One of the earliest arguments in support of coeducational schooling concerns the suggestion that it provides a more realistic and socially integrated environment than sex-segregated schooling (cf. Dale, 1971). Not surprisingly, from the outset, a number of research studies on the value of both types of schools have, among others, also addressed this issue. Yet, even though several of these studies did find favourable results in co-educational settings in terms of cohesiveness among pupils (Dale, 1971; Schneider and Courts, 1982; Foon, 1988; Hannan et al., 1996; Robinson and Smithers, 1999), some others indicated that pupils, girls in particular, identify more closely with their group in single-sex schools as compared with mixed-sex schools (Trickett et al., 1982; Mahony, 1985; Rowe, 1988; Lirgg, 1994; Kreienbaum, 1995; Shmurak, 1998).
In addition to these direct examinations, which provide contradictory data, there is a considerable body of literature that implicitly refutes the original 'natural atmosphere' hypothesis. For instance, regarding the nature of classroom interaction dynamics, it has been demonstrated that girls as well as boys tend to prefer same-sex peers from pre-school through secondary school, thus producing de facto gender segregation (Wilkinson and Marrett, 1985; Maccoby, 1990; Sadker et al., 1991). At the same time, in a number of studies, some of which are ethnographic in character, attention has been called to social practices in mixed school settings, such as differential expectations and treatment by teachers, gender-segregated extra-curricular activities, male dominance, disproportionate numbers of classroom interactions involving boys, gendered subject choice and girls' serving as a negative reference group (Delamont, 1990; Stables, 1990; AAUW, 1992; Sadker and Sadker, 1994; Lee et al., 1994; Abraham, 1995; Canada and Pringle, 1995; Shaw, 1995; Murphy, 1999). Generally, from this indirect evidence, it would appear that it is the girls' schools rather than the co-educational schools that offer benefits as far as fostering a sense of belonging or social bonding is concerned. Still, it would seem appropriate, because of the scarcity of immediate data on the affiliation among pupils in single-sex as compared with co-educational schools, to re-examine the aforementioned contradictory findings. This would appear all the more necessary since most of the earlier tests have been primarily descriptive - e.g. lacking proper controls and failing to develop a valid measure of sense of belonging.
The model specification
In order to examine the effect of mixed education, use is made of a multivariate model consisting of individual and school level variables, allowing more accurate estimation of the independent effect of school type. So, firstly, there is parental socio-economic status (SES) to be taken into account. The factor of social background may affect the parents' choice of either type of school as well as the children's feeling of belonging. In particular, as regards the feeling of belonging, it has been pointed out that pupils from the lower social classes feel less well integrated in school than middle-class pupils (Willis, 1977; Cox, 2000). Presumably the former are more likely to experience incongruity between schooling and parental values. Furthermore, given the fact that the home environment constitutes an important source of social capital which has important implications for other domains of social life (Rossi and Rossi, 1990), it should also be adjusted for an aspect of the quality of the parent-child relationship. For instance, pupils' perception of parental support not only affects social bonds beyond the parent-child relationship but at the same time appears to influence the way in which pupils experience their school environment (Brutsaert, 1993). In the same vein, a social psychological variable pertaining to the quality of the teacher-pupil relationship may function as a buffer against a potential negative school type effect. …