The educational integration of second generation southern Italian migrants to the north

Abstract

After WW2 Italy experienced a huge internal migration from the south to the northern Italian regions. More than two million individuals moved up north, and the majority of them settled down permanently. How were southern internal migrants integrated into northern Italian society? This work studies the assimilation of this migration flux from a long-term perspective, comparing the school outcomes of the children of southern migrants to those of both northerners' children and children of southern families who did not move. To this aim, logit models of three different school transitions are applied to data from the Italian Longitudinal Household Survey, a retrospective panel survey that includes detailed life-course information on a representative sample of roughly 11,000 Italians. There is no difference between the educational performance of both generation 2 and the mix generation and that of the northerners. However, strong and significant disadvantages were found with regard to generation 1.5, due to the disruption in individual school experience caused by the migration itself.

Headnote

Abstract

BACKGROUND

After WW2 Italy experienced a huge internal migration from the south to the northern Italian regions. More than two million individuals moved up north, and the majority of them settled down permanently. How were southern internal migrants integrated into northern Italian society? Despite the theoretical and substantial relevance of the topic, there has been little systematic research on it.

OBJECTIVE

This work studies the assimilation of this migration flux from a long-term perspective, comparing the school outcomes of the children of southern migrants to those of both northerners' children and children of southern families who did not move.

METHODS

To this aim, logit models of three different school transitions are applied to data from the Italian Longitudinal Household Survey (ILHS), a retrospective panel survey that includes detailed life-course information on a representative sample of roughly 11,000 Italians.

RESULTS

There is no difference between the educational performance of both generation 2 and the mix generation and that of the northerners. However, strong and significant disadvantages were found with regard to generation 1.5, due to the disruption in individual school experience caused by the migration itself.

CONCLUSION

The Italian educational system played an important role in facilitating the integration of the second generation of Southern immigrants, but it was less able to assimilate those who had already begun their studies in the south before following their parents to the north.

1. Introduction

Internal migration can be defined as the change over time in the distribution of the population within a country. Since the early works on migration (Ravenstein 1885), scholars have treated it as a complex phenomenon involving sociological, demographic, and economic aspects (Etzo 2008). Classical research on social stratification took this kind of geographical mobility into consideration, especially the movements from rural to urban areas and their impact on social mobility (Lipset and Bendix 1959; Blau and Duncan 1967). However, in recent decades the international debate has mostly focused on international migrations, because of their greater visibility and political importance from the point of view of Western scholars. Mass internal migrations still occur in developing countries, but they attract less attention from scholars living in the rich countries of the West.

There are still many good reasons to study internal migrants and their integration into the host society. From a theoretical point of view, internal migrations may help achieve better understanding of the concepts used to study migration in general. Currently, mass geographical movements of people are defined as migrations when they imply a change of residence from one country to another: this may be an instance of the 'methodological nationalism' affecting much contemporary social research (Wimmer and Schiller 2003; King and Skeldon 2010).

As regards empirical research, when internal migrations are studied, migrants and natives are typically found in the same dataset with fully comparable information. This is not the case for international migrants, in particular those born abroad (first generation) (Bonifazi 2008). Something similar can be said in regard to the comparison between movers and stayers: most of the recent work on the occupational and social careers of migrants is unable make a judgement as to whether migration was a good choice because the data does not allow comparison between the outcomes of migrants and of their peers who remain in the home country. While the need to conduct surveys in both origin and destination countries has been acknowledged since the late 1970s (Zachariah et al. 1980), the first ethno-surveys to make this comparison possible were only carried out in the following decades (Massey 1987).

All of these considerations apply in the case of the Italian grande migrazione, the huge internal migration from the south to central and northern Italian regions which took place in the aftermath of the post-WW2 reconstruction, from the late 1950s to the mid-1970s.3 During that period some two million southerners moved to the north, and the majority of them settled there permanently; a population movement comparable to moving a city bigger than Naples from the south to the north in the span of two decades. A different population had to be integrated into northern Italian society, but despite the substantive importance of the question of the long-run integration of southern migrants, which was well analysed and discussed during the 1960s and 1970s, there has been little systematic research on it and the available results are inconsistent (see Panichella 2014, pp.17-45 for a review of the literature on the topic).

This study contributes to filling this gap by studying the integration of the second generation of internal migrants. It asks if being a child of southern migrants is a penalty in school achievement. As highlighted by current research on international migrations, by the second generation school achievement is one of the key indicators of an ethnic group's integration into a given host society, both because of the importance of education per se and because it is one of the main determinants of life opportunities (Portes, Fernandez-Kelly, and Haller 2009; Heath, Rothon, and Kilpi 2008: 212).

This study contributes to the literature in several ways. First, most of the current studies on internal migration focus largely on the social integration of the first generation (see, for instance, Borjas, Bronars, and Trejo 1992; Greenwood 2007), while there is less evidence on the integration of the second generation of internal migrants. Second, unlike most work on this topic (as reviewed by Heath, Rothon, and Kilpi 2008), this study compares the educational performance of the second generation with both 'natives'4 (born in the north from northern parents) and 'stayers' (southerners who have not emigrated to the north). Third, to our knowledge, this study is the first to analyse systematically, with representative longitudinal data, the outcomes of the offspring of migrants from the south to the north of Italy. Finally, from a policy perspective, studying past internal migrations in Italy may provide information relevant to the management of current international migrants, because their patterns of integration into Italian society often closely resemble those of previous internal migrants ("ecological succession": Aldrich et al. 1985). Thus, despite major differences with respect to the current international migration wave, the case of the great migration from southern to northern Italy could provide interesting ideas and potential scenarios concerning the integration of current international migrants.

The paper is structured as follows. The next section summarizes the socio-economic background of the grande migrazione, the third section presents the research questions, and the fourth the analytic strategy of the study. The fifth section presents the data, variables, and statistical models, and the sixth the empirical results. The final section sums up the results, discusses them, and draws some conclusions.

2. The great Italian migration

2.1 Italy: one state, two countries?

The modern Italian state was born in 1861, as the armies of Piedmont, a kingdom in the north-west of present-day Italy, were able to conquer most of the country. Since the Middle Ages, Italy had been divided into a number of states with different political systems and socio-economic structures, and their unification into a single nation-state was by no means a simple process. The main cleavage was the one between north and south, generally called la questione meridionale (the southern issue) in Italian public discussion and politics. Since the immediate aftermath of unification, when notable resistance to the new rulers was expressed by segments of southern society, the geographical cleavage has often been seen as the main problem of the Italian national state (Gramsci 1995 [1926]).

