Notes on: Strand, S.  (2012): The White British–Black Caribbean achievement gap: tests, tiers and teacher expectations, British Educational Research Journal, 38:1, 75-101.   http://dx.doi.org/10.1080/01411926.2010.526702 This is based on the LYSPE 1 and addresses the long-standing concern about the educational attainment of minority ethnic kids. This one reviews national test data at age 7, 11 and 14 and public examinations at age 16. They show that attainment of Black Caribbean Black African Black other Pakistani and Bangladeshi groups tends to be below that of White British peers, while Chinese, Indian and Irish (!) pupils score higher than White British, at least in 2006. SES is the most frequently cited explanation, with differences of income poverty generally, and entitlement to FSM specifically. There is other data on occupational status, showing a gradient. However, there is still 'mixed success in accounting for the Black – White [gap]  school with SES measures' (76). There is a typical reduction of this gap but by more than 1/3, and sometimes less, and their own LSYP showed that socio-economic variables 'could account for the Bangladeshi gap and reduce the Pakistani gap by over 80% and the Black African gap by two thirds relative to their White British peers. However, the White British Black Caribbean gap was not reduced' (76 – 7) at mean age 14. There is some success with very young children — a reduction of 40% in kindergarten in maths, and 66% in reading. Parenting practices and home learning environment are also strong predictors especially in the early years. Parents involvement and educational aspirations for their kids become more prominent with older students. Even these, added to additional 'student factors such as attitude to school, academic self-concept, frequency of completing homework, school context and neighbourhood deprivation' was still unable to account for low achievement in this Black Caribbean group, nor to explain why they were the only ones to make less progress than White British students in their first three years of secondary school and indeed fell even further behind'. What about indirect or institutional racism? Gillborn has discussed this in particular and it might appear especially with ability grouping and curriculum tracking. Several US authors have suggested that Black students are disproportionately placed in lower ability groups or tracks and that this produces negative attitudes and behaviours and thus poor attainment [references page 77] however, prior attainment and measured ability also have an effect on the placement in ability groups or tracks — some researchers have reported that ethnic differences disappear after we control for these factors and SES. There is some work in England on the negative effects of ability grouping, but  these are largely 'small-scale ethnographic studies', and the results have been challenged as showing that Black discrimination occurs on any scale (Foster is cited). A tighter focus on enrolment on specific courses seems important for example on lower maths courses during grade 8 which produces fewer opportunities to lead to advanced classes at grade 10. Again evidence is mixed here at least in the US — a longitudinal study there in 2008 shows that Black students are no less likely than White ones to complete higher level maths courses in grades nine and 10 after controlling for prior attainment and engagement. In England we have differentiated test papers, 'tiering' in national test in science and maths 14 and in public exams in a wide range of subjects at 16 — these allow the award of a limited range of national curriculum levels. Teacher judgement is used to assign students to these different tiers, and the higher levels can only be achieved if the teacher has entered the student for the higher tier examination. This is supposed to be more efficient and more positive for the student, but this clearly introduces 'a social dimension to the process and there has been very little research on how this may impact on different ethnic or social class groups' (78). Gillborn and Youdell ( 2008) have suggested that minority ethnic students are less likely to be entered by teachers for the higher test tiers and have found this across a larger sample of 18 secondary schools — but these were selected because Black students were performing below the average anyway so this is not a representative sample and there is no control for students prior attainment — 'i.e. the study is not able to establish bias in secondary school practices in tier allocation' (78). There is so far no study using a large and nationally representative sample exploring all the variables student family school and neighbourhood — except this one. This one asks whether all ethnic groups are equally likely to be entered for higher tier papers in maths and science; whether differential patterns of entry can be explained by prior attainment; whether other factors like student family school and neighbourhood factors, home background, social class, differences in students attitudes and aspirations or motivations can explain these patterns; whether or not there are indications of bias in entry to the higher test tiers and if so what factors might account for it. The data uses Wave 1 of the LSYP in England, 2004. It used a two stage sample, a stratified frame based on school deprivation, region and admissions policy, with probability proportional to size, and then a sample of students from the schools. There were sample boosts for the six largest minority ethnic groups to provide 'representative samples' from each school. They were left with 14,503 students from 629 schools. They did face-to-face interviews with student and with both parents or carers and gathered national test results at 11 and 14. They created 28 variables associated with educational attainment [things like family background, ethnic group, SEC, mothers highest educational qualifications, FSM, home