principals taking a more active role in curriculum development and in the professional development activities concerned with improving the effectiveness of SCT. They are more likely to deploy resources in school to provide additional non contact time for teachers to participate in intra- and inter-school activities. Whether by reputation or location these successful schools attract more Hong Kong born children which results in higher attendance at kindergarten and greater parental participation in school activities coupled with a higher level of support for these pupils outside school, whether it involves completing homework, undertaking educational visits or borrowing books from the library. These schools also succeed in attracting more specialist mathematics teachers and therefore have fewer teachers with a mainly Chinese initial training qualification teaching the subject. The above factors combined seem to exert a greater influence on a school’s performance than do the other variables considered such as class size and teaching approach.
Although there is a trend indicating that in the classes of the more successful schools there are more sustained cognitively challenging interactions between the teachers and individual pupils and a preference for working in pairs rather than groups, these differences fail to reach statistical significance. The least successful schools, in fact, succeed in having more pupils ‘on task’ but this engagement doesn’t appear to be matched by equivalent levels of knowledge and understanding.
10. Relative performance of schools with high proportions of disadvantaged
subjects. In P1 the residual gain for the disadvantaged schools is -1.80 against +0.21 for the remaining schools in the sample (5% significance level, although with a very small effect size). From the end of P1 to the end of P2 the corresponding figures are -0.43 and +0.05 which is not a statistically significant difference. But in the following year to the end of P3 the difference is again significant at the 5% level. (-1.01 compared to +0.15 but with a very small effect size). In the P1 year there are no gender effects but by the end of P2 girls’ performance in English in the disadvantaged schools had deteriorated to the extent that the deficiency now has a medium effect size. In P3, girls performance in Chinese in the disadvantaged schools was significantly below that of girls in the normal schools (1% significance, small effect size) while boys in the two types of schools scored similarly. In P4 the deficiency of the disadvantaged schools in English persisted, and as predicted from P3 mathematics scores, the disadvantaged schools had now fallen behind significantly. Both boys and girls showed similar trends in mathematics but although effect sizes were small, they were higher for the girls implying a sharper difference. The trend in Chinese evident in P3 has been reversed and scores in the two types of school are similar. Whether this has any connection with the return to normal classes is hypothetical.
Table 10.1 Cohort 1 comparison in attainment of ‘standard’ and
‘disadvantaged’ schools Period
of test
Type of school
Chinese English Mathematics
Mean s.d. N Mean s.d. N Mean s.d. N
Start of P1
Standard 23.77 7.43 3417 21.70** 8.45 3417 20.35 7.22 3417 Disadvantaged 23.48 7.92 405 20.30 8.18 405 20.65 6.80 405 End of
P1
Standard 36.43 19.00 3385 52.73** 22.39 3385 43.69 24.62 3403 Disadvantaged 34.69 19.97 414 47.08 22.16 417 41.70 25.84 417 End of
P2
Standard 51.41 18.20 3406 58.29** 23.81 3420 53.51 20.68 3407 Disadvantaged 52.12 19.94 483 48.67 23.23 479 52.66 20.04 485 End of
P3
Standard 47.98˚ 18.39 3259 34.05** 22.09 3245 59.79˚ 21.47 3255 Disadvantaged 45.85 18.83 470 27.17 19.50 473 57.42 22.85 474 End of
P4
Standard 51.31 17.18 3147 40.83** 22.74 3161 51.29** 21.17 3150 Disadvantaged 51.88 17.26 454 33.16 20.11 456 47.37 21.25 458
**p<0.01; *p<0.05 small effect sizes
˚p<0.05 very small (negligible effect size)
10.3 Cohort 2 was taught in small classes up to the end of the P2 year. This allows a partial replication of the results for Cohort 1 and the results are presented in Table 10.2. As with Cohort 1 the disadvantaged schools had a higher proportion of mainland born children with all the attendant consequences.
However in the P1 year, although the trend in English replicates the finding from Cohort 1, the pupils from disadvantaged backgrounds out-perform their peers in the remaining experimental schools on the end of year tests in both Chinese and mathematics. Although the mean in Chinese for disadvantaged pupils on entry is the same as that of their peers in the remaining experimental schools at the end of P1 the former have gained a small advantage (p<0.01 but very small effect size). In mathematics the pre-test mean of the disadvantaged pupils is below that in the other experimental schools but by the end of P1
there is a significant gain (5% level) in favour of the disadvantaged schools although the effect size is again very small. This is confirmed when the residual gains for the combined scores are calculated. That for the disadvantaged schools is +2.07 against a value of -0.27 for the remaining schools in the SCT study sample (statistically significant at the 1% level but small effect size). This difference comes about because boys in the disadvantaged schools do exceptionally well.
