2014) and, in Spain, when the FOESSA survey changed its methodology from hetero-categorization to self-identification the total population estimate of Roma population in Spain went down from 970,000 to 560,000 (Hérnandez et al. 2019). published studies on Roma education are qualitative and also because of the complex nature of the phenomenon. Firstly, a search was performed in the scientific databases Web of Science and Google Scholar. Subsequently, different themes were identified and classified. Within each theme, shared aspects and differences among the publications were identified. Thirdly, the main study limitations as well as proposals of future lines of research were identified and analysed.

A Review of Quantitative Studies on Roma Education in Spain
The fact that there are no census or systematic statistical data about the Roma doesn't mean, however, that quantitative nationwide studies are completely non-existent. It doesn't mean, either, that the Spanish state has completely neglected the quantitative study on the Roma. What it really means is that those studies are sample-based, methodologically dissimilar, scarce and non-periodical. All the nationwide studies have been, in fact, conducted by only three organizations: the Centro de Investigaciones Sociológicas, the Spanish official organism responsible for the conducting of sociological surveys, the Secretariado Gitano Foundation (FSG), the biggest Roma NGO in Spain, linked to the Catholic Church, and the FOESSA Foundation, a think tank of Cáritas Spain, the biggest Catholic Church charity. All the surveys conducted by the FSG since 1991 have been undertaken in collaboration with the Spanish government. The CIS and FSG data are, thus, the closest thing we have to some sort of official statistics on the Roma.
The first quantitative nationwide study was the one conducted in the 1970s by the Asociación Secretariado General Gitano, a previous name of the FSG (FSG 1990). This study took the first snapshot of the Gitano society (by then migration of Portuguese Ciganos and Eastern European Rom to Spain hadn't yet started) in relation to demographics, family structure, housing, education, employment and economic activity, health, religion practices, cultural values and discrimination. It also gave us the first estimate of the Gitano population, 208, 344 people. The next one was something close to a comprehensive census of Roma dwellings and provided a second population estimate: 296,225 (FSG 1991). It was the pioneer of a series of dwelling censuses conducted by the FSG, in a non-periodical fashion (FSG 2007(FSG , 2018a. The total population estimates, always based on external categorization, provided by those censuses were 459,088 and 516,908, respectively. The 2007 study gave, for the first time, disaggregated data of the Roma subgroups, based on self-identification (93.7% Gitanos, 1.8% Ciganos, 3% Eastern European Rom). This dwelling censuses have become the main sample-building tool for subsequent surveys, introducing in them two significant underrepresentation biases (that of the nomadic and highly mobile Rom groups, and that of the most integrated Gitano families, which are not included in the census).
The FSG has been quite prolific in the conducting of other nationwide quantitative surveys on several topics. It is worth mentioning the last comprehensive study on social exclusion, with a section devoted to education (FSG 2019). This is in sharp contrast with the scarcity of studies undertaken directly by public organisms, only three: the 2007 survey undertaken by the CIS on its own, which investigated a variety of topics (CIS 2007) and the two National Surveys on the health conditions of Roma population (La Parra 2009;MSSI 2018). The FOESSA Foundation has been conducting surveys on the social conditions of the Spanish most impoverished population segments at almost regular intervals since the 1960s but only in the last three ( FOESSA 2008( FOESSA , 2014( FOESSA , 2019) they asked about the ethnic identity of the respondents, allowing to disaggregate the data on the Roma from those of the overall population. The FOESSA surveys themselves, however, because of their encompassing scope, don't contain very detailed information about the Roma: this must be found in works published separately (Laparra Navarro 2008; Damonti and Arza Porras 2014;Hernández Pedreño et al. 2019). The FOESSA surveys provide us with yet other population estimates and illustrate how difficult is to come up with an accurate figure: the 2007 survey (FOESSA 2008), using an external categorization criterion, estimated a population of 970,000; in the 2018 survey (FOESSA 2019), which changed to a self-identification criterion, the estimate was 560,6769. This last figure is closer to the one given by the 2015 FSG dwelling census (FSG 2018a) (but this study warned about the number having forcibly to be bigger, since only took into account the sedentary Roma, most of them Gitanos, living in highly ethnicized neighbourhoods). The estimate given by National Survey on the health conditions was 665,987 (La Parra 2009). As we can see, the actual population figure remains unknown. This must always be taken into consideration when discussing the data on Roma education.