Not surprisingly, there has been much debate on both the historical reasons for and the pattern over time of the north-south cleavage, and it is far from reaching a conclusion (Trigilia 2012; Barbagallo 2013; Felice 2013). Indeed, according to recent estimates the economic gap between the two areas was relatively small at the time of unification, and increased over time (Daniele and Malanima 2011). Figure 1 shows a widening gap over time, in particular during the first half of the 20th century. This pattern suggests that unification to some extent damaged the southern economy.5

Less attention has been paid to the gap between the two areas regarding education. Recent research in economic history is of help. It shows that the gap between the two parts of the country was much greater in regard to literacy and school attendance than it was in economic indicators. In 1871, when the available time series begin, the literacy rate was around 41% in the north and around 16% in the south (Felice 2007; see also Barbagli 1982). It took several decades for the south to catch up with the north in terms of primary and secondary school attendance. The new state immediately created a unified and centralised school system, inspired by the French model, but until 1911 the financing was left to the municipalities. The result was much lower funding for southern schools where there were fewer resources. Thus it was only between the two world wars that a convergence process took place in primary and secondary education participation (Felice 2007; 2013).

When the post-war reconstruction gave way to the economic boom that put Italy among the ranks of the industrial countries, the cleavage between the two parts of the country was still massive. The boom mostly involved the northern regions, while the south remained predominantly rural, poor, and traditional. In 1951 the labour force participation rate was 46% in the north and 37% in the south, while the shares of employment in manufacturing were 35% and 23% respectively (Daniele and Malanima 2011). With respect to educational infrastructure the differences between north and south were still huge during the 1950s (Perri 1971; Rossi-Doria 2008). For instance, the pupil/teacher ratio in 1950-1951 was 25:1 in the centre-north but 45:1 in the South. When, in 1959, a national plan for schools (Piano per lo sviluppo della scuola) was drafted, the percentage of classrooms to be built in order to meet total demand was much higher in the south: 65% of primary school classrooms had to be built in the south compared to 34% in the north, and 56% of lower-secondary classrooms versus 43% in the north (Salvemini 1962). In regard to school participation, according to the 1951 census the percentage of the population with at least a primary certificate (5 years) was 71% in the north and 51% in the south, while for lower-secondary school (8 years) the percentages were 11.8% and 7.4%, respectively.

2.2 The great internal migration

The rapid development of northern manufacturing combined with the socio-economic cleavage profoundly affected migration dynamics. Migration from the south to the north characterised Italian society from the early 20th century, and probably even before unification (1861) (Treves 1976). However, at first it concerned mostly educated people living in urban areas and from relatively high social backgrounds, while from the 1950s to the first half of the 1970s there was a mass migration of low-educated people from rural areas in southern Italy to the industrial cities of the north (Compagna 1959; Panichella 2012; 2014). In that period southern Italians also emigrated to other European countries such as West Germany, Switzerland, and Belgium. Such flows are easily explained in terms of the rural-to-urban model (Harris and Todaro 1970): the labour demand of manufacturing plants located in the northern Italian regions and northern European countries triggered the migration of people from the rural areas of the south where there was an excess of labour supply in the agricultural sector (Etzo 2011). The internal migration of the 1950s-1970s was the largest population movement in modern Italy; and perhaps in modern Europe, if the forced migrations that followed WW1 and, in particular, WW2 are excluded. According to our estimation based on official register data, between the early 1950s and the mid-1970s at least 2 million southerners moved permanently to the north looking for work, thus contributing to the modernization of the country.6 The pattern of the internal migration flow over time is charted in Figure 2.

To interpret the graph correctly it should be noted that in the 1930s the Fascist regime had imposed severe administrative constraints on population movements by requiring citizens to obtain authorisation to change residence. Thus the first peak in the graph depends on a number of irregular situations that were regularized by changes in the law in the late 1950s, often many years after the actual migration took place. The actual peak of the flow was reached in 1970. Thereafter the population movement declined, coming almost to a halt in the 1980s, and then resumed in the mid-1990s, albeit at much lower levels. Today Italy is mostly a country of immigration, but the current resumption of south-north migration has renewed interest in its study after two decades of almost complete neglect (Bubbico, Morlicchio, and Rebeggiani 2011; Panichella 2014).

Between the 1950s and the 1970s the south-north flow had most of the features currently associated with international migration: a one-way movement from a relatively underdeveloped to a relatively developed area; people leaving from the most backward zones of the sending countries and entering the lower occupational strata of the receiving countries; and (some) conservation of traditional customs and values (Reyneri 1979; Ginsborg 2003; Panichella 2014). Indeed, southerners who moved to the north of Italy were called 'immigrants' by northerners, even though they remained in their own country (Gabaccia 2000).

However, southern migrants were citizens of the same state: they spoke the same language, had voting rights, and could get jobs in the public administration (Foot 2001). Their children could attend free public schools.7 These circumstances favoured the stabilization of migrants and reunion with their families, so that the majority (about 66%) of the southerners who moved to the north did so permanently (Panichella 2014). They married and had children, rapidly producing a second generation of immigrants in the northern regions (Guetto and Panichella 2013).

The integration of the first generation was by no means easy. The early literature is full of descriptions of the hard times experienced by the first migrants, who faced relentless social discrimination (Fofi 1975). In most cases Italians from the south were, and still are, easily distinguishable from northerners because of their physical appearance and accent. Landlords could thus reject them as tenants. Nicknames with pejorative connotations spread (the one most frequently used was "terroni"), and political parties demanding a reduction in the influx of southern migrants stood in local elections, sometimes gaining a sizeable number of votes. Despite Italian being the national language, those from less educated families only spoke their local dialect (Fofi 1975). Nor was integration in the school system easy. In northern Italy, schools could not cope with such a high number of southerners and the migration caused psychological distress, with negative consequences for school attainment, especially for those born in the south who migrated during childhood, following their parents (Fofi 1975). Moreover, children of southern migrants did not receive much help with schoolwork from their families because their mothers had to work to supplement the family income, and they could not rely on familiar or other social networks for contact with more educated adults (Coleman 1988; Panichella 2014).

We know that migrants from the south generally experienced downward social mobility and entered the northern labour markets in the lowest occupational strata (Panichella 2014). But what happened in the longer run, with the second generation? The answer is not straightforward, because in the 1980s, when large international migration flows began to move towards Italy (Pugliese 2002; Colombo and Sciortino 2004; Ambrosini 2011), the interest in internal migration started to decrease. 8 Hence, whilst the (difficult) inclusion in northern society of the first generation of southern migrants has been the subject of several studies, less attention has been paid to the integration of their children.

Empirical work on the issue is relatively scant, and what does exist shows inconsistent results. On the one hand, a number of local studies confirm that a southern origin is still a source of disadvantage, especially in terms of educational opportunities (Ceravolo, Eve, and Meraviglia 2001; Badino 2012). On the other hand, Magatti (1998), in a general discussion of current social integration patterns in northern Italian society, states that southerners were gradually integrated without long-lasting cleavages due to their origins. The only study, to our knowledge, that uses national representative sample data found no migration penalty in the educational achievement of the offspring of southern migrants schooled in the north. However, a penalty existed for the so-called generation 1.5, those who started school in the south and then moved north with their parents, thereby disrupting their school careers (Impicciatore and Dalla Zuanna 2006).

