Cognitive outcomes
Non-cognitive outcomes
The costs of multi-grade
Several summaries of the advantages and disadvantages of multi-grade teaching in developed and developing countries are available (eg Pratt 1986, Miller 1991, Thomas and Shaw 1992, APEID 1989). All stress a different range of positive and negative outcomes, reflecting in part the variety of human and financial resource contexts in which different educational systems operate.
In industrialised countries much of the debate about multi-grade vs mono-grade arises in the context of rural depopulation among communities which have already achieved universal enrolment in primary education. The policy objective lying behind the research is often the reduction of costs and the closure of small schools. Those who advocate cost-savings highlight the high costs of keeping small multi-grade schools open, the cost-efficiency of transporting children to adjacent schools and, sometimes, the negative achievement and social effects of multi-grade on students. Those who resist the policy stress the role of the primary school in the life of the community in general as well as the benefits, especially social, for the children who study in a multi-grade structure. Sometimes the debate is pedagogic; sometimes economic. Members of the teaching profession discuss ideal ways of organising all schools and not simply those serving low and declining school-age populations. Multi-grade teaching is presented as a powerful pedagogical tool for promoting independent and individualised learning. In England and Wales for example the 1966 Plowden report was influential in encouraging "vertical grouping" as an alter native to "horizontal grouping", especially across the first three grades of primary education. The pedagogical argument was based on the possibilities for social development as well as peer and cross-age learning among children aged 5-7 years.
In Newfoundland and Labrador in Canada the debate is contemporary. Some argue that multi-grade classrooms offer exciting and challenging arenas for learning, and viable sites for high quality education; while others cast multi-grade classrooms as "unfortunate remnants and reminders of times past" in which children cannot possibly receive an education equal in quality to that provided in mono-grade classrooms (Mulcahy 1993). In Sweden, the general attitude to multi-grade is positive and many primary schools appear to be adopting multi-grade teaching out of choice. The heads of the 35% of primary schools which engaged in multi-grade teaching in 1987/8 reported a range of initial motivations for adopting multi-grade teaching strategies. Forty four per cent said that there initial motivation was purely educational, 35% purely resource-oriented and 20% a mix of both educational and resource reasons (Malmros and Sahlin 1992).
The general debate in the literature from industrialised countries is usually couched in terms of multi-grade versus mono-grade. Furthermore, because there is a high degree of congruence between age and grade in educational systems in industrialised countries the terms multi-age vs mono-age appear regularly in the literature on multi-grade teaching.
In developing countries the debate also arises in the context of rural and small populations, but these are often school-age populations which are growing rather than declining. General population growth and increased participation in schooling among communities which have not yet achieved universal access to primary education give rise to schools which tend to be located in areas where access to the next school is difficult, where facilities are already extremely limited and to which teachers generally do not wish to be posted. Because children tend to enter school at different ages and because promotion from grade is often non-automatic, there is often a lack of congruence between age and grade. The characteristics of comparison groups vary. Sometimes comparisons are made with mono-grade schools, sometimes only rural mono-grade schools, sometimes with schools which purport to be mono-grade but which have fewer teachers than grades and sometimes with schools where there is only one teacher for all grades.
Only rarely does the debate arise in developing countries as part of an educational discussion amongst teaching professions about the ideal way to organise schools in general, rather than in difficult rural areas. In general, multi-grade is viewed by the teaching profession as a second-class solution to educational problems which beset disadvantaged communities. Cost-savings often feature in the discussion, though usually in response to questions raised by agencies external to the country which are considering lending or aiding educational development. And, in contrast to the cost arguments rehearsed in North America and Europe, multi-grade is often presented as a strategy for rather than against cost-saving.
In the sections which follow, the findings from industrialised countries are separated from those from developing countries. Wherever possible the broader policy and educational context of the research will be noted. The majority of research studies located for this review focus on the cognitive and non-cognitive outcomes of multi-grade teaching for students and on the costs of multi-grade.
