Official Statistics

Recruitment diversity statistics June 21

Published 19 August 2021

Applies to England and Wales

HMPPS Recruitment Diversity Statistics Experimental Official Statistics Quarterly estimates: 1 July 2019 to 30 June 2021

Feedback on this experimental Official Statistics release should be sent to: [email protected]

Main Findings

Around one-fifth of applicants in the period Q3 2019 to Q2 2021 were from BAME backgrounds	Between 1 July 2019 and 30 June 2021, 19.8% of Prison Officer applicants and 19.1% of Operational Support Grade (OSG) applicants were from BAME backgrounds, compared to 21.4% of all HMPPS candidates
Around five percent of applicants in the period covered by this report had declared themselves disabled	For the period covered, 3.6% of Prison Officer applicants and 4.5% of OSG applicants declared themselves as disabled, compared to 5.6% of all HMPPS candidates
Around half of the applicants in the period Q3 2019 to Q2 2021 were female	For the period covered, 45.6% of Prison Officer applicants and 51.9% of OSG applicants were female, compared to 55.8% of all HMPPS candidates
Among Prison Officer and OSG applicants, there is evidence of a disparity in outcomes when comparing BAME applicants to white applicants	The ethnicity Relative Rate Index (RRI) was 0.58 for Prison Officer applicants and 0.69 for OSG applicants for the 2 years to June 2021 suggesting a disproportionately high number of white applicants completed the application process compared to BAME applicants
There is evidence of a disparity in outcomes when comparing female applicants to male applicants, for Prison Officer but not OSG candidates	The RRI for gender was 0.71 for Prison Officer applicants, but 1.15 for OSG applicants for the overall period covered by this report suggesting that the success of rate of male applicants was higher for Prison Officer roles but higher for female OSG applicants
There is evidence of a disparity in outcomes when comparing disabled applicants to non-disabled applicants for OSG candidates and Prison Officers	The disability RRI was 0.70 for Prison Officer applicants and 0.76 for OSG applicants suggesting that the rate of success for disabled applicants was lower than for non-disabled applicants
The proportion of HMPPS applicants accepting a final offer from BAME backgrounds was consistently below the target of 14%	In the two years to June 2021, 12.9% of successful applicants were from BAME backgrounds

Background

• Following the Lammy review ^{[footnote 1]}, HMPPS made a public commitment that 14% of all new recruits will come from BAME backgrounds by December 2020.
• As at 30 June 2021, 10.5% of all HMPPS staff declared themselves to be from a BAME background. On the same date, 8.6% of all Public Sector Prisons staff, 15.9% of all YCS staff and 15.6% of National Probation Service staff were from a BAME background.
• HMPPS recruitment campaigns are run in different parts of the country, recruiting from local populations, at different times. It should therefore be expected that the percentage of BAME applicants will fluctuate over time.
• The Lammy review also recommended that new data should be collected and published with a full breakdown by ethnicity. This experimental statistical release sets out to meet that recommendation.

As these are experimental statistics, we would welcome feedback as to how useful they are, whether different analysis would be preferable, or any other comments about them. If you wish to send any views you may have about these experimental statistics, please use the contact details at the end of this bulletin

Following an internal review of the content of this annex, it was decided to include additional information about recruitment by stage, across the whole of HMPPS. A cohort approach is used in tables 1-3, covering Prison Officer and OSG campaigns, with the progress of people through all later stages of the recruitment process shown in the quarter the candidate applied in, regardless of when the subsequent stages happened. For example, if a candidate applied in March 2020 their progress would be represented in Q1 2020 in the data tables, even if the later stages happened in Q2 or Q3 2020. This allows the comparison of the proportion of candidates with protected characteristics across each application stage. In tables 4-7, covering all HMPPS recruitment campaigns, a staged approach is used, with each stage of the application processes being recorded in the quarter that it took place even if that is not the same as the quarter of the initial application. This means that the proportion of applicants with protected characteristics in each stage can be compared over time.

Statistics are shown for the following stages of the application process, for campaigns that were run between 1 July 2019 and 30 June 2021:

• applications;
• invitations to assessment/interview;
• provisional offers; and
• acceptances of a formal offer.