This work studies the topic with a different dataset, the Italian Longitudinal Household Survey (ILHS). While our sample size is smaller than the one used by Impicciatore and Dalla Zuanna (Istat Multipurpose Investigation - Family, social subjects and conditions during infancy), it includes more detailed information, allowing us not only to study educational performance in terms of achievement but also to observe enrolment, dropping-out, and the choice of upper secondary education track.

We now move to our own research questions and to the analytic design that will enable us to exploit the detailed ILHS life-history data and contribute to this long-standing discussion.

3. Research questions

Briefly stated, this study asks whether the disadvantage of the first generation of southern migrants persisted into the second generation. Whilst the social disadvantages of the first generation can to a large extent be explained by the disruptive features of the migration process itself, the comparison of the outcomes of the second generation and of natives may be seen as the experimentum crucis when studying the long-run integration of migrants (Heath, Rothon, and Kilpi 2008).

The integration of the second generation is of course a complex phenomenon, as it involves the education, careers, and family behaviour of migrants' children. Given the timing of the great migration, the observation of the life course of the second generation of migrants is right-censored because most of its members were born in the 1970s and the 1980s. What we can observe in the ILHS data, collected in 1997 and updated until 2005 (more details below), is their educational career; indeed, school integration is used as a proxy for social integration. This does not seem too strong an assumption, because social stratification and economic research have repeatedly shown that education is one of the key predictors of occupational attainment in contemporary societies (e.g., Shavit and Müller 1998).

In particular, we focus on upper secondary school (high school), an educational level known to be crucial for the structuring of the inequality of educational opportunities (hereafter IEO); that is, the fact that school attainment is affected by ascriptive characteristics such as gender, social class, parental education, ethnicity, etc. (Breen and Jonsson 2000; Lucas 2001). In Italy this is true from the point of view of both vertical and horizontal stratification. Vertically, current recruitment practices of Italian firms consider an upper secondary qualification the minimum educational level that candidates must possess in order to be considered for most jobs, including skilled manual labour (Ballarino and Perotti 2011). Moreover, access to university is impossible without it. Horizontally, Italian upper secondary school is tracked, and despite a 1969 law that gives holders of any type of diploma university access, a strong hierarchy between tracks persists. The classical and scientific academic tracks (liceo classico, liceo scientifico) are the most prestigious and demanding, commonly seen as a direct route to university. The technical track (istituti tecnici) was originally intended to impart knowledge immediately useful in the labour market, but over time has become more academic. Finally, the vocational schools (istituti professionali) provide practical instruction for those wanting to enter the labour market early.

Unlike in other tracked educational systems (e.g., in Germany), in Italy the family of origin plays a crucial role in the choice of upper secondary school track because there is no formal system of teacher recommendation and no specific number of places available in each type of school. Thus students can enrol in whatever type of school they (or their family) prefer, irrespective of their previous school performance (Checchi and Flabbi 2007). Obviously, the three tracks are associated with different probabilities of enrolling at university. Graduates from the upper secondary academic tracks have about 88% probability of enrolling at university, while the probability decreases to about 32% for graduates from the technical track (istituto tecnico) and is as low as 17% for graduates from the vocational track (istituto professionale)9. Moreover, a number of studies show that students with higher social backgrounds are more likely to follow the academic tracks, while those with lower social origins are more likely to attend technical or vocational schools (Panichella and Triventi 2014).

4. Analytical strategy and research hypotheses

The aim of this paper is to study the effect of geographical origin - i.e., being born of northern parents ("northerners"), being born in the north from southern migrants to the north ("migrants"), or being born of southern parents who did not move to the north ("stayers") - on three educational outcomes, as displayed in Figure 3: a) the probability of enrolling in upper secondary school (line b); b) the upper secondary track choice (line c); and c) the attainment of an upper secondary diploma (line d). The third outcome is important because of the high dropout rate in Italian upper secondary schools (Ballarino, Bison, and Schadee 2011).10 The three outcomes refer to different dimensions of the IEO, the first and the third to the vertical and the second to the horizontal. All outcomes are also affected, of course, by social class of origin (lines b2, c2 and d2).

Social class of origin and geographical origin are also related to each other (line a), since migration implies a selection process (Borjas 1994). With respect to the stayers, migrants are favourably selected because they are on average more educated, ambitious, and motivated than their peers who choose to remain in the region of origin (Chiswick 1999). This argument also applies to Italian internal migration, even if during the economic boom the costs of migration were particularly low and the selection process accordingly became less strict (Panichella 2014). On the other hand, despite this positive selection, southern migrants were on average less educated than the northern population, and when they moved up north they were mostly concentrated in lower segments of the labour market.

Because of this selection process the social background of the children of internal migrants is, on average, different from those of both northerners and southerners. Thus one might think that this difference is the only factor accounting for differences in the educational achievement of the second generation of southern migrants compared to both northerners and stayers (Heath, Rothon, and Kilpi 2008). If this is the case, such differences should shrink or even vanish when family background is taken into account (composition hypothesis). On the other hand, being a child of migrants might have an impact per se on educational achievement, independently from other individual features. In this case, one would expect the educational achievement of internal migrants' children to differ from that of both northerners and southern stayers, even when their family background is controlled for (southern migrant effect hypothesis).

This general alternative must be qualified by specifying the pattern of the possible 'southern migrant effect' with respect to the three school outcomes observed (Figure 3). The hypotheses elaborated are summarized in Table 1 according to the educational outcomes (lines a, b, c in Figure 1) and the two terms of comparison: northerners (set of hypotheses 'a') and southern stayers (set of hypotheses 'b').

In regard to enrolment in upper secondary school (line b), the argument of the disruption effect of migration, as reported in section 2 above, would suggest that the second generation of southern migrants is penalized with respect to northerners (H1a). Conversely, they should be advantaged with respect to southern stayers because average school achievement was, and still is, higher in the north than in the south, while the opposite holds for family-background-related school inequality. Indeed, the transition to upper secondary school has been found to make the most difference in school achievement and inequality between the two geographical areas (Ballarino, Panichella, and Triventi 2014). Whilst in the north after WW2 transition rates to upper secondary school increased, with declining class differences, in the south this expansion was slower and class IEO remained substantially stable, producing a divergence over time between the two areas. Hence, the second generation of southern migrants might have benefited from such a divergence, improving their educational opportunities with respect to their peers who remained in the south (H1b).