North America
Evidence from North America on the cognitive outcomes of multi-grade compared with mono-grade suggests that children perform no better and no worse in multi-grade classes. Pratt (1986) reviewed thirty experimental studies conducted between 1948 and 1983 in the United States of America and Canada. In view of the high degree of congruence in North American schools between age and grade the multi-grade classes were described as multi-age classes. All the multi-age classes contained an age range of 2-3 years and the achievement variables studies were usually reading and mathematics scores on standardised tests. Pratt notes that
many of the studies suffer from imperfect control of differences between teachers and schools which elected or rejected multi-age grouping. Too few of the studies reported sufficiently complete data to allow more than a counting procedure for summation of the results.(Pratt 1986: 113)
Of the thirty studies reviewed thirteen showed inconclusive results, ten favoured multi-age classes and five favoured mono-age classes. However, eight of these studies were doctoral dissertations which, Pratt argues, are likely to be relatively rigorous in their design. Five of the eight generated inconclusive results, one favoured mono-grade and two multi-grade. Miller(1991) confirms this general picture with a review of twenty one studies from the US which suggested that students in multi-grade classes performed as well as students in mono-grade classes. The reviews of neither Miller nor Pratt contextualise studies in time or place.
England
Rather less research on multi-grade and mono-grade teaching has been reported from England. In a study conducted shortly after the publication of the Plowden Report advocating vertical grouping, Mycock (1967,1970) compared the academic achievement of children in two vertically grouped infants schools with two which were horizontally grouped. She found no difference between the groups in terms of vocabulary growth, reading accuracy and mathematical skill. Ford's (1977) review of the effects of multi-grade on cognitive development confirmed this. However both Mycock and Ford found differences in favour of multi-grade on a range of socio-emotional factors.
Further research was stimulated in the 1980s in response to a survey of primary education in England published in 1978 by Her Majesty's Inspectors (HMI). This survey was carried out at a time when a fall in the numbers of children entering primary school had led a number of school to adopt vertical grouping. The HMI report (1978) highlighted a number of problems with this type of class organisation and suggested that 9 and 11 year old children achieved more highly in mono-grade than in multi-grade classes. They also suggested that teachers in mixed-grade classes were less able to judge the ability levels of students and to match tasks to these levels.
The findings of the HMI were accepted uncritically by Government Ministers and Used as a basis for national policy which rejected mixed-age classes as an acceptable form of classroom organisation (Bennet, Roth and Dunne 1987). Nonetheless, falling rolls continued and "headteachers found themselves in the invidious position of having to implement mixed-age organisation because of falling rolls despite educational and political pressure not to do so" (Bennet et al 1987:43). However, as Bennet et al point out
it is always dangerous to base policy on the findings of one study since no piece of research is perfect, and the HMI survey was no exception. Four methodological problems are apparent1. No attempt was made to differentiate between different kinds of mixed-age class; classes with a few children of a different age group from the majority were treated in the same way as classes containing equal proportions of children of three or four age groups.
2. Achievement scores were found to be significantly poorer in inner-city schools but HMI did not indicate whether there were more mixed-age classes in such schools. If there were, the link between mixed-age and achievement may not in fact be caused by mixed-age classes.
3. Assessments of the match of task and child were based on the judgements of a large number of HMI. Questions must therefore be raised about the consistency of such judgements and the criteria used.
4. Doubts must also be raised about the validity of the standardised tests used given the marked differences in curriculum which children follow...The extent to which the items of any test will match the actual curriculum taught will therefore vary widely.
(Bennet et al 1987: 43)
This study by Bennet et al study focused on the extent to which teachers were able to cater for the extremes of ability within their classes. The study examined in detail whether teachers in twelve mixed-age and nine mono-age classes could provide suitable learning tasks for the three highest and three lowest ability children in their class. They found that teachers in all the classes tended to direct the content, level and pace of work towards the average, and, in consequence, underestimated the abilities of the high attainers and overestimated those of low ability. Although this finding confirmed a trend identified for some teachers in the HMI report it refuted the claim that teachers in mono-grade classes behaved any differently.