There is a particular focus on the application process for Prison Officer roles and OSG roles but statistics are presented covering all HMPPS recruitment campaigns in the period.

These experimental Official Statistics are used to monitor and evaluate the diversity of applications and appointments. These are newly developed statistics and are currently considered experimental so that users and stakeholders can be involved in the assessment of their suitability and quality.

Recently, there have been a number of recruitment stages that have changed, mainly in the intermediate stages of the recruitment process, such as more roles moving towards using assessment centres. It may not always be possible to capture these updates in the data straightaway which may result in an undercount of candidates at some stages. It is therefore recommended that the statistics in this annex are treated as estimates.

For tables 1 to 3, it is important to note that some candidates are still working their way through the recruitment pipeline as of 30 June 2021. Where this is the case, the candidate is not included in this release.

In the next release of these statistics, numbers will be updated if any information has been recorded to reflect candidates’ further progress. This means that figures shown for the more recent quarters are currently based only on a small proportion of the eventual final data, particularly for the later stages of the process. As the data is updated in future publications there could possibly be changes in the proportions and Relative Rate Index values used to compare the rate of success between groups. For this reason, the figures for the stages after application for the latest two quarters have not be shown in Tables 2 and 3.

In some campaigns a merit list may be held, whereby candidates who are successful at assessment/interview are put onto a waiting list until a suitable vacancy opens up. Where this is the case, a candidate may not be offered a post until sometime after the initial application and assessment/interview stages.

The diversity of candidates is monitored at various stages throughout the recruitment process. To assess the relative levels of success we examine the progress of all the candidates who have applied for posts on a quarterly basis. Candidates may apply for multiple roles but only one application per candidate per quarter is included in the analysis, and the application that is included is the one that made it furthest through the process.

For the campaigns included in this bulletin we examine the volumes of candidates who have progressed through each of the following stages of the recruitment process:

HMPPS Recruitment Stage	Description
Application	The candidate submits an online application for a role in HMPPS. Depending on the role applied for, there are varying processes to determine if a candidate is invited to an assessment. In the case of Prison Officer recruitment, there are online tests, including mathematical proficiency and situational judgement. For OSG recruitment, a manual sift is performed.
Applicants invited to assessment	All candidates who pass the application stage process will be invited to attend an assessment relating to the role applied for. For Prison Officer candidates, the assessment day consists of a literacy and language test, a fitness test, role-playing tests with actors playing the part of prisoners, and retake the situational judgement and mathematical skills test. For OSG candidates, an interview is conducted.
Applicant made provisional offer	Candidates who have been made a provisional offer will now undergo pre-employment checks including security vetting.
Applicant accepts formal offer	Candidate has confirmed acceptance of Formal Offer and this has been recorded on the recruitment system.

Representation percentages

Within the bulletin, representation percentages are presented for the following stages of the recruitment process:

• Initial applications;
• invitations to assessment/interview;
• provisional offers; and
• acceptances of formal offer.

The diversity statistics in this report are based on self-declared information that applicants have provided on the Oleeo recruitment system. Completion of the data on Oleeo is voluntary, and the information is not considered during the recruitment process. The declaration rates are typically high at above 98 per cent. Some applicants choose not to declare their ethnicity, disability status or gender, or do not complete the information – these applicants are grouped together and included in the ‘Unknown’ group. As this is not a meaningful category those who do not declare their information are not included within the analysis.

The declaration rate is calculated as the total number of valid declarations divided by the total number of cases. Where the declaration rate of a diversity characteristic is in excess of 60 per cent, the representation rate is also presented; this is calculated as the known declarations from the particular group divided by the total number of valid declarations of the characteristic. This represents the best estimate of the true level of representation among all candidates. To date, declaration rates in this data have far exceeded the 60 per cent threshold to allow us to present representation percentages for every stage in the recruitment process. However, it should be noted that for some campaigns in some quarters, the numbers involved can be very small and so any interpretation of them should be treated with caution. In addition, where there are two or fewer individuals in a cell in the accompanying Excel tables, these numbers have been suppressed to avoid disclosure.

