7.1 Three years of systematic observation have identified only a few significant differences in teaching approach between small and normal size classes.
Where variations exist these are mostly a consequence of changes in year group (more factual questions in P1; more challenging ones in P3) or have to do with subject differences (pupils are more often listened to when reading in English while in mathematics they are listened to when offering explanations).
In all years there has been little variation in the amount of individual help that pupils receive when they are the sole focus of the teacher’s attention. The main difference concerns the manner in which such attention is received. In smaller classes individual attention (mainly when part of a group or whole class) is given in bursts and is more likely to be sustained over the 30 second time unit between successive observations. In normal classes such attention consists of brief exchanges. The inference here is that in the smaller classes there is a possibility that more insightful discourse may occur. In a small scale study of English classrooms Hargreaves, Galton and Pell (1998) found that more complex discourse, built around sustained teacher-pupil interactions, was the main difference between small and large classrooms.
7.2 As discussed earlier in the present report (paras 2.7 and 2.8) this finding is in accord with the results from most other studies of class size reduction, even where the numbers of pupils in a class are around 15. Mathematically, reducing the class size from the high 30s to the low 20s affords the possibility of only limited increases in individual attention, particularly in the present
study where over 60% of most lessons consist of whole class teaching during which time the teacher is talking and no individual pupil is the focus of his or her attention.
7.3 However, there is a further factor that helps to minimise any significant variations between small and normal classes. So far, in the study comparisons have been made in terms of mean differences. This takes little account of the variation that undoubtedly exists between individual teachers (apart from using the standard deviation to determine the level of statistical significance and the subsequent effect size). Using mean scores runs the risk that the high level of interaction of one teacher on a certain category of behaviour is cancelled out by the low level of activity of another, thus masking real differences in teaching behaviour.
7.4 An alternative approach is to use a statistical technique known as ‘cluster analysis’. This is a procedure for placing individuals in groups such that the variation between members of any group is less than the variation between different groups. In the present study therefore we can use the technique to group teachers (and pupils) who have similar interaction patterns. In all, data from 395 classrooms were available for analysis. In a minority of cases a teacher who taught a P1 class was also observed when teaching P2 or P3 classes. However because the previous analysis showed that the main variations across the interaction categories was a function of the different year groups, the class rather than the individual was chosen as the unit of analysis.
7.5 Table 7.1 shows the characteristics of the 4 cluster solutions. Type 1 teachers are mainly identified by the relatively high amounts of attention they give to individual pupils (both boys and girls). They tend to prefer pair work and have the highest levels of sustained conversation of the four types. They are second only to Type 4 teachers in asking open-ended questions and highest on statements of ideas. Feedback tends to be of the informing rather than the correcting kind. These teachers might therefore be describes as individual and pair sustained enquirers.
Table 7.1 Characteristics of the four teacher types1
Observation categories Mean % of all observations Effect size Type 1
(N=119)
Type 2 (N=73)
Type 3 (N=119)
Type 4 (N=84) Questions
Which recall facts 3.90 4.89 3.31 7.14 M
offer closed solutions 8.95 7.35 13.95 13.61 M
offer open ideas 5.14 3.06 2.44 5.79 M
refer to task supervision 2.86 3.70 1.09 3.73 M refer to routine activity 1.19 0.46 0.28 0.75 S
All questions 22.03 19.46 21.07 31.03 M
1 In some categories rounding off the figures to two places of decimals may mean that the totals do not add exactly to 100.00.
Observation categories Mean % of all observations Effect size Type 1
(N=119)
Type 2 (N=73)
Type 3 (N=119)
Type 4 (N=84) Statements
of fact 10.65 9.37 15.93 9.84 M
of ideas 4.11 0.78 1.57 1.83 M
of task direction 16.12 25.86 15.33 19.01 L
of correcting feedback 5.72 6.07 6.39 4.60 None
of informing feedback 6.40 4.34 2.78 4.21 M
of behaviour feedback 4.78 2.47 4.51 4.07 S
of routine directions 5.78 4.47 5.30 8.85 M
All statements 53.55 53.36 51.81 52.40 None
Listening/watching pupils
report or explain 11.27 12.06 9.87 7.20 S
Read 5.57 2.28 10.44 5.14 M
in silence (monitoring) 8.83 12.93 6.81 4.50 M
All listening/watching 25.87 27.27 27.11 16.85 M Audience
individual boy 5.75 1.69 1.21 1.98 L
individual girl 2.61 0.82 0.64 0.52 M
boy for group 2.58 9.50 2.19 2.42 L
girl for group 1.79 9.18 1.12 2.48 L
boy for class 12.62 7.12 13.65 11.87 M
girl for class 9.38 4.02 9.58 9.15 M
Pair 8.38 0.50 1.51 2.10 L
whole group (no one in focus) 8.20 20.41 6.67 8.49 L whole class (no one in focus) 48.69 46.74 63.42 60.99 L
Sustained 15.96 9.51 10.40 11.67 S
Effect size: S=small; M=medium & L=large where 0.2<S <0.5 <M<0.8 <L (Cohen 1988)
7.6 Type 2 teachers favour the use of group work. They have the lowest level of questioning overall but the highest percentage of task direction statements, presumably to do with setting up the groups. They spend most time either listening to pupils report or explain or else monitoring the pupils’ activity.