Colombia
In chapter 2 the characteristics of Colombia's Escuela Nueva - the New School programme were described. Its key features are a multi-grade organisation allowing flexible, rather than automatic promotion, a rural-oriented curriculum and instructional materials designed for self study and individualised learning. An evaluation conducted in 1987 by the Ministry of Education in Colombia examined the achievement of grade 3 and grade 5 students in Maths and Spanish among a sample of 3,033 students drawn from 168 Escuela Nueva and 60 traditional rural schools. The traditional schools are described as mono-grade, following a national curriculum, providing no special attention to slow learners and not stimulating the students through special materials (Psacharopoulos, Rojas and Velez 1993, Colbert, Chiappa and Arboleda 1993). The mean scores showed that the grade 3 Escuela Nueva students scored higher in Spanish and Maths. Grade 5 students scored more highly in Spanish, but there was little difference in Maths. However, as Psacharopoulos et al (1993) point out, differences in gross means can disguise the effect of a wide range of factors on school achievement. The data were then reanalysed using an education production function in which a number of student, family, school and teacher characteristics were examined. These characteristics were
student - age, gender, repeater, works, hours watching TV
family - economic level of region, books at home, TV and radio at home, homework help
school - type of school, student-teacher ratio, electricity access, number of supervisory visits
teacher - gender, years of experience, educational background, residence and pay
Several of these factors had independent effects on achievement scores. For example, the scores of male students were higher than those of females in Maths; females performed better than males in Spanish; repeaters scored lower than non-repeaters. Higher scoring students were taught by university graduate teachers or by teachers who resided at the school. Nonetheless, taking all these factors into account, the effects of school type on the achievement of students i.e. whether the student was enrolled in an Escuela Nueva or a traditional school remained strong.
While the results of the analysis are compelling, several methodological puzzles remain. The first is the issue of "repeating". One of the variables in the model is repetition, and data are produced on repetition for both the Escuela Nueva students and "traditional school" students. The authors do not explain what repetition means in the context of Escuela Nueva where students work through self-instructional materials at their own pace. The second is the issue of comparison. Although the composition of teachers and grades in the traditional schools was not described by the authors, it appears that the comparison group consisted of students from at least three different types of school (i) unitary schools, with just one teacher for all grades, (ii) mono-grade schools, with one teacher for each grade, and (iii) with fewer teachers than grades (Velez personal communication). However, the reported analysis did not distinguish the groups in this way, and it would seem important to know whether the benefits of multi-grade remain when compared with all three types of traditional school. Third, and related to the first, is the issue of selectivity of the students in the samples. Because Escuela Nueva enables students to advance on the basis of mastery of module learning units rather than an assessment across a whole year, students advance at different rates. One student may spend sixteen months covering the modules equivalent to grade 3 while another may spend only seven months covering the same work (Velez, p.c.). Advancement is determined by unit mastery. Certainly the higher average age of the Escuela Nueva students classified as grade 3 and grade 5 students is consistent with this possibility. We are not told how the Escuela Nueva mastery assessment system operates, but we do know in general that mastery-oriented schemes of assessment require higher levels of performance from each child than the norm-referenced assessments which characterise much internal school assessment. Is it possible then that the average achievement of students entering a new grade in the traditional schools will be lower than that of Escuela Nueva students because of the nature of the assessment? This is a fascinating possibility which deserves further exploration, not only because of its implications for the evaluation of Escuela Nueva and other multi-grade systems, but also because of its implications for assessment and achievement within primary education systems more generally.
Indonesia
A study of the Indonesian "small schools" project, designed to help multi-grade teachers, is reported by Bray (1987). He reports that of 1,300 primary schools in Central Kalimantan, 460 have only one to three teachers. Self-study materials were developed for grades 4 to 6 and other types of material were developed so that adult volunteers could work with some children, releasing the teachers to work with children with particular difficulties... He reports that "the project students performed better in most subjects than did other students" (Bray 1987:43). However the study is not referenced and we do not know the characteristics of the comparison group.
Other studies from Indonesia also report achievement gains arising from multi-grade teaching. A 1975 evaluation of Proyek Pamong, described in Chapter 2, compared the performance of seven Proyek Pamong schools with seven control schools in rural communities similar to those served by the Pamong schools. Students were compared on conventional district inspectorate tests and block tests linked with self learning modules. Before the experiment began there were no differences in performance between the two groups. Not surprisingly the grade 4 and 5 Proyek Pamong students performed better on the innovatory block tests. Grade 4 students also performed better on conventional tests designed by the district educational administration, and grade 5 students performed no worse (Proyek Pamong 1975). However a thorough analysis of several Pamong evaluations some years later yielded inconclusive results (Thiessen quoted in Cummings 1986). Cummings notes that
with hindsight it is apparent that the tests used in many of these studies, being based on the conventional curriculum, did not adequately test the material taught on the PAMONG schools. Also, teachers in both PAMONG and non-PAMONG schools were known to manipulate scores.(Cummings 1986: 90)
Philippines
Positive results from the Philippines Project IMPACT were reported frequently (eg INNOTECH 1978) but Cummings (1986:89) cautions readers to interpret the results with care. He raises questions about the background characteristics of the non-IMPACT schools and students used for purposes of comparison. The educational and socio-economic levels of the students attending IMPACT schools were generally higher than those of students in the non-IMPACT schools and- the teacher characteristics more favourable. The cognitive achievement scores were generally higher for grade 4 students but showed no differences for students in the higher grades.