As far as the specific quantitative studies on Roma education are concerned, the studies are equally sample-based, discontinuous and can be reduced to a handful of works. The Spanish state annually publishes statistic reports on the education system since 1963, first through the INE (1964INE ( , 1985 and then through the Ministry in charge of the education system, which has gone under different names (MEC 1986(MEC , 1995MECD 2000;MEF 2019). None of these reports has ever included data on the Roma student population. In contrast, the reports do include the total amount of students by national origin (although no other specific information beyond that figure). To this statistical "invisibility" we must again add the absence of an official policy committed to the conduction of periodical nationwide studies on the educational situation of the Roma population. And this, in spite of the existence of specific policy instruments aimed at reducing the historical marginalization of Roma in Spanish society (MSSI 2012). The pioneer work by the Asociación Secretariado General Gitano contains some general data on literacy, dropout rates and maximum level of education attained (FSG 1990). From there we have to wait till 2006, with a study on the schooling of Roma girls in secondary education (FSG 2006). The CIS survey (2007) included only two questions about education (CIS 2007). The most relevant and comprehensive studies up to date were the ones conducted by the FSG on primary (FSG 2010) and on secondary education (FSG 2013). To these we must add the comparative survey conducted by the European Union Agency for Fundamental Rights on Roma from all European countries (FRA 2014). Since then no other specific quantitative study on Roma education has been conducted in Spain. Relevant nationwide quantitative data on education, though, can be extracted from the FOESSA (Laparra Navarro 2008; Damonti and Arza Porras 2014;Hernández et al. 2019) and FSG surveys on social exclusion (FSG 2019).
There is a plethora of studies about the educational situation of Roma in Spain, and their number has gone up at an increasing rate as decades went by (Haz-Gomez et al. 2019). However, all those other works are of an ethnographic, fundamentally qualitative, nature (74.3%, according to Haz-Gomez et al. (2019) with an additional 10.3% combining ethnography with limited quantitative data collection) and all of them use some of the mentioned studies as their source of nationwide quantitative information.

Roma Education Studies: The Hegemony of a Perspective Focused on the Roma Educational "Anomaly"
In spite of 40 years of social and education policies aimed at achieving the full convergence of Roma with the rest of society, this goal is still very far from being attained. Whether coming from quantitative or qualitative studies the evidence is overwhelming: high levels of absenteeism, low performance and high drop-out rates that increase sharply in the transition to the second cycle of secondary education, being significantly higher in the case of Roma girls. Here are some illustrating figures from the most recent nationwide quantitative studies. We want to call your attention to the differences in estimates between sources, which clearly demonstrates the already discussed inaccuracy of the available data: • The illiteracy rate of Roma overall population was 7% in 2013 and 5.5% in 2018 (Hernández et  Regardless of their inaccuracy, the figures reflect the undeniable existence of a huge educational gap between the Roma and the overall Spanish population. Some studies lead to hypothesize that this gap might also exist when compared with minorities of migrant origin. The Ministry of Education doesn't publish any data about the educational achievement of migrant students and unlike the case of the Roma, there is no nationwide survey and the number of ethnographic studies is low. 3 Thus, as far as the education field is concerned, the migrant population suffers from its own kind of "invisibility". Nonetheless two regional quantitative studies have been published recently, shedding some light on the matter. The first one shows drop-out rates of secondary migrant students in Catalonia to be similar to the rate for native Spanish (Moroccans, 15.9%, Latin Americans, 15.5%, Chinese, 13.6%) with the only exceptions of Sub-Saharan Africans (40.9%) and Pakistanis (35%) (Bayona and Domingo 2018). The second gives a general drop-out rate for the migrant student population of Murcia, 43%, which is also lower than the one for the Roma (Frutos and Barba 2018).
