Scottish prison population statistics technical manual

Information on data presented in the Scottish Prison Population Statistics reports from 2019-20 onwards.


Data Quality Assurance at SPS

PR2 is the sole management information system used by SPS in managing individuals in custody.  It is of critical operational importance to ensure the correct housing, management and release of individuals held in prison establishments.  Internal training and quality assurance are used to maintain accurate prisoner records for these purposes.

Prison officers are given training on the use of PR2 when they attend the SPS College (SPSC) prior to being deployed operationally. The training focuses on the aspects of PR2 the officers are most likely to use, depending on whether they are Operations (i.e. security, visits, movements) or Residential (i.e. cell sharing risk assessments, integrated case management).  The importance of accurate information being inputted is stressed during this training.

Each prison establishment has a Business Improvement Manager (BIM) who runs regular business objects reports from PR2 to check the accuracy of information held. Database Amendments are sent to the central Digital Solutions Team to correct any errors in the data or Database Administer is used to correct within the database itself. Validation controls also exist within PR2 to ensure data accuracy. Where possible, users choose from a pre-populated list of values instead of entering free text.

Accredited Official Statistics Assessment

The Scottish Prison Population Statistics are currently undergoing assessment by the UK Statistics Authority for compliance with the standards required of National Statistics. The interim result of this assessment by the Regulator is available from the UKSA website.

Requirement 2 from this initial assessment requires a fuller explanation of the quality  of data collected for operational purposes in prisons as they pertain to the production of Official Statistics. To that end the prison statistics team has conducted a Quality Assurance of Administrative Data exercise, the results of which are summarised in the next section.

Quality Assurance of Administrative Data (QAAD) Findings

A review of the operational context for the collection, processing and compilation of data from the prisoner records system, PR2, was conducted in March-May 2024. In line with guidance from the Office for Statistics Regulation, four practice areas were examined for compliance with the standard:

  • Operational context of administrative data collection
  • Quality Assurance principles, standards and checks by data suppliers
  • Communication with data supply partners
  • Statistics Producers' Quality Assurance investigations & documentation

In the following section we examine the first two of these practice areas: the operational context and data standards applied by SPS in populating the prisoner records system database.

This section of the review involved interviews with data owners and operational experts on how prison staff interact with the system, how there inputs are validated and reviewed and institutional safeguards and contraints related to data quality. For this exercise the information used in the construction of the cellWise data and the annual accredited official statistics was the sole focus, but conclusions drawn about these data apply equally to other statistical publications drawn from the same source.

Data sources

The quality of the resulting data maybe linked to source that information is drawn from:

  • For internally generated information like index reference numbers and date-time information, the values are generated automatically as a result of system events. Such events may require additional pieces of information to be submitted by the operator at the same time, for example providing detail of a new cell being allocated or the grounds for a liberation.
  • Transcribed information is provided in documentation from external sources, the most common being warrants for imprisonment generated by Courts. This information is typically entered by hand by the operator. This may be verified against the retained paper copy of the warrant.
  • Subject-Reported information is provided directly by the person in custody. This information typically relates to issues of their person circumstances and identity – addresses, ethnicity, religion etc. can fall under this category.

Collection Environment

The circumstances under which data relating to this source are entered into the prisoner records system include:

  • During admission, when one or more people arrive at reception. Legal documentation is presented to the receiving staff and transcribed into the system. The person arriving is asked a number of questions about their identity which are also logged in the system.
    Reception can be a very busy, noisy and stressful environment, particularly when a large number of people arrive at once, for example from court custody.
  • During induction. People arriving to prison are given a one-to-one induction meeting with prison staff shortly after their arrival – usually no more than a day or two after reception. Any information recorded at admission can be checked and amended in this meeting, and any missing information can be requested.
  • During normal operations: if the circumstances of a person in custody change over the course of their stay, prison officers may amend the information held about them on the system – usually when this information is volunteered by the data subject or when documentation of status changes are required.