Much of the evaluation research referred to above also examined a range of non-cognitive outcomes of multi-grade teaching, including friendship patterns, self-concept and esteem, social development, work and school attitudes and civic behaviour.
North America
The non-cognitive outcomes examined in Pratt's (1986) review include children's friendships, self concept, altruism and attitude to school. Summarising the studies on friendship patterns he notes
the general picture that emerges...is one of increased competition and aggression within same-age groups and increased harmony and nurturance within multi-age groups
Overall he claims that social-emotional development of children is either accelerated in multi-age groups or shows no difference. He located fifteen studies between 1948 and 1983 which studied the effects of multi- and mono-age grouping on socio-emotional variables, most commonly self-concept and attitude towards school. Nine of the fifteen favoured the multi-age groupings, none favoured the conventional grouping, while six were inconclusive. Reviewing in more detail seven doctoral theses he finds that three favoured the multi-age groups while four yielded inconclusive results. Miller's review (1991) also indicates a number of favourable outcomes of multi-grade teaching. Citing his previous review of twenty one quantitative studies he claims
in terms of affective measures...multi-grade students out-performed their single grade counterparts at a statistically significant level.
He also reviews a number of qualitative studies which do not include comparisons with mono-grade schools. Among the positive benefits of multi-grade he cites social interdependence, independence, community involvement, self-reliance and cooperation. Ford's (1977) review lists a number of claims which have been made for multi-grade, among them
a greater sense of belonging and confidence; relationships with a wider age range of children; well adjusted personality; good work attitudes and high aspirations; better teacher-student rapport; low stress; stronger self concept of older, slower students; and better personal and social development
Ford challenges a number of these claims through reference to studies mainly from the US, but also from the UK. She claims that the research findings are mixed on the reduction of anxiety levels, friendship patterns, and on personal and social adjustment. She confirms more positive self concepts and greater self-esteem, benefits from a lengthened teacher-pupil relationship, more positive attitudes to school and better attitudes to work. However she also points out that there are important differences between the UK and the US practices of multi-grade which can influence findings. For example UK multi-grade classes tend to cover a wider age range than in the US, and pupils in UK multi-grade classes tend to spend more years with the same teacher.
England
Although Mycock's (1967) study of two vertically and two horizontally-grouped infants classes in England did not show any difference in measures of cognitive achievement, there were a number of differences in socio-emotional development. Children in the vertically-grouped classes experienced less stress on admission to school and a greater speed of socialisation into the school culture, a greater range of social interaction, better work attitudes, a closer and more secure relationship with the teacher and higher levels of aspiration on specific learning tasks.
Colombia
The Colombian studies referred to above also examined the effects of Escuela Nueva on measures of creativity, civic behaviour and self-esteem. Early evaluations had credited the programme with positive effects on self-esteem and civic behaviour, but had found no difference between students in the Escuela Nueva schools and the traditional schools on measures of creativity (Colbert et al 1993).
The more complex analysis of the same data performed by Psacharopoulos et al (1993) confirmed the positive effect for civic behaviour and null effect for creativity but not the positive effect for self-esteem. The factors which seemed to affect creativity positively were whether a student was older, had work experience, had not repeated a grade, had a TV at home, and was taught by a male teacher with higher pay, greater years of experience and who lived in the school. The factors which seemed to affect self-esteem positively were whether a student was older, female, a non-repeater, access to more books at home, was taught by a teacher without a university degree or by a teacher who lived in the school.
Indonesia
Cummings summarises the research from the Proyek Pamong experience in Indonesia and cites a study of the effects of Pamong on study habits, self confidence, initiative and cooperation. Although the study reported significant positive effects Cummings suggests that the analysis was flawed, making it difficult to draw confident conclusions about the effects of Pamong on the attitudes of students. Bray's (1987) brief account of the Kalimantan small schools programme suggested that children in multi-grade small schools working with self-instructional and programmed material were more self reliant than children in mono-grade schools.