This phenomenon is what we refer to as the "Roma educational anomaly", something that has been consistently reported in all European and American countries having Roma minorities (Meunier 2007;UNESCO 2007). Radojević (1984) wrote of it as a "peculiarity" and Jurova (1994) labelled it a "handicap". This educational gap has been perceived by many scholars and policymakers as one of the fundamental dimensions of the Roma social exclusion, where education and social marginality are caught up in a systemic loop, being at the same time cause and consequence of each other. A situation that constitutes a major social problem and has, consequently, led to an enormous proliferation of research on the subject, the studies being carried out by different social actors concerned with the plight of the Roma collective: researchers from all sorts of fields, public policy-making institutions, NGOs and Roma organizations.
The first relevant academic works on Roma education came from the United States (Vogel 1979;Vogel and Elsasser 1981;Kaldi 1983) and Britain (Reiss 1975;Worrall 1979) as a part of the boom on ethnicity studies of the 1970s. They will take off in the rest of the Western world from the 1980s onwards. According to Liegeois (1994), schooling is the only field regarding the Roma minority to have been subjected to systematic analysis in almost every European country (Radojević 1984;Jurova 1994;Liegeois 1999;Gomes 1999;Myers and Bhopal 2009;Trentin et al. 2006;Forray 2002;Ringold 2001;Kyuchukov 2000). Spain, being one the countries with the largest Roma population, pioneered the foray into this sociological territory (González 1970;Garrido 1980;Iniesta 1981;Palacios and García 1986). In the twenty-first century we see also studies proliferating in the Americas (Bereményi 2007a, b;PRORROM 2007;Bel 2011;Niquetti 2013;Gewald et al. 2013;Walsh et al. 2008;Hancock 2013;Silverman 2012 ) although with a lower number of publications than in Europe, where Roma populations are larger. And yet, the phenomenon has not been covered or studied in all its possible dimensions. This is so because most research, whether in Spain or other countries (EUMC 2006), has predominantly focused on the "anomaly" rather than the more encompassing and multidimensional phenomenon of Roma education itself. Why this has been so can only be hypothesized. Perhaps it has to do with the predominance the applied dimension of the research about Roma has attained in the last decades, at least in Europe, fueled by the abundant funding coming from social inclusion programmes. Whatever the reason, it seems that Social Sciences have tended to regard Roma education with the eye of a clinician, as a social malady to be cured rather than as a mere sociological subject to be simply analyzed. This could explain why the literature is overwhelmingly focused on the negative factors of the Roma education phenomenon (Abajo and Carrasco 2004), because they are the most conspicuous and the most alarming ones for a majority of researchers that don't only search for explanations but for urgent solutions, for a way to "normalize" the "anomaly" (even if this "normalization", is understood in the light of the postmodern paradigm and entails "inclusion without identity loss"). The Roma education field has been hijacked by the agenda of social inclusion policies. As the sociological and educational "anomaly" persists till this date, the number of studies keeps growing in both continents at an ever quicker pace (Helakorpi et al. 2020 The explanatory frameworks developed to account for this "Roma educational anomaly" have been numerous but all of them have at least one thing in common: they all agree the phenomenon is of a systemic nature and stems from some complex set of interrelated factors. A review of the literature has allowed us to make a list of the different factors identified by the researchers and sort them out in four broad categories resulting from the intersection of two dichotomous categorical axes: (a) An endogenous/exogenous axis: that is, factors associated with the social and cultural traits of the Roma themselves versus factors that have to do with the characteristics of the majority society. (b) A structural/cultural axis: factors that have to do with social institutions, institutionalized roles and functions in the political economy versus factors stemming from the systems of values (those of the Roma and those of the non-Roma populations). Combining both axes, factors can be classified into a four-fold categorical scheme (endogenous-structural, endogenous-cultural, exogenous-structural, exogenous-cultural factors).