On the basis of H1a and H1b, a 'selection scenario' can be envisaged concerning both the track choice (Figure 3, line c) and the probability of dropping out (line d). This scenario is based on the differential selection process taking place over school transitions (Mare 1980, 1993). This argument suggests that the population at risk is not the same across transitions because members of the disadvantaged groups are more selected in the early transitions, and that the selection takes place on the individual characteristics favouring achievement.11 In regard to the comparison with the northerners, if the second generation is at a disadvantage in the probability of enrolling in upper secondary education (H1a), then those who enrol might be more selected not only in terms of ability but also where motivation and aspiration are concerned. 12 Since this positive selection compensates for the negative impact of disruption, we hypothesize that the offspring of the migrants that decide to enrol are more likely to choose the academic track than northerners (H2a), and less likely to drop out before completion of secondary school (H3a). Conversely, in regard to the comparison with those who stayed in the south, if daughters and son of southern migrants have a higher probability of enrolling at upper secondary school because the IEO is higher in the south of Italy (H1b), then we would expect them to be less likely to choose the academic track (H2b) and more likely to drop out (H3b) than their peers whose families stayed in the south.

Migration studies often subdivide children of migrants into specific categories based on their age at arrival in the receiving society. Typically, a distinction is made between those who are born in the host country (generation 2), those who arrived at 0-5 years (generation 1.75), at 6-12 years (generation 1.5), and 13-17 years (generation 1.25) (Rumbaut 2004). Such classification is relevant to this work because those of southern parentage who are born in the north of Italy might have had better educational opportunities than those who migrated during childhood. The former did not experience school disruption as they were entirely schooled in the north. Moreover, they are likely to have begun their educational career when their parents had already amortised most of the migration costs and reached residential and economic stability. Conversely, in the case of those children of southerners who followed their parents' move to the north, the social disruption implied by the geographical mobility of the family might have had negative effects on their school careers. Hence, we expect that the educational disadvantage faced by the second generation is higher, in all educational outcomes, for those who migrated during childhood (H4).

5. Data, variables, and methods

5.1 Data and variables

The data used are from the Italian Longitudinal Household Survey (ILHS), the main Italian social mobility survey. ILHS includes five waves. The first (1997) was a retrospective survey of a representative sample of 4,956 Italian households, where all household members more than 18 years old were interviewed. Thereafter, four prospective waves were fielded every second year between 1999 and 2005, collecting new information about previous respondents, any offspring reaching 18 years of age, and all new household members. The unit of analysis selected for the survey is the household, and the reference population is the set of households resident on Italian territory and registered at municipal registry offices at the end of 1996. Interviewed households were selected by means of a two-stage stratified sampling procedure: the 8,104 Italian municipalities were assumed as primary sampling units and assigned to 42 strata defined by two variables, the region and the type of municipality (metropolitan, adjoining, other). Despite its relatively small sample size, ILFI data have been the most important dataset in the studies of social stratification in Italy for at least fifteen years. For more information on the representativeness (even at regional level) of ILHS data and on its sampling procedure, see Schizzerotto (2002) and http://www.soc.unitn.it/.

The ILHS collects detailed information on the entire life history of the interviewees, including social origins, education and vocational training, geographical mobility, career, family, and fertility behaviour. Given our research questions, we restricted our sample to those born between 1950 and 1985, and adopted a 'transition approach' (Mare 1980), further restricting our sample to those respondents who had achieved at least a lower secondary school qualification, ending up with 5,446 cases.

The three dependent variables, as shown in Figure 3 above, were a) probability of enrolling in upper secondary school; b) track choice; and c) probability of achieving a final diploma (diploma di maturità) enabling matriculation to university. The variable concerning the school tracks was operationalized in three categories: a) academic track, b) technical track, and c) vocational track.

The main independent variable, the geographic origin of Italian children, was operationalized into five categories. First, the south-north divide was dichotomized, defining as 'southern' individuals born in seven regions (Abruzzo, Campania, Apulia, Molise, Calabria, Basilicata, Sicily, and Sardinia) and as 'northern' those born in the remaining regions. To these two groups of 'stayers', three types of migrant children were added, subdivided according to their age at arrival in the receiving society. We ended up with five groups: 1) northerners: born in the centre-north, both parents born and living there; 2) gen 2 mix: born in the centre-north, with one native parent and one migrated from the south; 3) gen 2: born in the centre-north or migrated when aged less than 6, with both parents born in the south; 4) gen 1.5: born in the south, migrated to the centre-north with the parents when aged 6-1313; 5) southerners: born in the south, to southern parents, and living there.

In order to control for other relevant background factors and related composition effects biasing the association between school outcome and geographical origin, the models included three control variables. The first was social class of origin, measured by the social class of the main breadwinner in the family according to the Italian version of the EGP class scheme (Ballarino and Cobalti 2003): Bourgeois (EGP I-II), White collar (II-IIIa), Urban petty bourgeois (IVab); Farmer (IVc); Urban working class (IIIb, V-VIIa); and Farm worker (VIIb)14. The other control variables were gender and year-of-birth dummies.

Table 2 shows the descriptive statistics of the analytical sample.

On inspection of the table, it may come as a surprise that Gen 2 and Gen Mix have an advantage in terms of social background, being over-represented in the bourgeois and white collar categories. This is because southerners from 'good' social backgrounds have always migrated to the north, even if this positive selection became less strict during the 1960s and 1970s. Indeed, the crisis of the southern agricultural sector, combined with the spread of Fordist manufacturing in the north, increased the propensity for the lower strata of southern society to migrate (Panichella 2012).

5.2 Methods

Logit models were estimated to study the vertical dimension of IEO. The probability of enrolment was estimated including all graduates from lower secondary education, while the analysis of the probability of not dropping out included all enrolled in upper secondary education. The horizontal dimension (track choice) was studied by means of multinomial logit models, considering only those who enrolled in upper secondary education. The results are reported in terms of average partial effects (APE), which are easy to interpret and make the comparison across models less problematic (Mood 2010). For each analysis we estimated three models: Model 1 only included geographical origin as regressor, Model 2 added social class, and Model 3 added gender and year-of-birth dummies. The relatively limited size of the sample did not allow more complex specifications, including (for instance) an interaction between class of origin and geographical origin, or one between the latter and cohort/year of birth.

A set of robustness checks with different models and specifications was performed. First, because the ILHS dataset does not include individual-level information on parental place of birth, geographical origin was collected by linking the record of each child to the characteristic of his/her parents. This procedure had a limitation because it only enabled detection of the members of generation 2 and generation 2 mix who were still living with at least one parent in 1997. Moreover, if a parent had died or divorced before 1997, it was impossible to detect second-generation mix. In that case, individuals were classified according to the geographical origin of the parent living in the household.

However, based on information collected on parents, it was possible to know how many children had already left the household and were thus excluded from the sample. In our case, we were able to gain information on 61.2% of the members of generation 2, while the rest have been classified as northerners. This is a problem faced by many studies using data from household-sampling surveys that do not collect information on parental geographical origin (Impicciatore and Dalla Zuanna 2006). We addressed it by eliminating those interviewees born before 1950 and controlling for year of birth. Another robustness check, which we took from the latter paper, was to further restrict the analysis to interviewees aged 20-35 in 1997. The results did not change (results available on request from the authors).