The APEID Studies
The UNESCO/APEID study referred to earlier synthesised findings from country reports on multi-grade from twelve countries in the Asia and Pacific Region. The study listed four advantages of multi-grade teaching situations, all of them non-cognitive:
students tend to develop independent work habits and self study skillscooperation between different age groups is more common resulting in a collective ethics, concern and responsibility
students develop positive attitudes about helping each other
remediation and enrichment activities can be more discreetly arranged than in normal classes
(APEID 1989: 5)
It is clear from the country reports that these advantages represent the views of conference participants and are not grounded in systematic study. At the same time it is important to note that they stress the non-cognitive advantages and make no claims for the cognitive. This is consistent with most of the findings reported above, save perhaps those of Escuela Nueva, which claims superior cognitive outcomes for multi-grade schools. The claims about independent study, cooperation and socio-civic behaviour attract some support from a few studies in the US and Colombia. However in general, systematic data on non-cognitive outcomes from multi-grade settings in developing countries are so few that it is difficult to substantiate the APEID claims through an appeal to research findings.
Thomas and Shaw's (1992:11) summary of the effects of multi-grade draws on some of the studies above but also includes studies from India, Pakistan, and Togo. Although their generalised claims are a little incautious, few could disagree with the gist of their concluding lesson
although somewhat scanty, the evidence emerging from the developed and developing worlds leads to the conclusion that multi-grade schools are just as effective as single grade schools in educating students. In some cases, students have attained higher levels of achievement in academic subjects as well as in social-civic indicators than their single grade counterparts. Peer tutoring, repetition, self learning and improved opportunities for socialisation are important ingredients for success. More importantly students in multi-grade classes "learn to learn".One may conclude that when programs are correctly implemented, students may attain higher achievement levels and improve social skills. But students in multi-grade schools which fail to adopt effective pedagogical techniques tend not to perform as well as their counterparts in single grade schools. The lesson to be drawn from this is that in order for a multi-grade school work well teachers must master and use effective teaching practices, be supported through training programs, and have appropriate texts and materials at their disposal.
The high costs of maintaining small multi-grade schools in North America and Europe has often been used as a rationale for closing down small schools. The costs of transporting, and sometimes boarding students at mono-grade schools are offset by substantial savings on teacher and ancillary staff costs and maintaining buildings and lands. The need to reduce costs often leads to the closure of the multi-grade school. In developing countries, on the other hand, the cost argument is presented rather differently. Multi-grade schools offer cost-savings, especially in situations where school-age populations are growing. As Thomas and Shaw (1992:8) point out
among the most obvious costs of setting up a multi-grade classroom are: furnishing and equipping the classroom, and providing self-learning materials and textbooks. Gains can be expected in terms of increased efficiency of the system resulting from lower repetition and dropout, and more efficient use of human and capital resources. It would not be unusual for expenditures in a multi-grade school to result in higher costs per student but lower costs per graduate, thus resulting a cost-efficient option. The approach becomes cost-effective when it results in increased achievement.
The few studies of the costs of multi-grade have almost always been conducted in the context of foreign-funded support for education and most appear to indicate that multi-grade is not a high-cost strategy for rural schools, especially when compared with mono-grade schools in similar settings (eg Cummings 1986:91-92 Psacharopoulos et al 1993:275, Colclough with Lewin 1993: 1302,138). Psacharopoulos et al (1993) point out that the quality benefits of the Escuela Nueva programme were achieved at a unit cost per student just 5-10% higher than those in traditional rural schools. They urge a degree of caution in interpreting this result, and point out that a full cost study has not yet been carried out. The cost of teacher training in the Escuela Nueva programme for example was three times higher than traditional teacher training.
However the nature of the comparison is important when evaluating results and comparing within and across countries. For example, should the cost and quality benefits of multi-grade in the Escuela Nueva programme be compared with traditional rural primary schools in Colombia, or with schools in urban centres? (Colclough with Lewin 1993:138). Or should the benefits of multi-grade be compared with the outcomes for children not attending school at all, a possibility which, in some countries, remains distinct.