Although most of the works take into account, to some extent, all four dimensions, explanatory frameworks can be ordered along a continuum from those who give more weight to the endogenous factors to those stressing the exogenous ones. On the first end of the continuum it is more likely to find works written by non-Roma academics and, more particularly, the earliest studies (Garrido 1980;Iniesta 1981;San Román 1984;Fernández Enguita 1999). On the second end we can locate those works conducted by NGOs (some of them Roma) and more recent academic studies (Flecha and Soler 2013;Padilla et al. 2017), some of them conducted by an increasing number of Roma researchers (Heredia Maya 1980; Santiago and Maya 2012; Macías-Aranda and Redondo 2012). A very general chronological trend can be observed for the studies to move away from strongly endogenous culturalist explanations towards approaches that give more weight to exogenous factors such as social exclusion, discrimination and deficient education policies. The change is at least partially the consequence of NGO's and Roma associations becoming important agents in the Roma studies field as well as the climate of political correctness brought to the Roma studies by the national and European social policy agendas. It can also be partially seen as a consequence of the accelerated process of acculturation and integration experienced by the Roma, a process that has steadily eroded some of the endogenous factors that, in the eyes of many scholars writing just some decades ago, seemed to be paramount for the explanation of the "educational anomaly". This passage has also taken place in the midst of an incensed international debate initiated by some Roma academics like Hancock (2010) and led in Spain by some of the Gitano intelligentsia linked to Roma associations and by the Centro de Estudios Gitanos integrated in the CREA (Community of Research on Excellence for All), a leading research group from the University of Barcelona that counts among its members several Roma scholars (Macías-Aranda and Redondo 2012). This current has a strong exogenous stance and has accused studies leaning towards the endogenous factors of being fraught with an ethnocentric and essentialist bias. In Spain, the critics have been particularly harsh towards the work of Fernández Enguita (1999) which has been overtly labelled as "anti-Gitano" (Macías-Aranda and Redondo 2012: 79).
In the following pages we describe the main factors in the four sets of categories: 1. Endogenous-structural factors (that originate in the specific forms of social organization of the Roma population and their functional/class articulation with the more encompassing social system/political economy) A kinship organization in highly ethnically endogamic extended families which function as a selfsegregating mechanism in all social dimensions, including that of education (Vallés 2017 (Gamella 2007;Piemontese 2017).
All this factors point out at presenting the Roma as a group that constitutes a society within the society, one with clear-cut boundaries but, more importantly, characterized by the survival of pre-modern structural traits (kinship-basedvs. individual-based relations, informal vs. formal economic bonds, intellectual and technological backwardness, gender inequality). From this point of view, some have argued that comparing the Roma to other ethnic minorities in European societies is misleading and that their unique socio-cultural characteristics made them much more similar, instead, to the indigenous peoples in postcolonial states (Banach 2001;Botonogu 2007;Cesarino and França 2017;Andrade Martínez 2018). Some of those postcolonial states, like Colombia and Brazil, have, as a matter of fact granted to their Roma populations a legal status quite similar to that of their Native American ones (Colombian Government 2010; Salloum and de Vasconcellos Figueira 2019). Although not very often explicitly acknowledged in the Roma studies field, the Roma have, as a matter of fact, been considered by many as just another "indigenous people" and the classic theory of modernization, the one that makes formal school education and modernity two synonym concepts and sees traditional pre-modern sociocultural characteristics as obstacles to educational attainment (Harbison and Myers 1964;Anderson and Bowman 1965;Aran et al. 1972;Rival 2000;Paprock 2006), has almost inadvertently sneaked into the explanatory frames. In the light of this theory the premodern Roma sociocultural structures are seen at odds with the formal education systems of modern states in very similar ways as has been described for indigenous peoples around the world (Thompson 1978;Eversole 2005;McEwan 2004;Marie et al. 2008;Gordon and White 2014;Bosco 2017) even leading to some comparative studies between indigenous and Roma students educational attainment (Levinson and Hooley 2014). This cluster of factors can also be related to the so called Theory of Cultural Discontinuity, which has been applied to explain the underachievement of Native American students (Ogbu 1982;Garrett 1995;Tyler et al. 2008).