ownership, parental aspirations, resources and involvement, family discord — quarrels with the student — SEN, truancy absence, contact with social services, exclusion, student educational aspirations, homework, academic self-concept, attitude to school, school type, neighbourhood deprivation,pp 96 – 98]. All students complete test in English maths and science at the end of year nine and the typical grade at age 11 is expected to achieve level 4 and at 14 five or six. The highest level that can be achieved in English and science is seven, and eight in maths. Science is available in two tiers, the first one leading to levels four and five. If a student entered for the higher tier fails to achieve level 5 there is a lower chance of getting a four and the student risks being graded  unclassified. The professional judgement of the teacher decides and this is 'influenced by the teacher's perceptions of how students will cope with the demands made on them' (80). The tiering structure for maths is even more complex. There are four tiers, leading to levels four, five, six, and seven, again with the risk of a U grade 'if a student entered for a higher tier fails to achieve the expected level' (81) They used logistic regression for the science test and ordinal regression for the maths tests to identify the 'unique (net) contribution of particular factors to variations in the tier of entry while other background factors are controlled'. This is to manage prior attainment levels and other socio-economic factors and student factors in particular. So one (first) model includes only ethnic group. If there is disproportionate entry to the higher tier, this still does not indicate the existence of bias because there may be 'actual differences in attainment between ethnic groups' so a second model estimating prior attainment is required and this is done in national English, maths and science tests at age 11. Family background is added in a third model to include matters such as social class, educational qualification, FSM and so on and the effects measured. In the final model all the variables were eligible for inclusion: they all impact independently on attainment, but the less significant ones are 'progressively removed to create parsimonious models' (81). Descriptively, 12% of White British students achieve the highest level in the science test, but only 6% of Pakistani and Black African students and 5% of Bangladeshi and Black Caribbean students. 46% of White British students are entered to this higher tier, 38% are Bangladeshi, 33% of Black African, 28% of Pakistani and 28% of Black Caribbean. We can calculate the odds ratios for different ethnic groups, the chances of being entered for the higher tier relative to the odds for White British students. — Pakistani and Black Caribbean students are only half as likely, Black African and Bangladeshi students more likely but 'significantly underrepresented' (82). We then compare rates to prior attainments as indicated by age 11 average test marks. 'Prior attainment accounts for a substantial proportion of the variation in tier entry' (p. 83) and the odds ratios for 'Pakistani, Bangladeshi and Black African students are no longer significantly different from White British students, suggesting the tier entry decisions are broadly in line with students prior attainment'. However for the Black Caribbean students, the odds ratio only rises ( to 0.66 to 1) which still means they are significantly less likely to be entered for the higher tier than White British students of the same prior attainment [ orig emphasis]'. Even after including family background, Black Caribbean students still continue to be underrepresented, with a slight increase in their odds ratio, and the full contextual model revealed that other variables were associated, such as gender, or having mothers with a degree, or coming from higher and lower managerial and professional homes, having parents actively involved with the school who monitored their children and had higher educational aspirations, students who completed homework regularly and had high academic self-concept, had not truanted or been involved with the police, excluded from school or lived in a high deprivation neighbourhood. All these were 'statistically significant' but still 'explain relatively little additional variance' over prior attainment, and still did not account for the underrepresentation of Black Caribbean students in entry to the highest tier. It is still the case that for every three White British students entered for the higher tier only two comparable Black Caribbean students are entered (odds ratio 0.64 to 1). Pakistani students also appear to be underrepresented 'although to a less marked extent' (84). For mathematics, Black Caribbean students are the lowest attaining ethnic group at age 14 and only 1/3 attain level 6 or above compared to 55% of White British. Pakistani 38% Black African 39% and Bangladeshi students 40%. Black Caribbean students are substantially underrepresented at 25% — 'more extreme than for any other ethnic group' (84). Again they produced a base model, and then systematically included the variables. Prior attainment showed that age 11 test marks were 'strongly correlated with tier of entry, with odds ratios the same for Pakistani and Bangladeshi students as for White British students, and actually better for Black African and Indian students.. Black Caribbean students are only two thirds as likely to be entered for higher tiers as White British students 'with the same age 11 mathematics test score' (85). Adding family background shows that 'Pakistani and Bangladeshi groups joined the Black African and Indian groups in being overrepresented in the higher tiers after accounting for their high level of socio-economic disadvantage'.  