As another robustness check, linear probability models (LPMs) were estimated in place of the logit model; the analyses were separately replicated with each of the three macro-areas included in the 'north' category (north-west; north-east; centre). We estimated a set of 'unconditional models', including both all individuals eligible for upper secondary education and the overall sample.15 We also included in our model a set of regional dummies and an interaction term between year of dummies and region of residence, in order to control for composition effects related to the fact that members of the second generation of southern migrants were over-represented in some of the region-by-year combinations. In all cases, the results were almost identical to those presented in what follows. They are not reported for lack of space, but are available from the authors on request.

6. Empirical results

The analysis starts with the first educational outcome, the probability of enrolling in upper secondary school (Table 3)16. According to the three specified models, it is clear that geographical origin has an independent effect on the probability of enrolling at upper secondary school, because its effect does not change when social class of origin (Model 2) and other individual characteristics are controlled for (Model 3). Concerning this transition, therefore, the composition hypothesis can be ruled out.

However, the southern migrant effect appears to be negative only for the case of gen. 1.5, those who migrated from the south aged 6-14: in Model 3, with a full set of controls, they have 15 percentage points' less probability of entering upper secondary school than their northern peers. It may be interesting to add, in order to compare geographical and class penalties, that this difference is similar to that between the petty bourgeoisie and the bourgeoisie (19 percentage points), but is smaller than the difference between urban working class and bourgeoisie (26 percentage points).

Given that they started school in the south and finished it in the north after their families had migrated, we would attribute the disadvantages faced by members of generation 1.5 to school disruption. The negative effects on school achievement of school disruption are generally well known (Astone and McLanahan 1994). Consequently, one could state that the source of disadvantage is school disruption associated with the geographical move from the south to the north, not the latter per se. This is indeed a good argument, and we tested it. All models were re-estimated with a different specification of the independent variable, splitting northerners and southerners into two groups to distinguish between those who had had at least one episode of geographical mobility when aged 6-14 and those who did not. 17 No significant effect was found for the 'moving' categories, meaning that what makes the difference is not school disruption per se but disruption due to parental migration from the south to the north (results available from the authors on request).

However, in contrast to the hypothesis above (H1a), other children of southern migrants fare much better: those born in northern Italy (gen 2) are statistically indistinguishable from northerners, and those who have one southern and one northern parent (gen 2 mix) have even more probability (7 percentage points) of enrolling in upper secondary school than the reference category. On the other hand, comparison with those who stayed in the south supports our migrant advantage hypothesis (H1b) only for gen mix and gen 2, who are more likely to enrol in upper secondary school, while in the case of gen 1.5 (who moved north while children) no statistically significant difference is observed.

We now move to the analysis of the second educational outcome concerning the choice of upper secondary track. Hypotheses 2a and 2b were tested by means of a 'conditional' model on a set of three outcomes, estimated only on those interviewees who entered upper secondary school. The results of this model - which reproduces the 'selection scenario' with a different, smaller, and selected population at risk - are shown in terms of difference in the probability of enrolling in the academic and technical tracks rather than in the vocational track (baseline outcome).

As predicted by our hypotheses, when those who did not access upper secondary school were excluded there was no sign of any kind of 'southern penalty': even the weaker group, those belonging to gen 1.5, were more likely than their northern peers to choose the liceo instead of the vocational tracks. Thus the selection scenario of H2a regards only gen 1.5. Something similar holds for the comparison with the offspring of those interviewees who stayed in the south, who showed a propensity to enrol in an academic track higher than that of both the northerners and gen 2, and similar to that of the mixed generation and gen 1.5. Hence, H2b is confirmed only for generation 2, which has less probability of enrolling in the academic track, but not for gen 1.5 and gen mix.

The advantage of southerners and gen 1.5 is consistent with the 'selection scenario', since these categories experience a stronger selection in the transition from lower to upper secondary school, and those who 'survive' are more selected on ability, effort, or any other feature favouring school achievement than those of the other groups. However, the advantage of generation mix is different. This group, in fact, not only has a greater probability of enrolling in upper secondary education but also an advantage in the probability of entering the academic track. Thus their advantage might depend on different factors such as the availability of local networks via the northern-born parent or other parental characteristics (curiosity, an open mind, etc.) that affect both the probability of choosing a partner with a different geographical origin and the probability of effectively helping children in their school experience.

Finally, we turn to the 'southern migrant effect' for the third educational outcome considered, the likelihood of dropping out upon enrolment in upper secondary school (Table 5). In this case there is no evidence of a southern migrant penalty: even in Model 1, without controls, all the confidence intervals (CIs) include 0, and the addition of the entire set of controls widens the CIs but does not move them. This holds for the comparison with both northerners and with the southern stayers. Thus, both H3a and H3b are not corroborated.

To sum up, once the migrants' disadvantage is eliminated, in this case by differential selection in access to upper secondary school, their school outcomes are fairly similar to those of the northerners. However, this conclusion should be further investigated by interacting geographical origin with the school track attended. Unfortunately, the small sample size prevented control of whether the migrants' penalization in the probability of dropping-out changed across different tracks.

7. Conclusions

This paper has studied the educational integration of the offspring of southern Italians who migrated to the north. The question addressed has been the long-run integration of migrants, and the indicator was whether the second generation of internal migrants had different educational opportunities with respect to both northerners and southern stayers.

In general, our results show that such a difference exists, even when other relevant factors are taken into account. There is a southern migrant effect on school achievement which is independent of composition effects. However, the evidence shows that this effect only takes the form of a penalty in some cases, as it varies according to the circumstances of the family's migration and the school outcome considered.

As regards access to upper secondary school, evidence shows that only members of gen 1.5 who moved to the north as children of school age are disadvantaged compared to northerners, while those who were born in the north from southern parents (gen 2) are not distinguishable from the northerners, and those with just one southern parent (gen 2 mix) are advantaged. Compared with those whose parents remained in the south, children of migrants are at an advantage in the case of gen 2 and gen 2 mix, while the members of gen 1.5 are indistinguishable from those who stayed in the south.

Concerning the further outcomes, i.e., track choice and the completion of secondary school upon enrolment, a 'selection scenario' was hypothesized according to which differential selection may produce different patterns of inequality across transitions, as observed by research on the sociology of education. As expected, the results confirm that the children of southern migrants who moved north aged 6-14 underwent a strong selection at the moment of enrolment in upper secondary school. This makes the surviving children more selected in terms of ability and motivation. It is for this reason that their track choice does not show any penalty with respect to both the northerners and the southern stayers. Conversely, they show a higher propensity than the former group to enrol in both the upper (liceo) and the intermediate tracks (istituto tecnico). The same goes for southern stayers who enrol in upper secondary school, as they undergo a stronger process of selection than their northern peers. Similarly, no effect of southern origin was found, even without controlling for family background and other relevant factors, on the probability of dropping out after enrolment and so not achieving a final diploma and being unable to access university.