2. Endogenous-cultural factors (related to an idiosyncratic Roma habitus stemming from their secular history as an "outsider" group in Western societies) Among them, the following ones has been mentioned as explanatory factors of the Roma educational "anomaly": The very low cultural capital of Roma families (Levinson 2007;Fernández Enguita 1996); a sort of historical ethos that rejects the execution of highly abstract cognitive tasks of the kind demanded by the last years of compulsory and post-compulsory education curriculum (San Román 1984;Fernández Enguita 1999); an undervaluation of school and a sort of skeptical attitude towards its declared function as mechanism of upwards social mobility (Gkofa 2017a, b;Powell 2016;Fernández Enguita 1999;San Román 1984) that leads to high rates of absenteeism and the disengagement of families from the educational process (Martínez and Alfageme 2004); the anomie caused by the disadjustment between a new set of personal expectations and goals in life instilled in the Roma youth by acculturation processes and the structural possibilities (and perceived probabilities) of achieving them from their marginal position within the social structure (Bereményi and Carrasco 2015;Soriano et al. 2011); the cognitive dissonance between the individualistic cultural goals of mainstream society acquired at school and those of the collectivist traditional Roma families; the absence of role models of school success (Gypsy teachers, intellectuals and professionals) within the Roma community itself (Fernández Enguita 1999;Vargas and Gómez Alonso 2003;Álvárez-Roldán et al. 2018).
Studies that focus on this approach tend to see the Roma educational situation in a similar way as that of other disenfranchised ethnic minorities and/or particularly vulnerable segments of the working class in Western societies. The reference model here is the African American subproletariat in the U.S. and two theoretical frameworks which were developed to explain their school underachievement. The first one is the Theory of Cultural Deprivation (Bernstein 1961;Taba 1964;Hunt 1966;Chopra 1969). According to it, most Roma families have severe cultural deficits that limit their capacity to support their children's school careers: illiteracy and functional illiteracy, insufficient language proficiency (although Spanish Gitano have lost the original Romani language, they speak a sort of substandard Spanish and they are completely disadvantaged in those regions where a second regional language is taught at school) and insufficient command on the varied array of skills required of them by the education system (knowledge of the bureaucratic organization and rules of the school system itself, social and cultural skills to interact with the rest of the school community, pedagogical resources, etc.). The second one is the Theory of Oppositional Culture (Ogbu and Simons 1998;Ogbu 2008;Downey 2008), an application to African Americans (and other highly marginalized "involuntary" ethnic minorities, which Ogbu and Simons oppose to the "voluntary" minorities of migrant origin with higher expectations of success) of the Theory of the Culture of Resistance developed by Willis in his ethnography on an English working-class school (Willis 1977). In the light of this theory, the rejection and skepticism towards education shown by many Roma (another "involuntary" minority) is seen as an alternative set of cultural values stemming from their perception of mainstream society as a hostile world in which their chances of success are very limited. The values, goals, role models and expectations transmitted by the school cause anxiety and low self esteem because they are perceived as unattainable. The underrating or utter rejection of academic goals step in as a coping mechanism to avoid anomie and psychological harm. The school is seen, then, as a prison where society forces the Roma to stay and absenteeism and underachievement are oppositional mechanisms of resistance that become a subset of cultural practices in themselves, even a habitus in Bourdieu's sense (Bourdieu et al. 1977). In conclusion, what at the end of the day is discussed here is how the values, attitudes, expectations and practices of Roma vis-à-vis education are part of a more encompassing underclass culture, in the sense identified by the Culture of Poverty Theory (Lewis 1963;Leacock 1971). A culture of poverty that becomes a self fulfilling prophecy keeping most Roma trapped in the state of destitution and privation consistently described by all the studies (CIS 2007;FOESSA 2014FOESSA , 2019, in spite of the increasing battery of public policies addressed to them. 4 Not all the diffidence of Roma towards school comes from values related to poverty, though. According to some explanations, the Roma strong sense of cultural identity also plays a part (Fernández Enguita 1999): the school would be seen by some Roma as an acculturating institution, triggering a fear of social and cultural assimilation very much alike to the fear of "whitening" described in Afro American students (Pruitt 2004;Picower 2009).