They might be described group task monitors. Both Type 3 and Type 4 teachers favour whole class teaching. Whereas Type 4 teachers are distinguished by high levels of questioning across all five categories stated in Table 7.1. Type 3 teachers have high levels of statement of fact and of correcting feedback. When listening to pupils it is mostly to hear them read rather than silently monitoring pupils while they work. We might thus describe Type 3 teachers as whole class instructors while Type 4 practitioners might more aptly be named whole class questioners.
7.7 In Table 7.2 the proportions of these teacher types in small and normal classes are examined. The analysis is shown separately for the 3 year groups given the possibility that as pupils advance through primary school the teaching approach changes. In general, there tends to be a mix of teaching approaches in the small classes, particularly in the P1 year. In normal classes, however, one type tends to dominate in the different year groups. Thus 60% of teachers
in P1 normal classes are whole class instructors while in P2 nearly half (46%) are whole class questioners and 54% of P3 normal classes use a mixture of individual and pair work to promote sustained enquiry. These findings suggest that when teaching small classes many teachers were still experimenting with different approaches, even after three years, while in normal classes there seems to be a progression from direct instruction in P1 to more interactive class teaching in P2, while in P3 pupils are given more responsibility for managing their learning so that the teacher’s support becomes more individualised. A further finding which is common to all clusters emerges from Table 7.1 is that irrespective of whether a pupils is the focus of the teacher’s attention individually, as part of a group or as part of a class, it is boys who in every case receive more attention than girls.
Table 7.2 % of teaching types in small & normal classes in P1 to P3 years Teacher types P1 classes P2 classes P3 classes
small normal small normal small Normal Individual and pair sustained
enquirers
31 21 41 15 17 54
Group task monitors 21 6 14 15 31 5
Whole class instructors 23 60 23 23 34 28
Whole class questioners 25 13 23 46 17 5
Number of teachers 181 48 44 13 70 39
7.8 It is also possible to investigate the links between specific subjects and teaching types. This is done in Table 7.3. Here because of small numbers it is not possible to conduct the analysis across the three year groups so the results are an aggregate of P1, P2 and P3. For normal classes whole class instruction is a favoured teaching approach used across all subjects although for English individual instruction and pair work also form strong elements of typical practice. In small classes, however, there is more variation. In Chinese there is evidence of the increasing use of group work and in mathematics a switch to individual and pair work with an emphasis on challenging questions rather than factual statements. English shows some change although individual enquirers and class instructors appear to remain the dominant teaching approach. These findings support the comments based on the consultant’s school visits in the first two years of the study, where it was suggested that English teaching, in particular, had tended to regress in the use of whole class instruction. Both Tables 7.2 and 7.3 suggest a willingness among practitioners to experiment in the use of different teaching methods when faced with a smaller class. Further analysis shows a shift away from group task monitoring towards more challenging individual and pair work among teachers with the greatest experience of SCT during the three years in which observations were made.
Table 7.3 % of teaching types in the three core subjects
Teacher types Chinese English Mathematics
small normal small normal small Normal Individual and pair sustained
enquirers
20 28 29 42 37 24
Group task monitors 27 10 12 0 25 11
Whole class instructors 26 48 35 40 18 40
Whole class questioners 27 14 24 18 20 25
Number of teachers 94 29 101 38 96 37
7.9 A similar analysis can be carried out with the pupil observations. Again four substantial clusters emerged. Because of the large number of categories only those discriminating between the clusters which help to identify the pupil types are shown in Table 7.4. They bear a remarkable similarity to pupils in English primary classrooms who were first identified in the ORACLE (Observational Learning and Classroom Learning Evaluation) study and subsequently replicated in a repeat of the original research two decades later (Galton et al. 1980, 1999). Type I were described as solitary workers because they are on task for almost the entire lesson even when the teacher is engaged with other pupils elsewhere in the classroom. Although they sit in groups for nearly 70% of the time they work on their own. Their contact with the teacher mainly involves being part of his/her audience when nobody in the class is in focus. Such pupils were first identified by McClelland (1963) in North American science classes and were described at the time as ‘undeflected workers’. McClelland likened these students to human cannonballs because they remained on course, despite the vagaries of the teaching. In the SCT study they form 43.8% of the observed sample.