Representation percentages allow comparison of the diversity proportions at the various recruitment stages outlined above. This is particularly useful for the application stage, as it provides a picture of the diversity of the pool of applicants, and how closely they represent the general population, or the diverse make-up of a local population where a Prison Officer and OSG recruitment campaign is localised to a particular geographic area. It is also particularly useful at the formal offer stage to illustrate any differences in the diversity proportions at the final stage. However, representation among those receiving a formal offer reflects both the representation among applicants and rates of success for each group at the preceding stages.

To make valid comparisons across time or across different groups requires a measure of disparity of outcomes on a standard scale. This standardised measure of disparity of outcomes is described as the Relative Rate Index (RRI).

Further details on the RRI can be found in the Further Information section at the end of this publication. Caution should be used interpreting these figures as in many cases they may not be statistically significant, particularly where they are based on very low number of applicants. Statistically significant values are indicated in the tables.

Detailed Results

COVID-19 had a significant disruption on the recruitment process in HMPPS, particularly for Prison Officer recruitment during part of 2020. This manifested in two ways – firstly in the number of campaigns and therefore applicants in Q2 and Q3 (only 622 and 3,632 applicants for Prison Officer posts in Q2 and Q3 2020 respectively, compared to over 10,000 in the previous four quarters); and secondly in affecting the subsequent stages after application.

SECTION 1: Prison Officer and Operational Support Staff (Tables 4 to 7)

These tables use a cohort approach with the progress of people through all later stages of the recruitment process shown in the quarter the candidate applied in, regardless of when the subsequent stages happened.

Prison Officers (Summary Table 1a and Table 2)

Figure 1: BAME representation for Prison Officer candidates at each stage of the application process, Q3 2019 to Q2 2021

BAME candidates made up 19.8% of all Prison Officer applicants, and 12.6% of formal offers accepted between July 2019 and June 2021. The statistically significant RRI value of 0.58 suggests that BAME candidates had a disproportionately low success rate compared to white candidates. This was mainly driven by a large proportion of BAME candidates not reaching the assessment stage (RRI of 0.65^{[footnote 2]}).

Figure 2: Disability representation for Prison Officer candidates at each stage of the application process, Q3 2019 to Q2 2021

Disabled candidates made up 3.6% of all Prison Officer applicants, and 2.6% of formal offers accepted between July 2019 and June 2021. The statistically significant RRI value of 0.70 suggests that suggests that disables candidates have a lower success rate than non-disabled candidates. The stage that causes the largest disparity is between being made a provisional offer and accepting a final offer (RRI of 0.78^{[footnote 2]}).

Figure 3: Female representation for Prison Officer candidates at each stage of the application process, Q3 2019 to Q2 2021

Female candidates made up 45.6% of all Prison Officer applicants, and 37.2% of formal offers accepted between July 2019 and June 2021. Relative comparison of the overall rates for the 24-month period by gender gave a statistically significant RRI value of 0.71 indicating females have a lower rate of success compared to males through the recruitment process. It is not clear which stages are driving this disparity.

Operational Support Grades (Summary Table 1b and Table 3)

Figure 4: BAME representation for Operational Support Grade (OSG) candidates at each stage of the application process, Q3 2019 to Q2 2021

For the period July 2019 and June 2021, BAME candidates made up 19.1% of applicants and 14.0% of formal offers accepted. Comparison between BAME candidates and white candidates gives a statistically significant RRI value of 0.69 showing that BAME candidates have a disproportionately lower success rate. This was shown to be driven by the low proportion of BAME candidates being invited to assessment following an application.

Figure 5: Disability representation for Operational Support Grade (OSG) candidates at each stage of the application process, Q3 2019 to Q2 2021

Disabled candidates made up 4.5% of all OSG applicants, and 3.5% of formal offers accepted between July 2019 and June 2021. The statistically significant RRI comparison of disabled and non-disabled candidates (0.76) indicates that disabled candidates have a lower success rate through the recruitment process. It is not clear which stage has the biggest impact on this disparity.

Figure 6: Female representation for Operational Support Grade (OSG) candidates at each stage of the application process, Q3 2019 to Q2 2021

Female OSG candidates made up 51.9% of all OSG applicants, and 55.4% of formal offers accepted between July 2019 and June 2021. The comparison of male and female candidates gives a statistically significant RRI value of 1.15. This indicates that female candidates have a higher success rate through the recruitment process though it is not clear at which stage this mostly occurs.