10.4 By the end of P2, however, these advantages have disappeared. There are no significant differences between the two groups of schools in Chinese and mathematics while in English the poor performance of the disadvantaged pupils relative to their peers in the remaining schools continues. When the combined scores are calculated the mean residual gain for the disadvantaged schools is now -1.61 against a value of +0.232 for the remaining schools. This change mainly comes about because of the deterioration of the girls’
performance, thus replicating the pattern that emerged with Cohort 1. The passage through P3 for Cohort 2 reflects the findings with Cohort 1 for Chinese and English, but having returned to normal classes the disadvantaged schools mathematics scores fell significantly below those of the normal schools. It should be noted that the same effect occurred with Cohort 1 in P4, again when classes returned to normal. However, the dip in both cohorts in the disadvantaged schools on returning to regular classes was larger than for the other SCT schools. The inference can therefore be made that without spending this time in small classes the gap between the disadvantaged schools and the rest would have been even bigger. From this it may be concluded that smaller classes enhance the mathematics performance in schools with high proportions of disadvantaged pupils. The effect is more noticeable with girls although it is also present with boys to a lesser degree.
Table 10.2 Cohort 2 comparison in attainment of ‘standard’ and
‘disadvantaged’ schools Period
of test
Type of school
Chinese English Mathematics
Mean s.d. N Mean s.d. N Mean s.d. N
Start of P1
Standard 64.75 25.84 2836 53.66** 27.51 2836 56.25 22.22 2836 Disadvantaged 64.74 25.84 375 48.50 26.37 375 54.58 20.99 375 End of
P1
Standard 37.99 19.91 2829 52.36** 23.43 2822 44.64 24.57 2813 Disadvantaged 40.22˚ 20.27 368 48.16 22.55 388 47.49˚ 23.67 388 End of
P2
Standard 52.21 19.39 2672 53.83** 24.11 2654 54.18 21.72 2665 Disadvantaged 53.45 21.16 376 48.67 22.87 374 54.40 22.48 379 End of
P3
Standard 49.17 18.75 2895 34.65** 22.62 2902 60.66** 21.82 2871 Disadvantaged 49.20 19.57 458 28.04 20.00 454 56.96 22.70 459
**p<0.01 small effect size
˚p<0.05 very small effect size
10.5 Thus over the two Cohorts pupils in the disadvantaged schools initially hold their own or do slightly better than pupils in the remaining ‘standard’ schools in Chinese and mathematics by the end of the P1 year, although the effect sizes are very small. By the time pupils enter the P3 year however the advantage has moved towards the remaining experimental schools. In both
Cohorts the small classes in the disadvantaged schools favours boys rather than girls and it is mostly the deterioration in the latter group’s performance that reduced the advantages gained in P1. In English, however, pupils in the disadvantaged schools are handicapped on entry to primary school as their initial P1 scores are significantly below those of pupils in the remaining schools. The position worsens as these pupils move through P2 and P3.
However, it can be reasonably argued that if these pupils from disadvantaged schools had remained in normal size classes then the deficit in English and perhaps also in Chinese and mathematics would have been even greater. This thesis can be tested by examining the differences in the control classes.
10.6 Examining differences within the control groups of all 37 experimental schools does not allow the progress of the same pupils to be followed over three years. P2 pupils in the control classes were tested in 2004/05, moved to P3 in 2005/06 and P4 in 2006/07. P1 pupils in Cohort 3 were not tested until 2006/07 and moved to P2 in the 2007/08 school year but these pupils had a possible advantage in that some of the teachers had, by that point in time, considerable experience of small classes in the intervening years. Hence it is difficult to interpret result so only the P2, P3 and P4 comparisons are shown in Table 10.3.
Table 10.3 Comparison of P2, P3 & P4 attainment in ‘standard’ and
‘disadvantaged’ control groups Period
of test
Type of school
Chinese English Mathematics
Mean s.d. N Mean s.d. N Mean s.d. N
Start of P2
Standard 54.50 20.02 3839 64.85* 21.36 3799 60.23 23.24 3829 Disadvantaged 52.40 20.23 437 58.84 19.31 443 59.94 22.57 439 End of
P2
Standard 54.00* 18.21 3864 59.42** 23.44 3854 51.43˚ 20.56 3858 Disadvantaged 50.82 19.05 456 49.31 23.31 457 49.23 20.99 453 End of
P3
Standard 46.42 17.01 3185 31.85** 21.07 3207 60.66 20.57 3262 Disadvantaged 45.23 17.24 424 26.32 19.32 427 58.45 20.40 428 End of
P4
Standard 54.90 17.67 3487 41.58** 23.33 3480 52.30 21.12 3487 Disadvantaged 53.26 17.97 422 34.54 21.59 421 50.91 21.72 50.91
**p<0.01; *p<0.05 small effect sizes
˚p<0.05 very small effect size
By the end of the P2 year the disadvantaged schools all trail the other experimental schools. The difference for English is greatest (1% significance level); that for mathematics the smallest (5% level) although because of the large standard deviations the effect sizes are small. By the end of P3, however, the disadvantaged schools have caught up in Chinese, have almost done the same in mathematics but in English the position has continued to deteriorate.
As with the previous comparison with small classes it is the relatively poor performance of girls, particularly, in mathematics that produces the results.
Disadvantaged boys, match the overall performance of their peers in the remaining schools and the difference between their scores and those of the girls results in a large effect size. When the three scores are combined the
residual gain for the shift from P2 to P3 is +0.74 for the disadvantaged schools against -0.11 for the remaining schools with standard populations of pupils.