7.10 Type 2 pupils (22.4% of the sample) were labelled intermittent workers in the ORACLE study. These pupils have the highest levels of distraction. There is a slight emphasis on sitting in pairs and the data suggests that while these pupil work when the teacher is present they can easily be distracted when the teacher is either elsewhere, monitoring class activity or when these pupils are part of the teacher’s audience when sitting as a class at the front of the room.
Table 7.4 Characteristics of the four pupil types
Observation categories Mean % of all observations Effect size Type 1
(N=690)
Type 2 (N=353)
Type 3 (N=368)
Type 4 (N=165) Target pupil activity
Target’s behaviour
On-task 94.49 73.37 90.24 65.45 L
Distracted/partially distracted 5.07 23.23 8.10 19.39 M
Observation categories Mean % of all observations Effect size Type 1
(N=690)
Type 2 (N=353)
Type 3 (N=368)
Type 4 (N=165) Target pupil’s location
In base 99.31 98.44 97.76 78.48 L
Out of base 0.40 0.99 1.77 19.09 L
Teacher’s activity
Elsewhere in the classroom 66.59 45.54 69.84 60.91 S
Monitoring 2.72 22.80 6.66 6.52 M
Seating
Alone 0.15 24.72 5.77 6.82 L
Pair 0.15 24.72 5.77 6.82 S
Group 69.35 45.05 73.03 60.76 S
At front of class 17.14 8.92 4.01 20.91 S
Mode of working
Individual 59.24 60.34 15.29 40.61 L
Pair 2.39 2.90 21.81 6.21 M
Group 16.27 5.45 52.38 30.76 L
Class 22.10 31.23 10.53 22.42 S
Target pupil-adult activity Target’s role
Seeks interaction with adult 0.98 1.84 1.49 10.61 M Part of teacher audience 71.38 72.59 21.67 46.36 L Content
Task work/supervision 71.59 72.31 26.09 41.21 L
Feedback on work: praise 1.09 0.92 1.36 7.27 M
Setting
T gives individual attention 0.00 2.27 1.49 4.55 S T interacts with whole class 65.80 61.40 13.25 39.39 L Target-peer activity
Target’s role
Target initiates contact 4.57 3.68 25.53 9.24 L Target responds to contact 7.14 4.32 33.45 14.39 S Target sustains interaction 0.00 0.00 2.92 5.45 M Mode of contact
Speaking-listening 7.97 5.67 43.86 18.94 L
Task2
Same 12.32 7.22 62.50 25.45 L
Different 0.04 1.27 0.75 5.78 M
Effect size: S=small; M=medium & L=large where 0.2<S <0.5 <M<0.8 <L (Cohen 1988)
7.11 The third type of pupils was called active collaborators in the ORACLE study.
They make up 23.4% of the observed Hong Kong sample and are working on their task for over 90% of the lesson, even when the teacher is engaged elsewhere in the classroom. They generally sit and work in groups where they both initiate and respond to other pupils working on the same task. In many
2 Task here refers only to situations where the target pupil is interacting with either one or more peers.
The totals do not add up to 100% since such interactions are only a proportion of the target’s behaviour.
ways these pupils are similar to the solitary workers, the main difference being that they tend to operate in collaborative settings rather than on their own.
Finally the fourth type of pupil, originally labelled attention seekers, in the ORACLE study are perhaps more aptly described as attention grabbers because they not only seek but also more often succeed in obtaining the teacher’s undivided attention. They have the lowest levels of on-task behaviour and tend to be frequently out of their base area, either because they are required to sit at the front of the class or they are perhaps seeking assistance or, more likely, given the relatively high levels of distraction chatting to another group. Such conversations tend to be sustained. These pupils more often attempt to seek the teacher’s presence or else succeed in gaining his/her attention. We might hypothesise that some of these pupils are perhaps in need of reassurance since the teacher’s feedback often consists of praise. These attention grabbers form 10.5% of the sample.
7.12 There are no significant ability variations across the four types of pupil and neither do the proportions vary between the three subject domains. There are however gender differences. Of the 690 solitary workers 378 or 54.8% were girls whereas among the attention grabbers 98 of the 165 of this type of pupil (59.4%) were boys. Table 7.5 looks at the distribution of the various pupil types between small and normal classes and between year groups.
