Scottish Government

APPENDIX 2: SAMPLES

A2.1 The analyses compare the participation and attainment in national qualifications of young people in the Phase 2 EMA Pilot areas with that of young people in other areas of Scotland who attended schools with similar characteristics to those in the EMA pilot areas.

The Phase 1 comparative Sample

A2.2 The sample consisted of young people who attended nine schools in East Ayrshire, and nine schools selected from North and South Lanarkshire. Sample details were provided by the local authorities from SEEMIS, and were linked to the SQA data. Table A2.1 compares the characteristics of the sample in the Phase 1 pilot and control areas.

Table A2.1: Characteristics of comparative sample in Phase 1 pilot and control areas

The Phase 2 comparative sample

A2.3 It was more difficult to select a "control group" of schools comparable to those in the Phase 2 EMA Pilot areas, because the extreme levels of deprivation and low levels of attainment found in Glasgow schools are not matched elsewhere in Scotland. The local authorities selected to pilot the EMA were deliberately chosen on the basis of high levels of poverty and low educational attainment. Although we have chosen a control sample to mirror as closely as possible the characteristics of the EMA areas, we should be aware that the EMA areas include areas of more extreme deprivation and we would therefore expect the EMA sample to have lower levels of educational participation and attainment than the control sample.

A2.4 There are 48 schools in the Phase 2 EMA Pilot areas, 31 in Glasgow, 10 in Dundee and 7 in West Dunbartonshire. A "control group" of 48 comparator schools was selected to match as closely as possible the schools in the EMA pilot areas in terms of the following characteristics:

• % of students living in low SES areas (based on MOSAIC classification),
• mean area deprivation based on the Scottish Executive's deprivation index,
• % with free meal entitlement (an indicator of low household income),
• proportion with low S4 Standard Grade attainment,
• average number of credit-level awards at S4 Standard Grade,
• S4 Standard Grade point score.

A2.5 The sample was selected from a dataset of all state-funded secondary schools, except those in East Ayrshire, and the Highlands and Islands. We ranked the relative disadvantage of schools on each of the measures listed above, and calculated a final rank on the basis of average rank on all variables. Using the final rank we examined the distribution of schools with disadvantaged intakes (Table A2.2).

Table A2.2: Numbers of secondary schools in phase 2 EMA pilots and other EAs with disadvantaged/advantaged intakes

A2.6 We find that of the 66 Scottish schools with the most disadvantaged intakes, half are in the Phase 2 EMA pilots, and of these 25 are in Glasgow. A smaller proportion of the next most disadvantaged category is in the EMA pilots (7/65). There are however, some Phase 2 EMA schools in the average and more advantaged categories. In order to select a sample of non- EMA schools which could be used as a control group for comparison, we included all the schools in the 20% most disadvantaged category. Other non- EMA schools were selected for the control group on the basis of proximity in the overall ranking to the less-disadvantaged EMA schools. The final sample consisted of 48 EMA schools and 48 control schools (Table A2.3).

Table A2.3: Location of schools in the Phase 2 comparative sample

A2.7 Table A2.4 shows the average distribution of characteristics in the Phase 2 EMA pilot and control samples, and highlights the higher levels of area deprivation and free meal entitlement in the EMA pilot sample than the control sample. A slightly smaller proportion of the pilot sample lived in low SES areas (based on the MOSAIC classification), and the average level of attainment in the EMA pilot sample was a little higher than the control sample. We take account of these factors in the statistical models

Table A2.4: Characteristics of comparative sample in Phase 2 pilot and control areas

Note: Based on analysis of variance, ** denotes significant difference at the 99% confidence level