Explanation of our schools data
Home > About

Explanation of our schools data

You may have received a letter containing data about your school. This page contains an explanation of how the figures were created. If you have any questions or require further information please feel free to contact us.

 

Data sources

This modelling is based on the Department for Education Key Stage 5 destination data. We use the revised version of this data, released on 24 January 2019 titled “Destinations of KS4 and KS5 pupils 2017: key stage 4 institution level tables (revised)”.1 These data relate to students who were deemed to be at the end of key stage 5 in the 2015-16 academic year and therefore progressed to university or other destinations in the 2016-17 academic year. Data on more recent cohorts was not available at the time of modelling. We match this data set with Department for Education Key Stage 5 attainment data for the 2015-16 academic year.2 Our modelling is restricted to state-funded mainstream schools and colleges only.

Handling of the destination data

There are 2,380 schools/colleges in the destination data for the year in question. For each institution, there is a breakdown of statistics including:

> The number of Key Stage 5 leavers in 2016-17 (the ‘KS5 cohort’).
> The percentage of those Key Stage 5 leavers progressing to a ‘top third’ university.3

For 402 of these schools/colleges, the data on progression to top third university is redacted to protect the privacy of students. This happens where either there are fewer than 11 students in cohort or the outcome in question relates to 1-2 students. Where the cohort contains fewer than 11 students, we drop them for the purpose of our modelling as we have no way of knowing what value the ‘progression to top third’ variable should take for these schools/colleges. We retain 2,338 rows in our data.

For all the other redacted rows, we know this relates to either 1 or 2 students progressing to a top third university. We make the most optimistic assumption possible - that these rows relate to 2 students - and then turn this back into a percentage by using the KS5 cohort as the denominator. In our final destination data, we have 2,338 rows.

Handling of the attainment data

There are 3,246 schools/colleges in the attainment data for the year in question. For each institution, there is a breakdown of statistics including:

> The average point score for each academic qualification entered.
> The percentage of A level students achieving at least three levels at grades AAB or better, at least two of which were in facilitating subjects.

Note: The facilitating subjects are: Biology, Chemistry, Physics, Mathematics, Further
Mathematics, Geography, History, English Literature, and Classical/Modern
Languages. These are the subjects which are most often required for entry into selective universities.4

Again, some of this data is coded as NE (which means there were no relevant entries) or SUPP which means the data is suppressed because there are 5 or fewer pupils or students covered by the measure. We replace the NEs with zeros and the SUPPs with N/A.

Modelling with the destination and attainment data

All 2,338 rows in the destination data merge successfully with the attainment data. We know that demographic factors (such as the proportion of disadvantaged students in a school/college) are highly predictive of progression to selective universities. However, we are primarily interested in the relationship between attainment and progression, so we model progression to top third universities as a function of:5

> Average point score per academic entry
> Percentage of A level students achieving at least three levels at grades AAB or better, at least two of which are in facilitating subjects

When we drop rows with missing data for these variables, 2,064 rows remain.

Using our model, we want to identify schools/colleges which performing worse than we would expect. To this end we:

> Calculate the expected rate of progression top third universities based on the attainment variables in the model.
> Calculate what one additional student going to a top third university would equate to in percentage point terms.
> Identify the schools where the differences between the expected and the actual rate (in percentage points) is bigger than the equivalent of one student going to a top third university.

There are 873 schools/colleges which meet these criteria. Of these, 24 are selective and we drop them from our sample, leaving a total of 849 schools where the actual rate of progression is lower than the expected rate of progression.

 

[1] These data can be downloaded via the Department for Education website: https://www.gov.uk/government/statistics/destinations-of-ks4-and-ks5-pupils-2017

[2] These data can be downloaded via the Department for Education website: https://www.gov.uk/school-performance-tables

[3] The top third most selective universities are defined using the mean UCAS tariff score from top 3 A level grades of entrants. You can find out more on the DfE website: www.gov.uk/government/collections/statistics-destinations

[4] The accountability measures are formally defined here: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/746566/16-18_Accountability_Measures_Technical_Guide_Oct18.pdf

[5] Note: that including the number of students at the end of 16-18 study who entered for at least one advanced (level 3) academic qualification does not add any additional explanatory power so we do not use this variable.