SECTION 2: Overall HMPPS Recruitment by Sector (Tables 4 to 7)

These tables use a staged approach with each stage of the application processes being recorded in the quarter that it took place even if that is not the same as the quarter of the initial application.

HMPPS recruitment (Table 4)

In the last two years to June 2021, there has been 396,029 applicants to HMPPS recruitment campaigns, 89,380 applicants were invited to an assessment, 20,814 were made a provisional offer and 16,779 accepted a formal offer.

In this time, there were 83,437 BAME applicants making up 21.4%. The proportion of BAME applicants has steadily increased from 19.1% in Q3 2019 to 22.5% in Q2 2021. However, there were 2,114 BAME applicants to accept a formal offer at 12.9% of the total, ranging over the period from 12.3% in Q3 2019 to a high of 13.8% in Q2 2020.

Figure 7: BAME representation HMPPS candidates at each stage of the application process, Q3 2019 to Q2 2021

During the period, there were 21,959 disabled applicants to HMPPS recruitment campaigns. This amounted to 5.6% of all applicants, increasing from 4.9% in Q3 2019 to 7.2% in Q2 2021. There were 813 disabled applicants to accept a final offer, 5.0% of the total, ranging from 3.9% in Q3 2019 to a high of 6.2% in Q4 2020.

There were 219,406 female applicants to HMPPS recruitment campaigns. This amounted to 55.8% of all applicants, ranging from 51.3% in Q1 2020 to 64.0% in Q2 2020. There were 9,037 female applicants to accept a final offer, 54.5% of the total, ranging from 49.5% in Q4 2019 to a high of 59.1% in Q4 2020.

Breakdowns for PSP & YCS, Probation and HQ & Area Services are provided in Tables 5 to 7.

Further information

Technical information

The diversity statistics in this report are based on self-declared information that applicants have provided on the Oleeo recruitment system. Completion of the data on Oleeo is voluntary. Some applicants choose not to declare their details or do not complete the information – these applicants are grouped together and included in the ‘Unknown’ group. This is not a meaningful category, and accordingly those who do not declare their information cannot be included within analysis of the outcomes.

Any records downloaded from the Oleeo system with a time stamp after 23:59 on the last day of December 2020 have been removed. Some records do not have a timestamp for some of the recruitment stages. Where this is the case, we have assumed that where a candidate has reached a stage, he or she passed all the previous stages, even if they don’t actually have the relevant timestamps for all those earlier stages.

A single candidate can make multiple applications. Where this is the case, duplicates have been removed to leave a single instance of each candidate in each quarterly cohort (an individual can appear a maximum of 6 times in the table). The application which got furthest in the process is selected. If there are several applications from different quarters at the same stage, the earliest submission is selected.

Within the Prison Officer applications, the vast majority of roles are advertised as “Prison Officer”; however, recently there have been a small number that are advertised as “youth officer”. This is a Prison Officer role that aimed particularly at Prison Officers who will work with juvenile offenders in the YCS. This Youth Officer role has been included with Prison Officers in this publication.

Relative Rate Index (RRI)

The RRI, or Relative Rate Index^{[footnote 3]}, gives a standardised measure of difference between groups. This has been adopted in line with the Judicial Appointments Commission use of the RRI for presenting similar data, which was reviewed by MoJ statisticians in February 2020^{[footnote 4]}, and the RRI was widely used within the Lammy Review^{[footnote 1]}. These statistics use this metric to give a standardised view of any disparity of outcomes on recruitment by diversity characteristic.

The RRI is calculated by dividing the rate of success (i.e. the percentage of those who applied that accepted a formal offer) for one group by the rate of success for the other group with a diversity characteristic, thus creating a single standardised ratio measure of relative difference in outcomes between those two groups. If the rates were the same, the result would be a value of 1, which would indicate identical rates for both groups (that is a parity of outcomes). Deviation from 1 in either direction suggests a difference in the rates of outcomes.

Considering differences in the two success rates on a relative basis on a standard scale, enables direct comparison of any disparity of outcomes between groups across exercises and across time. By convention, the RRI has been calculated with the under-represented group as the numerator, with the baseline reference group as the denominator. However, calculation in the reverse direction is equivalent and would result in the same interpretation.