In sum, no evidence was found of a generalized 'southern migrant penalty' in school achievement for the offspring of southern Italians who had migrated to the north. On the contrary, our results suggest a remarkable degree of heterogeneity for the second generation, as hypothesized by H4. Gen 1.5, in particular, including those who moved north with their parents when aged 6 to 14, stands out as being disadvantaged in access to upper secondary school, the key school transition in the Italian system. We attribute this disadvantage to school disruption, because gen. 1.5 started school in the south and finished it in the north after their family migrated.

There is a great deal of evidence indirectly supporting our interpretation. As noted above, the earlier literature largely testifies to the difficulties encountered by the children of southern migrants when inserted in northern school classes (Fofi 1975). Current research on school quality in Italy shows that on average it is lower in the south than in the north (Bratti, Checchi, and Filippin 2007), and all available information points to the persistence over time of this gap. Moreover, other social disruption issues beyond the school system have to be considered because they may have had a negative impact on the children's school behaviour and performance. It is possible to speculate that early difficulties related to the change of peers, or to the different accent, or to any other factor related to the change of school and social environment, may have pushed the children of migrants, in the eyes of their teachers, into the category of mediocre students, resulting in demotivation and negative peer effects. As well known, a small early disadvantage can incrementally turn into a substantial gap in a 'Matthew effect' pattern (Merton 1968).

Another possible factor causing school disadvantage for gen 1.5 children is that their schooling in the north took place within a shorter time span after the migration episode than their peers in gen 2. Thus their parents were likely to have been much busier integrating themselves into the host society, in particular from an occupational point of view, which would have left them with little time and few resources to support their children's school performance by helping them with homework, meeting teachers, etc. In a sense, this is also a dimension of the school disruption process related to geographical mobility whereby the child loses the social networks, both familial and extra-familial, that are often helpful for his/her school performance (Coleman 1988).

What about the other groups of children of southern migrants? When controlling for other relevant background factors, those schooled in the north (gen 2) were not statistically distinguishable from their native peers in terms of the probability of enrolling in upper secondary school, while those with a northern parent (gen 2 mix) were advantaged. First, they did not experience school disruption, as they were entirely schooled in the north; second, when they entered school their parents were to some extent settled in the northern host society, giving them more time to care for their children and stronger social networks to rely on.

The case of gen 2 mix is of particular interest. Their advantage over their peers from gen 1.5 may depend on the local networks available via the native parent, or on some unobserved characteristic of the parents correlated with both the probability of choosing a partner from a distant place and the probability of effectively helping children in their school experience. Curiosity and an open mind may be among the relevant unobserved factors. Moreover, because they come from different cultures, mixed parents have a larger array of educational tools to use with their children. Finally, couples in which the partners come from different places and cultures are typically more at risk of dissolution, and those that manage to survive up to the point of having and raising children may be positively selected with respect to traits such as patience and devotion, which are helpful in children's schooling.

Besides comparing the offspring of southern migrants with northerners, our evidence also provides a comparison with the offspring of southern families who did not move. This is a key comparison, for it provides answers to questions such as whether the decision to emigrate was a good one or not. The results show that also in this case, controlling for other relevant factors, the timing of the family migration makes a difference. Those schooled in the north had more chance of enrolling in upper secondary school, in particular in the technical-vocational tracks, than those whose families had stayed in the south, while those whose schooling was disrupted (gen 1.5) had the same outcomes as the stayers. Hence - at least from this point of view - on average, migration was a good decision: the migratory choice of the parents allowed their children to access upper secondary school, whose vocational and technical tracks were expanding in tandem with the great expansion of northern manufacturing during the 'economic boom' of the late 1950s and 1960s. Consequently, they were able to join the ranks of the skilled working class and lower-level technicians (Ballarino, Panichella, and Triventi 2014). Of course, more detailed data on their careers is required to check their outcomes in the long run.

Finally, in regard to the debate on the educational integration of migrants, we conclude that the case of southern Italians in the north testifies to the importance of schooling in the host society. Our results confirm that northern Italian schools were to a large extent able to integrate the children of the southern migrants who were schooled in the receiving regions (Impicciatore and Dalla Zuanna 2006). This achievement was substantially helped by the fact that the second generation of internal migrants did not suffer one of the main difficulties encountered by many current international migrants: the difference in language. However, even speaking the same language was of no use in the case of those southern children who interrupted their school careers to follow the northbound movement of their parents. In their case, northern Italian schools proved unable to effectively manage the shocks brought about by the disruption of their schooling.

There are still some points that deserve further investigation. First, data limitation does not allow the study of the integration of the second generation of internal migrants in different areas of the north, for instance distinguishing the north-west from the north-east and the centre. Second, it would be useful to study the educational outcomes of the second generation looking not only at the upper secondary level but also at access to and the achievement of tertiary education. Finally, as previously mentioned in the text, ILHS data does not include information on parental place of birth; hence it does not allow identification of the members of the second generation and second generation mix that were not not living with at least one parent in 1997. We addressed this problem with a wide range of robustness checks (see section 5.1), but this selection could have affected our estimates because, as one grows older, residing in the home with parents may be associated with particular characteristics. Unfortunately, to date there are no surveys that allow the study of the educational inequalities of internal immigrants' children while avoiding this problem. For this reason it is necessary to have more complete data that allow a direct identification of both generation 2 and generation 2 mix.

8. Acknowledgements

Previous versions of this work were presented at the EducEight Conference, Northumbria University, Newcastle-upon-Tyne, July 2012; at the Espanet Italia Conference, Università "La Sapienza", Rome, September 2012, and in a seminar at the European University Institute (EUI), Florence, January 2013. The authors would like to thank all those who participated, and in particular Albert Arcarons, Fabrizio Bernardi, Hector Cebolla Boado, Daniel Horn, Emmanuele Pavolini, and Loris Vergolini for useful comments.

Footnote

3 For the sake of readability, in the rest of the paper we will use the terms 'south' and 'north', but the latter term should be understood as 'centre-north'. See below, in the data section (5), how the geographical areas are defined and operationalised.

4 Second generation migrants are also natives, but for sake of simplicity we will use the term 'natives' only for those born in the north from northern parents.

5 More recent estimates by Felice (2007; 2009) show a bigger economic gap between northern and southern regions at the time of unification. However, the difference is not relevant to the argument made here.

6 Data are collected by the Italian National Statistical Institute (Istat) (source: Movimento anagrafico della popolazione, 1955-2002). See Panichella (2014) for details.