3. Exogenous-structural factors (derived from the institutional architecture of the national welfare and education system) The weakness and ineffectiveness of affirmative action policies at all levels of administration (FSG 2010(FSG , 2019Salinas Català 2005); the limited lobbying and political effectiveness Roma organizations have displayed so far vis-à-vis the education governance bodies (Ministries, regional governments, etc.) (Jiménez González 2012); the highly spatial segregation or Roma population and their subsequent concentration in so-called "ghetto schools" as a consequence of enrolment schemes based on school zoning (Santiago and Maya 2012;Parra et al. 2017); the deficient facilities and equipment in those "ghetto schools" attended by the majority of the Roma population (FSG 2013;Brüggemann 2015); the inexistence of educational programmes or strategies aimed at the nomadic or highly mobile Roma population (Piemontese 2017); teachers' lack of training in intercultural pedagogical tools (Luciak and Liegl 2009;Olivencia 2012;Herrero et al. 2017); the system's failure in preventing the chronic absenteeism of Roma students (Bereményi 2007a, b;Toro et al. 2017); the government strategy to tackle the Roma educational problem by putting most of them in a simplified remedial curriculum (Aulas de Educación Compensatoria, Remedial Education Classes) instead of increasing the quality of education (Díaz-Aguado Jalón 1999;Iguacel 2009;Santiago and Maya 2012); the cultural alienation provoked by an assimilationist school model devoid of Roma cultural references and the inexistence of specific affirmative action programmes for the training and hiring of Roma teachers (Smith 1997;Casa-Nova 2010;Calvo Buezas 1989Vargas and Gómez Alonso 2003;Moreno 2006;Liegeois 2004;Parra et al. 2017).
4. Exogenous-cultural factors (derived from the ethnically-based system of prestige, the set of ethnic stereotypes and prejudices and the dynamics of interethnic relations in the mainstream society) All recent surveys on racism, xenophobia and ethnic prejudices have shown Roma are one of the most negatively stigmatized and discriminated against ethnic minorities in Spain (Cea D'Ancona and Valles Martínez 2018;Colectivo Ioe 2003;Calvo Buezas 1995, 1990b. Laparra Navarro (2008) calls them "the most segregated ethnic group in Spanish society". The effects of this die-hard racism would be projected onto the education system and practices in similar ways as the described for other discriminated groups, such as Black students in the Anglo Saxon countries (Kohli et al. 2017;Smith 2005;Graham and Robinson 2004;Gillborn and Kirton 2000;Troyna 1984): the prejudices and stereotypes about Roma that still persist among teachers (Vargas and Gómez Alonso 2003;Peček et al. 2014;Trentin et al. 2006;Liegeois 1998Liegeois , 2004, students (Calvo Buezas 1990a), the curriculum itself (Jiménez González 2018;Calvo Buezas 1989) and society in general (McGarry 2014;Calvo Buezas 2012;Saul and Tebbutt 2005) and their consequences (Golem, self-fulfilling prophecy or social-mirroring effects, among others) (Calvo Buezas 2012;Calvo Buezas 1990a, b); the even more serious effects of the most explicit racism (Fekete 2014;Hancock 2010;Flecha and Oliver 2004;San Román 1986).

Emerging Approaches and Research Gaps in Roma Education Studies
Starting in the 1970s most of the literature on Roma and education has focused on the negative aspects of the phenomenon. Some authors (Abajo and Carrasco 2004;Jiménez González 2012;Macías-Aranda and Redondo 2012) explain this as the consequence of a biased approach to the matter underpinned by an implicitly common theoretical-epistemological position of a culturalist nature. Despite the already discussed variety of frameworks, critical authors, some of them Roma themselves, contend that many of the studies implicitly see educational underachievement and Roma ethnicity as one and the same thing. As some author has put it: "Compulsory education till the age of 16 doesn't match the cultural patterns of the Roma population" (Martínez and Alfageme 2004: 307). Jiménez González (2012), a Gitano sociologist, points out that one of the problems that contributes to the social exclusion of Roma in Spain is their invisibility at different levels and that to the other "invisibilities" (territorial, statistical) we should add one in the educational field: that of the Roma with average or outstanding academic performance. Focusing the lens on the deficiencies of Roma education, the argument follows, contributes to generate a partially misleading explanation of the phenomenon and to distort the perceived image of the Roma community itself, making it more difficult to single out and analyze the role and weight of any existing positive dimensions (Macías-Aranda and Redondo 2012). The most recent research by Álvarez-Roldán et al. (2018) could be brought up as an illustration of this. The ethnographic study is conducted in a school where the already mentioned "anomaly" is, in fact, the only reality: a school where the dropout rate of Gitano students is 100%. Instead of pointing out at the exceptionality of this extreme situation (exceptional even for the Roma) they use this exceptional environment as a laboratory to explore and explain the causes of the whole ethnic group educational underachievement.