10.7 By the end of P4 while both boys and girls in the disadvantaged schools continue to score significantly below the remaining schools in English (1%
level but small size effect) they maintain parity in both Chinese and mathematics, although in the latter subject girls in the disadvantaged schools again underachieve. Partly because of this factor and the poor English performance, when the residual gains from P3 to P4 are calculated using the combined scores on all three subjects, the value for the disadvantaged schools is negative (-0.83) against +0.11 for the remaining sample. Although this difference is statistically significant (5% level) it yields an extremely small effect size.
10.8 The data in Table 10.3 offers an interesting contrast to the earlier comparison involving smaller classes. Apart from English where the performance in the disadvantaged schools continues to deteriorate more sharply in every year irrespective of class size, the pattern appears to be asymmetrical. In normal size classes the disadvantaged schools start with a deficit in P2 which is presumably carried over from P1 and at the end of the year this gap in performance has been enhanced. But by the end of the P3 year the difference has been eliminated in Chinese for both genders and in mathematics for boys although not for girls. In P1, English again being the exception to the rule, small classes appear to give disadvantaged schools a better start which is maintained into P2 but then falls away in P3. This effect can be seen by examining the year-to-year residual gain scores. In Cohort 1 the P2 residual gain (using the aggregated score across the three subjects) for the small class sample was -0.43 for the disadvantaged schools and +0.05 for the remainder.
In the sample of normal size classes the corresponding figures are -2.71 and +0.31 respectively. But by the end of P3 whereas the values in the small classes were -1.01 and +0.15 those in the normal ones were in favour of the disadvantaged schools with values +0.74 and -0.11 respectively. Thus on the basis of this analysis smaller classes do allow pupils in disadvantaged schools to catch up in P1, provided the initial attainment of pupils in these schools is not too far behind as it tends to be in English. But any advantage gained has been lost by the end of P3 where as the earlier analysis of the TSA scores also suggested differences between small and normal size classes become relatively insignificant.
10.9 A similar analysis can be carried out using the attitude and motivation measures. In Cohort 1 these were first measured at the end of the P1 year.
Figure 10.1 shows the comparisons between the remaining 32 schools with standard population distributions and the disadvantaged schools. While the motivation and self esteem of pupils in the disadvantaged schools was significantly lower at the end of P1 (1% level very small effect size) the difference had largely disappeared by the end of P2 and this was also true of P3 (although the scale used in Figure 10.1 tends to magnify small differences).
Girls’ motivation and self esteem declined more sharply than did that of the boys. With subject attitudes only English at the end of P1 showed a significant difference (5% level). At the end of P2 and P3, however, both Chinese and
mathematics scores were significantly lower in the disadvantaged schools mainly due to a faster decline in boys’ dispositions. Effect sizes in all cases were very small. Combining all four measures into a single learning orientation score yielded a residual gain from P1 to P2 of -0.03 for the disadvantaged schools while that for the remaining schools in the SCT study sample is zero. Between P2 and P3 the corresponding figures were -1.0 (disadvantaged) and +0.16 (remainder) which although statistically significant at the 5% level only yields a very small effect size.
60 65 70 75 80 85 90
% max score
Mot (D) Mot (S) Chi (D) Chi (S) Eng (D) Eng (S) Math (D) Math (S)
Figure 10.1 Attitudes of Cohort 1 (disadvantaged (D) v standard (S) schools)
P1 P2 P3
10.10 For Cohort 2 data was collected from the beginning of the P1 year so that comparisons could be made through P1 and P2 while the pupils were in small classes. By the end of the P1 year English scores in the disadvantaged schools were higher, mainly due to the boys’ contributions. Boys had the lowest attitudes towards English at the start of P1 but had caught up by the end of the year. By the end of P2 the trend for English has been maintained but the attitudes to Chinese have become negative in disadvantaged schools mainly because the girls’ attitudes decreased faster than those of the boys. These differences were all significant at the 1% level although with small effect size coefficients in all cases. Examining the residuals using the combined learning orientation measure showed very little change overall. For the P1 year the value for the disadvantaged schools was -0.01 against zero for the remainder.
For the change during the P2 year the corresponding figures were -0.02 and zero respectively. Thus taking the two cohorts together the smaller classes would appear to have a minimal effect on pupils’ motivation and self esteem or on the attitudes to the three core subjects. When gender differences are examined therefore the results lead to an interesting, if tentative conclusion.
Although generally boys have poorer attitudes towards English, being in a small class in a disadvantaged school brings a slight improvement. In Chinese disadvantaged classes, although the attitudes of boys decline, those of the girls drop at a faster rate, which is again contrary to the usual trend. These attitude shifts coupled with the slight improvements in boys’ attainment suggest that small classes are benefiting these disadvantaged male pupils although in every case the effect sizes are small.
60 65 70 75 80 85 90
% max score
Mot (D) Mot (S) Chi (D) Chi (S) Eng (D) Eng (S) Math (D) Math (S)
Figure 10.2 Attitudes of Cohort 2 (disadvantaged (D) v standard (S) schools) start P1 end P1 end P2