7 Private education has never been quantitatively relevant in Italy.

8 In fact, economists, sociologists, and demographers continued to analyse the territorial movements between Italian regions, but the main research problem was that migration from the south had almost disappeared, despite the persistence of dramatic socio-economic differences between northern and southern Italy

9 Authors' calculation on ILHS data (see below). See also Ballarino and Panichella (2014). Besides the three main tracks, there are also art schools (liceo artistico, scuola d'arte), teacher training schools (scuola magistrale, istituto magistrale), and language schools (liceo linguistico), whose curricula, labour market value, and position in the track hierarchy have fluctuated notably over time. They are therefore often excluded from analyses (e.g., Cappellari and Lucifora 2009) because they have always hosted a lower proportion of students with respect to the three main tracks. We included them in our typology as a fourth residual category, but did not report the results.

10 To complete the picture, the probability of completing the academic track upon enrolment should also be considered, but the sample does not include enough cases for a reliable estimation of this outcome.

11 Unfortunately, ILHS data do not contain information on ability, nor on final grades and failures at lower-secondary level. Thus we were not able to conduct a strict empirical test of this proposition for our sample.

12 This hypothesis is related to the literature showing the greater motivation and ambition of children of many migrant groups compared to natives (Birnbaum and Cebolla-Boado 2007).

13 If this group is extended to include all those who migrated with their parents before entering school, even when aged 0-6, the results do not change. The numbers did not allow their consideration as a separate category.

14 We refer to the standard EGP, as presented in Breen (2004). The Italian scheme differs with respect to the white collar class, which includes most low-level professionals (EGP II) and the skilled working class (EGP IIIb) (Ballarino, Barone, and Panichella 2014). However, when the standard EGP was used the results concerning the association between family geographical origin and school achievement did not change (results available on request).

15 The model for track choice is reported in Table A1 in the Appendix, while the others are available on request from the authors.

16 For the sake of brevity and readability of the tables, this section only reports the estimates of our main parameter of interest: the association between geographical origin and school outcome. Full results, including estimated coefficients for social class, are available from the authors on request.

17 Geographical mobility is defined as a movement of residence across provinces. There are around 100 Italian provinces, depending on the period of interest, and it is easy to assume that such a movement implies a change of school.

References

References

Aldrich, H.E., Carter, J., Jones, T., Mc Evoy, D., and Valleman, P. (1985). Ethnic Residential Concentration and the Protected Market Hypothesis. Social Forces 63(4): 996-1009. doi:10.1093/sf/63.4.996.

Ambrosini, M. (2011). Sociologia delle migrazioni. Bologna: Il Mulino.

Astone, N.M., and McLanahan, S.S. (1994). Family structure, residential mobility, and school dropout: A research note. Demography 31(4): 575-584. doi:10.2307/ 2061791.

Badino, A. (2012). Strade in salita. Figli e figlie dell'immigrazione meridionale a Torino. Roma: Carocci.

Ballarino, G., Barone, C., and Panichella, N. (2014). Social background and education in occupational attainment in 20th century Italy. European University Institute (Working Papers Series/Report no.: EUI SPS; 2014/03).

Ballarino, G., Bison, I., and Schadee, H. (2011). Abbandoni scolastici e stratificazione sociale nell'Italia contemporanea. Stato e mercato 3: 455-494.

Ballarino, G. and Cobalti, A. (2003). Mobilità sociale. Roma: Carocci.

Ballarino, G. and Panichella, N. (2014). Origini familiari, scuola secondaria e accesso all'università dei diplomati italiani, 1995-2007. Scuola democratica 2: 365-392.

Ballarino, G., Panichella, N., and Triventi, M. (2014). School expansion and uneven modernization. Comparing educational inequality in Northern and Southern Italy. Research in Social Stratification and Mobility 36: 69-86. doi:10.1016/ j.rssm.2014.01.002.

Ballarino, G. and Perotti, L. (2011). Aziende lombarde e diplomati tecnico-professionali: criticità, punti di forza, strumenti. Milan: Chamber of Commerce (research report).

Barbagallo, F. (2013). La questione italiana. Il Nord e il Sud dal 1860 ad oggi. Roma-Bari: Laterza.

Barbagli, M. (1982). Educating for Unemployment: Politics, Labor Markets, and the School System - Italy, 1859-1973. New York: Columbia University Press.

Blau, P. and Duncan, O.D. (1967). The American Occupational Structure. New York: Wiley.

Bonifazi, C. (2008). International migration in Europe: new trends and new methods of analysis. Amsterdam: Amsterdam University Press. doi:10.5117/97890535689 41.

Borjas, G.J. (1994). The economics of immigration. Journal of economic literature 32(4): 1667-1717.

Borjas, G.J., Bronars, S.G., and Trejo, S.J. (1992). Self-selection and internal migration in the United States. Journal of Urban Economics 32(2): 159-185. doi:10.3386/ w4002.

Bratti, M., Checchi, D., and Filippin, A. (2007). Da dove vengono le competenze degli studenti? I divari territoriali nell'indagine OCSE PISA 2003. Bologna: Il Mulino.

Breen, R. (2004). Social mobility in Europe. Oxford: Oxford University Press. doi:10.1093/0199258457.001.0001.

Breen, R. and Jonsson, J. (2000). Analyzing Educational Careers: A Multinomial Transition Model. American Sociological Review 65: 754-72. doi:10.2307/265 7545.

Brinbaum, Y. and Cebolla-Boado, H. (2007). The School Careers of Ethnic Minority Youth in France. Success or Disillusion? Ethnicities 7: 445-474. doi:10.1177/ 1468796807080237.

Bubbico, D., Morlicchio, E., and Rebeggiani, E. (2011). Su e giù per l'Italia. La ripresa delle emigrazioni interne e le trasformazioni del mercato del lavoro. Milano: Franco Angeli.

Cappellari, L. and Lucifora, C. (2009). The "Bologna Process" and college enrollment decisions. Labour Economics 16: 638-647. doi:10.1016/j.labeco.2009.08.009.

Ceravolo, F., Eve, M., and Meraviglia, C. (2001). Migrazioni e integrazione sociale: un percorso a stadi. In: Bianco, M.L. (ed.). L'Italia delle diseguaglianze. Roma: Carocci: 83-116.

Checchi, D. and Flabbi, L. (2007). Intergenerational mobility and schooling decisions in Germany and Italy: The impact of secondary school tracks. IZA Discussion Paper No. 2876.

Chiswick, B.R. (1999). Are immigrants favourably self-selected? The American Economic Review 89(2): 181-185. doi:10.1257/aer.89.2.181.

Coleman, J.S. (1988). Social capital in the creation of human capital. American journal of sociology 94: 95-120. doi:10.1086/228943.

Compagna, F. (1959). I terroni in città. Roma-Bari: Laterza.

Colombo, A. and Sciortino, G. (2004). Gli immigrati in Italia. Bologna: Mulino.

Daniele, V. and Malanima, P. (2011). Il divario Nord-Sud in Italia. 1861-2011. Soveria Mannelli: Rubbettino.

Etzo, I. (2008). Internal migration: A review of the literature. MPRA Paper No.8783.