In order to mitigate this bias, a different approach is emerging in Spain (Aubert and Valls 2003;Abajo and Carrasco 2004;Flecha et al. 2009;Gamella 2011;Jiménez González 2012;Castilla Vázquez 2012;Flecha and Soler 2013;Arranz and González 2014;Girbés-Peco et al. 2015;Amador 2016;Longás et al. 2016;Macías-Aranda et al. 2019) and other European countries (Derrington and Kendall 2007;Dezső 2013;Brüggemann 2014;Gkofa 2017a;Goga 2019), one that focuses on the study of Roma students with successful education trajectories, aiming at identifying the assets and strengths that may exist within the ethnic collective, rather than their shortcomings and weaknesses, and trying to find patterns of success in individuals or families. The point of departure of this approach is a two-fold one. On the one hand a rejection of the cultural essentialism perspective, on the other hand, a double and bidirectional hypothesis about the role of the cultural factor in the students' educational attainment: first of all, the assumption that its weight as a factor leading to failure is not as important as presumed by the previous literature but, more importantly, that it does not always have to be a hurdle in the school performance of ethnic minorities. This approach is trying to apply to the Roma the findings of a whole set of studies on education and ethnic minorities conducted in the United States (Fordham and Ogbu 1986;Portes and Zhou 1993;Ford and Harris 1997;Portes and Rumbaut 2001;Rhamie 2007;Byfield 2008). This line of research found that ethnic factors may operate in a positive way, not as obstacles, but as drivers or boosters of educational attainment. Fordham and Ogbu's (1986) and Ogbu's studies (1995) showed how different groups of Afro Americans and Hispanics with similar cultural traits presented different grades of educational achievement depending on the particular characteristics of the social context in which they were inserted. This put the blame on the education system itself, not the ethnic culture. This hypothesis has been tested by several studies on the Roma: At a European level, the FRA survey showed great differences in the achievement of Roma populations across the continent (FRA 2014). 5 Levinson compared the educational situation of students from two Roma communities in Southwest England. In the first community, the school attainment was similar to that of non-Roma pupils. Dezső (2013) reported the positive results achieved by a high quality educational project aimed at the Roma population, the Gandhi School, in Hungary. In Spain the studies have been led by the CREA within the 5 years macro research "Includ-ed" project funded by the European Commission (Flecha et al. 2009). Several works have shown how the quality of Roma education can be successfully improved with the adequate pedagogical programmes, without apparent cultural loss. They have also detected the emergence of cultural narratives and discourses among the leaders of the Roma associative movement advocating education as a tool for community social development and a path to achieve greater and better opportunities for the future (Macías-Aranda et al. 2019;Girbés-Peco et al. 2015;Flecha and Soler 2013). Coming back to the United States, collaborators (1993, 2001) observed that, across all the studied minority groups, those students who maintained a stronger ethnic identity and had only selectively acculturated performed better that those who had lost their identity (dissonant acculturation) and how this was especially the case in highly discriminated groups. Derrington and Kendall (2007) and Gkofa (2017b) have tested Portes' theory of selective acculturation on successful Roma students, arriving at the same conclusion. In Spain this line of research is still completely underdeveloped and constitutes the first of the research gaps identified by our literature review.