When comparing rates and the RRI, it is important to consider uncertainty, to avoid the over-interpretation of fluctuations in outcomes that may be the result of chance alone. Tests of statistical significance have been conducted on the estimates in this publication that are based on the full 24-month period in order to ensure the interpretation of the RRI values take full account of the probability that the observed outcome may be the result of chance alone. Where a RRI is significant at the 95% confidence level, this has been denoted against the RRIs in Table 1 with a ‘**’. For the intermediate RRI values, significance has been determined using the ‘4/5th rule of thumb’^{[footnote 2]}.

In addition to consideration of the extent to which chance variation is involved in any apparent difference in the rates, it is important to consider what is known as practical significance– whether the actual magnitude of the apparent difference in the rates is sufficient to have a material impact, or whether the difference is small and of little material impact. To represent a meaningful disparity of outcomes, an apparent difference must be both statistically significant (unlikely to be the result of chance alone), and of sufficient magnitude to have practical significance.

To further aid interpretation using the ‘4/5th rule of thumb for adverse impact’ ^{[footnote 2]}, RRI values that fall within a range of 0.8 to 1.25 (the zone of tolerance) are not likely to indicate a difference in outcomes resulting in a disparity having practical significance. This sets a range around parity, within which fluctuations at least in part due to natural variation would not be taken as evidence of a disparity of outcomes representative of adverse impact to one group. This does not imply that an RRI falling outside of this range is indicative of the presence of a meaningful disparity.

As such, both statistical significance and practical significance where an RRI falls outside of the range 0.8 to 1.25 should be generally required to establish evidence of a disparity. Though more rigorous significance tests were used on the total RRI values.

Findings suggestive of a disparity of outcome do not necessarily imply issues within the recruitment process and may reflect other differences between groups, such as level of experience. As rates are the combined result of representation among applicants and those accepting a formal offer, it is essential that a disparity of outcomes is interpreted in conjunction with a view on representation among applicants and success at each stage of the recruitment process.

Notes and Conventions

The following symbols are used within the tables in this experimental statistics release:

~	values of two or fewer which have been suppressed
*	percentage suppressed due to small numbers. Where small numbers are present, percentages are highly volatile and potentially misleading.
-	denotes relative rate index not calculated due to suppressed values
..	denotes data not available.

Experimental Statistics

The statistics in this report are classified as experimental Official Statistics. Experimental Official Statistics are statistics that are in the testing phase and not yet fully developed. This report has been designated experimental the methodology has only recently been adopted and the results published. As such, the methods and approach used in this report are subject to modification. More information about the different types of Official Statistics can be found here: https://uksa.statisticsauthority.gov.uk/about-the-authority/uk-statistical-system/types-of-official-statistics/

Feedback relating to the content of this release should be sent to: [email protected]

Related statistics

These experimental Official Statistics have been published alongside Her Majesty’s Prison and Probation Service (HMPPS) Workforce Statistics Bulletin, as at 30 June 2021.

Contact

Press enquiries should be directed to the Ministry of Justice press office:
Tel: 020 3334 3536
Email: [email protected]

Other enquiries about these statistics should be directed to:

Kieran Maclean
Data and Evidence as a Service
Ministry of Justice
10 South Colonnade
London
E14 4PH

Email: [email protected]

Next update: 17 February 2022 URL: www.gov.uk/government/collections/national-offender-management-service-workforce-statistics

© Crown copyright
Produced by the Ministry of Justice

Alternative formats are available on request from [email protected]

1. https://www.gov.uk/government/organisations/lammy-review ↩ ↩²

2. RRI values falling outside the range 0.80 to 1.25 are outside the range within which the magnitude of the difference in rates would not normally represent evidence of a disparity of outcome. Dan Biddle. Adverse Impact and Test Validation: A Practitioner’s Guide to Valid and Defensible Employment Testing. Aldershot, Hants, England: Gower Technical Press. pp. 2–5. ISBN 0-566-08778-2. Toward a Coherent Test for Disparate Impact Discrimination: Peresie, J.L. 2009 ↩ ↩² ↩³ ↩⁴

3. Altman DG (1991) Practical statistics for medical research. London: Chapman and Hall ↩

4. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/869028/Methodology_Review_Report.pdf ↩