Table 7.5 % of pupil types broken down by class size and year group Solitary
workers
Intermittent workers
Active
collaborators
Attention grabbers
N Class
Small 43.2 21.2 25.2** 10.4 1177
Normal 45.4 25.8 18.0 10.8 399
Year
P1 44.0 24.6 20.2 11.2 913
P2 41.9 20.7 23.3 14.1 227
P3 44.3 18.6 30.0** 7.1* 436
** = P< 0.01: * = p<0.05 (small effect size)
7.13 The main difference between the small and normal classes lies in the increase in number of active collaborators in the former. As pupils move from P1 to P3 the percentage of this type of pupil also increases (significant at the 1%
level) while the number of attention grabbers decreases (significant at the 5%
level). Most of the changes occur in P3. When the percentages in the P3 year are broken down by class size it emerges that the increase in active collaborators arises from a decrease in both the intermittent workers (down from 28.2% to 13.2%) and the attention grabbers (down from 11.5% to 4.6%).
These are the groups with the highest levels of off task behaviour. In both cases the level of significance reaches 1% and result in medium effect sizes.
7.14 It is also possible to examine the relationship between teaching and pupil types and this is shown in Table 7.6. Only the profile of the whole class instructors differs from the overall distribution giving rise to a small effect size at the 1%
significance level. The fact that these class instructors have a higher
proportion of pupils ‘on task’ may, however, be no guarantee that it necessarily results in increased understanding, since when the teacher is addressing the class and these pupils are listening there may some who are uncomprehending. The fact that, overall, pupil classification appears independent of teaching approach (at least in terms of this typology) suggests that the different pattern of behaviour exhibited by certain pupils may partly be a function of their personality. It might, for example, be that the attention grabbers consist mainly of pupils who are shy, anxious introverts who need to seek constant reassurance.
Table 7.6 Variation of % of pupil profiles by teacher types
Pupil Type Teacher type N
Individual and pair sustained enquirers
Group task monitors
Whole class instructors
Whole class questioners
Solitary workers 36 42 53** 42 690
Intermittent workers 25 19 21 22 353
Active Collaborators 26 32 17 23 368
Attention Grabbers 13 7 9 12 165
Total Percentage 100 100 100 100
1576
N 528 240 498 310
** =p<0.01 (small size effect)
7.15 It is also possible to examine the relative effectiveness of the different teaching approaches in small classes as they manifest themselves in the descriptions of the various teacher types. This possible for P1 and P2 classes in each of the subjects since only Cohort 1 were in small classes in P3 and numbers of teachers in the various cells then becomes impossibly small. For each analysis the scores of the pupils (both attitudes and attainment) were converted to percentages, ranked and divided into four quartiles. For each teacher type the top, bottom and combined middle range percentages were then aggregated.
7.16 At P1 none of the four teacher types’ pupils make significantly more progress in attainment in all the three subjects but there are some differences in respect of attitudes. When Chinese ‘learning disposition’ scores (combined attitude and motivation) were analysed then pupils in the top third range who were taught by type 1 teachers (individual and pair enquirers) had significantly higher scores (p<0.05; large effect size). In English, however there were no significant differences for either attainment or attitude but in mathematics it was the type 4 teachers (whole class questionnaires) whose pupils in the top third range exhibited the strongest mathematics learning disposition.
7.17 At P2 there were no differences either in Chinese attitudes or attainment but in English, the pupils of whole class questioners (type 4) who were in the top third range did best overall in attitudes and attainment (p<0.01; large effect size). Interestingly among the English small classes in P2 there were no type 2 (group task monitors) teachers. In mathematics at P2 there were neither attitude nor attainment difference across all four teacher types. A similar analysis could not be conducted for P3 since only Cohort 1 classes were small ones and numbers did not allow an extended analysis of this kind.
7.18 The reason for so few significant effects probably lies in the small numbers of classes involved. There were 69 P1 and 41 P2 teachers of small classes. By the time these were sub-divided between three subjects and four teacher types the range of pupils in any combination of category (subject x type) ranged from 1 to 10. In an effort to improve numbers in each cell the aggregated score in both attainment and learning disposition were also analysed. No differences in attainment emerged but the P2 pupils of Type 3 (whole class instructors) had the poorest learning disposition while those of the whole class questioners (type 4) had the best (p<0.05; large effect size).
7.19 In summary, the main conclusion to emerge from this analysis is that while in normal classes most teachers stuck to the whole class instructor mode, within the small classes there continued to be a degree of experimentation across the remaining three approaches. This in part may account for the inconsistency in the attitudes and attainment differences since teachers had not yet decided in their own minds which activities were suited to the use of a particular approach. In the use of group work, for example there was still a high proportion of disengaged pupils, a fact confirmed by the observations of the consultant on his school visits. Looking at ways of devising group tasks so that more pupils participated for a greater part of the time was clearly an issue which teachers were still having to grapple with.
8. Impact of small classes on the Territory-wide System Assessment (TSA)