Etzo, I. (2011). The determinants of the recent interregional migration flows in Italy: a panel data analysis. Journal of Regional Science 51(5): 948-966. doi:10.1111/ j.1467-9787.2011.00730.x.

Felice, E. (2007). I divari regionali in Italia sulla base degli indicatori sociali (1871- 2001). Rivista di politica economica 97: 359-405.

Felice, E. (2009). Estimating regional Gdp in Italy (1871-2001): Sources, methodology and results. Madrid: Universidad Carlos III de Madrid, Dept. de Historia Económica e Instituciones: 09-07 (WP in Economic History).

Felice, E. (2013). Perché il Sud è rimasto indietro. Bologna: Il Mulino.

Fofi, G. (1975). L'immigrazione meridionale a Torino. Milano: Feltrinelli.

Foot, J. (2001). Milan since the miracle: City, culture and identity. Oxford: Berg.

Gabaccia, D.R. (2000). Italy's many diasporas. London: UCL Press.

Ginsborg, P. (2003). A history of contemporary Italy: society and politics, 1943-1988. London: Palgrave Macmillan.

Gramsci, A. (1995, original edition 1926). The Southern question. Toronto: Guernica.

Greenwood, M.J. (1997). Internal migration in developed countries. In: Handbook of population and family economics (1): 647-720. doi:10.1016/S1574-003X(97) 80004-9.

Guetto, R. and Panichella, N. (2013). Geographical mobility and reproductive choices of Italian men. European Sociological Review 29: 302-315. doi:10.1093/esr/ jcr068.

Harris, J.R. and Todaro, M.P. (1970). Migration, unemployment and development: a two-sector analysis. The American Economic Review 60(1): 126-142.

Heath, A.F., Rothon, C., and Kilpi, E. (2008). The Second Generation in Western Europe: Education, Unemployment, and Occupational Attainment. Annual Review of Sociology 34: 211-35. doi:10.1146/annurev.soc.34.040507.134728.

King, R. and Skeldon, R. (2010). 'Mind the gap!' integrating approaches to internal and international migration. Journal of Ethnic and Migration Studies 36(10): 1619- 1646. doi:10.1080/1369183X.2010.489380.

Impicciatore, R. and Dalla Zuanna G. (2006). A difficult social mobility: the education of children of Southern parents emigrated to Central e North Italy. Genus 63: 155-201.

Lipset, S.M. and Bendix, R. (1959). Social Mobility in Industrial Society. Berkeley: Univ. Calif. Press.

Lucas, S.R. (2001). Effectively Maintained Inequality: Educational Transitions and Social Background. American Journal of Sociology 106: 1642-1690. doi:10.1086/321300.

Magatti, M. (1998). Tra disordine e scisma. Le basi sociali della protesta del Nord. Rome: Carocci.

Malanima, P. and Daniele, V. (2007). Il prodotto delle regioni e il divario Nord-Sud in Italia (1861-2004). Rivista di politica economica Marzo-Aprile 2007: 267-315.

Mare, R.D. (1980). Social background and school continuation decisions. Journal of the American Statistical Association 75(370): 295-305. doi:10.1080/01621459. 1980.10477466.

Mare, R.D. (1993). Educational stratification and observed and unobserved components of family background. In: Shavit, Y. and Blossfeld, H.P. (ed.). Persistent Inequality. Boulder: Westview Press: 351-376.

Massey, D.S. (1987). The ethnosurvey in theory and practice. International Migration Review (21): 1498-1522. doi:10.2307/2546522.

Merton, R.K. (1968). The Matthew effect in science. Science 159(3810): 56-63. doi:10.1126/science.159.3810.56.

Mood, C. (2010). Logistic regression: Why we cannot do what we think we can do, and what we can do about it. European Sociological Review 26(1): 67-82. doi:10.1093/esr/jcp006.

Panichella, N. (2012). Le migrazioni interne nel secolo scorso: vecchie e nuove forme a confronto. Stato e mercato 32(2): 255-282.

Panichella, N. (2014). Meridionali al Nord. Migrazioni interne e società italiana dal dopoguerra a oggi. Bologna: Il Mulino.

Panichella, N. and Triventi, M. (2014). Social Inequalities in the Choice of Secondary School. Long-term trends during educational expansion and reforms in Italy. European Societies 16: 666-693. doi:10.1080/14616696.2014.939685.

Perri, P. (1971). Scuola e Mezzogiorno. La scuola e i suoi riflessi sullo sviluppo del Mezzogiorno. Vibo Valentia: Qualecultura Editrice.

Portes, A., Fernandez-Kelly, P., and Haller, W. (2009). The Adaptation of the Immigrant Second Generation in America: A Theoretical Overview and Recent Evidence. Journal of Ethnic and Migration Studies 35: 1077-1104. doi:10.1080/ 13691830903006127.

Pugliese, E. (2002). L'Italia tra migrazioni internazionali e migrazioni interne. Bologna: Mulino.

Ravenstein, E.G. (1885). The laws of migration. Journal of the Statistical Society of London 48(2): 167-235. doi:10.2307/2979181.

Reyneri, E. (1979). La catena migratoria: il ruolo dell'emigrazione nel mercato del lavoro di arrivo e di esodo. Bologna: Il Mulino.

Rossi-Doria, M. (2008). Scuola e Mezzogiorno. Napoli-Roma: L'ancora del mediterraneo [originally published btw 1955 and 1971].

Rumbaut, R.G. (2004). Ages, Life Stages, and Generational Cohorts: Decomposing the Immigant First and Second Generations in the United States. International Migration Review 38:1160-1205. doi:10.1111/j.17477379.2004.tb00232.x.

Salvemini, G. (1962). L'istruzione del Mezzogiorno con particolare riguardo all'istruzione professionale. In: Salvemini, G (ed.). Industria, Servizio, Scuola. Bari: Laterza.

Schizzerotto, A. (2002). Vite ineguali. Disuguaglianze e corsi di vita nell'Italia contemporanea. Bologna: Il Mulino.

Shavit, Y. and Müller, W. (1998). From School to Work. A Comparative Study of Educational Qualifications and Occupational Destinations. Oxford: Clarendon.

Treves, A. (1976). Le migrazioni interne nell'Italia fascista. Torino: Einaudi.

Trigilia, C. (2012). Non c'è Nord senza Sud. Perché la crescita dell'Italia si decide nel Mezzogiorno. Bologna: Il Mulino.

Wimmer, A. and Schiller, N.G. (2003). Methodological Nationalism, the Social Sciences, and the Study of Migration: An Essay in Historical Epistemology. International Migration Review 37(3): 576-610. doi:10.1111/j.1747-7379.2003. tb00151.x.

Zachariah, K.C., Conde, J., and Nair, N.K. (1980). Demographic aspects of migration in West Africa (Vol. 1). World Bank Staff Working Paper No. 414.

Appendix

(ProQuest: Appendix omitted.)