Black Caribbean students only improve their odds ratio  to 0.72 to 1. In the full contextual model, boys gain an advantage over girls, as do students in the higher for social classes, those with mothers with any level of educational qualifications, high parental educational aspirations, greater parental supervision, the provision of home computers and private tuition, high educational aspirations and self-concept, good rates of completing homework, and an absence of negative factors including SEN, absence from school, exclusion, contact with the police and attending high deprivation schools or living in high deprivation neighbourhoods. Again these factors were all significant although with small impact and added only 4.5% of the variance. Even within this there are 'statistically significant and large differences in entry to test tiers for two ethnic groups' (86) — Black Caribbean students are underrepresented relative to White British with odds ratios of 0.65 to 1, and Indian students overrepresented, with an odds ratio of 1.42 to 1 So the results show consistent underrepresentation for Black Caribbean students. It's not a result of their prior under attainment nor differences in gender, social class, maternal education, FSM, home ownership or single-parent households. Nor are factors like exclusion from school, SEN, truancy rates or other family school and neighbourhood factors. Overall 'the evidence points to systematic underrepresentation of Black Caribbean students' especially for tiered maths and science: it is 'substantially smaller [but still there] for the English test which is not tiered' (87). There is more underrepresentation after age 11 for Black Caribbean groups, but not for Black African Pakistani or Bangladeshi groups. This means that there is 'no evidence of bias in secondary schools teachers' allocation of students to tiers for these ethnic groups. However, this is not the case for Black Caribbean students… And the evidence suggests bias in secondary school teachers' allocation of students to tiers' (87). [I think bias refers to a statistical defeinition here] These results may still 'not of themselves demonstrate bias in tier entry decisions'. Real bias might only be established if initial test marks for any Black Caribbean students were higher than those of White British students entered for the same tier, showing that more able Black Caribbean students were held back. If this were so, those able Black Caribbean students should get higher marks within the tier they were entered for. However, there are other variables affecting performance — prior attainment, social class of the home and so on, and these may make it 'unlikely that Black Caribbean students would have a higher mean test mark than White British students within a tier', and anyway they generally get lower mean marks within a tier. There is also 'a more complex relationship between teacher expectation and tiering'. Decisions on allocation are required at least six months before the tests 'and may often be made substantially in advance of this' (88). Students may already be placed in ability groups and perhaps prepared for specific teirs by studying different material. Tiering makes explicit what teachers expect, but this is 'typically revealed well in advance of the test'. So lower marks could actually be a response 'to become demotivated and to try less hard'. Tiering allocations might actually be best seen as illustrating 'wider teacher expectation effects' and need to be put in this wider context. Overall, 'the fact that this underrepresentation in the higher tiers is specific to one ethnic group and persists even after taking account of extensive additional explanatory variables suggests a significant cause for concern' (88) and requires other explanations. Other indications of bias might be found in differences in permanent exclusion rates, presence in school action or SEN programs. Mixed White and Black Caribbean students are twice as likely as White British students to find themselves there or 1.5 times more likely to be identified with behavioural emotional and social difficulties than their White counterparts, 'even after student level controls for age gender entitlement to FSM and neighbourhood deprivation'(89). There is research to suggest that teacher judgements can be 'distorted by affective factors such as perceptions of their behaviour' [references page 89]. Bennett et al. (1993) said that perceptions of behaviour was 'a significant component of their academic judgements' [what was the evidence I wonder?]. If the behaviour of Black Caribbean students is more challenging or 'if teachers perceive their behaviour is more problematic', their academic ability may be underestimated and this is shown by some ethnographic studies in English secondary schools [lots of Gillborn and Rollock]. Gillborn and Youdell also suggests that teachers are risk averse with entry to higher tiers 'reflecting a desire to protect students from failure' and this may apply especially to Black Caribbean students who  were seen as more likely to be disaffected. 'There is a general agreement that Black Caribbean students have the most conflict in relations with teachers… But there are fundamental disagreements about the causes of the behaviour' (89). There may be peer pressure to adopt urban or street subcultures and not 'act White'. Others emphasise greater surveillance at school or 'pre-emptive disciplining' leading to a distinct subculture. 'It is likely that both sets of factors are involved and feed off each other in a vicious cycle of amplification (Pilkington, 1999, page 414)' (90). Teacher grades are multidimensional and reflect judgements of 'effort, participation, attendance and behaviour', and judgements of parents. Test scores may be likely to be less influenced but are not entirely independent. This study does not investigate different teaching groups and their effects or other school effects which may account for achievement gaps. There are implications for assessment policy, including whether or not teacher judgements should be given greater emphasis on assessing levels to enter students.