The other research gaps identified by this review are also related to this emergent field of studies: -There are very few longitudinal studies tracking the evolution of Roma educational achievement along time (Salinas Catalá 2009;), whether at a national or local level, from a quantitative or a qualitative ethnographic approach. Virtually all studies offer a still picture of the phenomenon, comparing the Roma situation with that of other ethnic groups at a given moment in time. Studying it, instead, from a chronological point of view, comparing the present Roma educational situation with that of the Roma of previous decades would help to see the "anomaly" in other perspective, as a situation that is capable of improvement. -There is a total absence of studies on the academic achievement of Roma students from the upper and middle classes, which could amount to a 10% of the total Roma population (FSG 2019). The FSG already warned (FSG 2018a, b) that research on the more affluent Roma is methodologically complicated, since they are spatially dispersed and constitute small minorities in areas (and schools) of a majority non-Roma population. For that reason, all studies and surveys, including those on education, have been conducted in places with a high concentration of Roma population, where the poor and marginal Roma live. The study of the Roma upper and middle class population could shed light on the role of selective acculturation in educational achievement. -There are very few studies about Roma students in post-compulsory education.
The review has not found a single one on vocational schools and only three on Roma students in higher education 6 (Gamella 2011;Carmona et al. 2017;Vargas del Amo 2018). The FSG has estimated their number in between 2 and 3% of the total Roma population (FSG 2013, 2019) but there is no way to calculate the margin of error of that estimate. The studies are also quite scant at the European level (Kende 2007;Pantea 2015;Garaz and Torotcoi 2017;Gkofa 2018) and non existent in the Americas. -There is not a single study on Roma professionals: quantitative figures, ethnographic studies, nothing.

Conclusions: Closing the Research Gaps on Roma Education as a Way of Improving Roma Social Inclusion
The literature review has shown there is a historical lack of statistics and periodical nationwide quantitative studies on Roma education in Spain that may lead to an inaccurate assessment of the educational situation. This problem is worsened by the fact that most of the studies are excessively focused on the negative aspects of the phenomenon. This has the effect of introducing a bias that magnifies the so called Roma educational "anomaly", making it appear not only bigger than it probably is but also as an almost insoluble problem, as it tends to see it in essentialist terms, as tantamount to being Roma rather than as a phenomenon which can be separated from the ethnic ascription and subjected to change (and improvement) throughout time.
We agree with the most recent critical currents that this bias may be acting, even though in an unacknowledged, unintended way, as a tool that reinforces the negative stereotypes and the cultural stigmatization that the majority society projects on the Roma and, in that sense, many of the studies might end up becoming, paradoxically, a contributing factor, an obstacle, to the educational attainment of Roma students, one more piece of the self-fulfilled prophecy mechanism that also works in other dimensions.
To counteract the deleterious effects of the data "invisibility" and the essentialist bias, the emergent approaches that focus on the positive dimensions of Roma education must be encouraged and the identified research gaps must be filled because the information obtained could be useful for the design of more effective inclusion policies. A chronological approach, for instance, would show the huge educational improvement undergone by the Gitano population in the last 40 years. The illiteracy rate in 1978 was 58% among the 10-24 age group (FSG 1990), but only 1% in 2014 (FRA 2014). If 27% of Gitanos have completed primary and only 1% compulsory secondary education in 1978 (FSG 1990) they were 98.6% and 40%, respectively in 2012 (MSSI 2012;FSG 2013). This positive trend is likely to continue in the future and we will need periodical studies to track it and divulge it to the Spanish population. A future conduction of studies on the top 10% of upper and middle class Gitano families will probably also provide a much more positive picture of the binomial Roma-education. And finally, more research is needed on the Roma enrolled in vocational schools and universities, as well as on Roma professionals. Although still an inaccurately measured phenomenon, there is no doubt that Roma students in Spanish universities are no longer anecdotal as they used to be (or as they still are in other countries) and show a growing trend in numbers since at least the beginning of the 2010s. It is essential to render the presence of Roma in the university more visible through further research on the subject. An ongoing study conducted by ourselves with forty Roma undergraduates from the five Valencian public universities is currently attempting to do just so: we are testing the selective acculturation theory, since all of them show strong Gitano identity, and we are looking for common characteristics in order to determine whether they are just exceptional individuals who have succeeded in spite of some ethnic and cultural barriers or the product of structural drivers of some sort, stemming from the ethnic group itself or from more encompassing social dynamics.