Input Validation

Constraints on the type of information that may be entered into specific fields improves the accuracy of the information recorded, by preventing mis-spellings, invalid entries and missing values. Such input validation controls are used for some parts of the prisoner records system, and take a variety of forms depending on the data type being entered:

  • Automatic generation: data items are generated automatically due to system activity. These are typically date-time values or index numbers to identify records. As values are logged when a system change occurs these need no further validation, although in some cases revisions are required where input is delayed.
  • Input value list: the inputs permitted are restricted to a list of valid values. This includes an external address database used to populate address fields, as well as categorical fields with a set list of valid responses.
  • Free text: In some cases such constraints on data entry are not appropriate. These may be numerical values, like the length of a sentence, or non-standard format identifiers like names and certain justice-system codes.

Quality Assurance of retained values

The extent to which information is quality assured operationally depends on its history and utility in the prison estate and the wider justice system.

  • Upstream: the quality of the data is assumed based on its source and the number of administrative checks applied before it arrives to the prison data systems.
  • Operational: inaccuracies in data entry can lead to real-world impacts. In particular, the mis-assignment of cell address would lead to the inoperability of in-cell telephones and is therefore typically quickly corrected. Other information QA’d in this way may be updated by officers in light of risk assessments, improved sources or data subject-led amendments.
  • System Administrators (Business Information Managers, or BIMs) carry out routine checks of new information entered on the system by operators and can check, clarify and amend that information via a standardised process.
  • Third Party checks occur when users other than those entering or administering prisoner data need to use the information in the course of business. Such agents include Case Workers, Immigration Services, Justice Social Work agencies, etc. Any required amendments can be sought via System Administrators.
  • None: in a few cases the information gathered has no direct operational value, and is used solely for reporting on the population. In such cases, Scottish Government analysts using such data are required to verify unexpected values with system administrators before it can be used in Official Statistics.

Prison Statistics Variables Quality Matrix

The following table provides an overview of the analytical variables used in the Official prison population statistics based on the categories ascribed to their input data from the sections above.

Table 4.1: QAAD Data Quality Information by Output

Output

Variables

Data Source

Collection Environment

Input Validation

Quality Assurance

CellWise Spine

  • Individuals
  • Prisoner days

Prisoner Number

Generated

Admission

Automatic generation

Operational/ System Admin

 

Establishment Name

Generated

Admission

Automatic generation

System Admin

  • arrivals

Cell Allocated Date

Generated

Operational

Automatic generation

Operational

  • departures

Cell Vacated Date

Generated

Operational

Automatic generation

Operational

Liberations

Liberation Date

Generated

Operational

Automatic generation

System Admin

 

Liberation Type

Generated

Operational

Input value list

System Admin

 

HDC outcomes

Generated

Operational

Input value list

Third Party

Prisoner characteristics

Age

Transcribed (DoB)

Admission/ Induction

Automatic generation

Upstream

 

Disability

Subject-reported

Admission/ Induction

Input value list

Operational

 

Ethnicity

Subject-reported

Admission/ Induction

Input value list

None

 

Gender

Subject-reported

Admission/ Induction

Input value list

Operational

 

Marital Status

Subject-reported

Admission/ Induction

Input value list

None

 

Sexual Orientation

Subject-reported

Admission/ Induction

Input value list

None

 

Veteran status

Subject-reported

Admission/ Induction

Input value list

None

Geographic Data

Local Authority

Subject-reported

Admission/ Induction

Input value list

Upstream/
Third Party

 

Deprivation

Subject-reported

Admission/ Induction

Input value list

Upstream/
Third Party

 

No Fixed Abode status

Subject-reported

Admission/ Operational

Input value list

Third Party

Warrant Spans

Legal Status

Transcribed

Admission/ Operational

Input value list

System Admin

Data relating to

Index Offence

Transcribed

Admission/ Operational

Input value list

System Admin

 

 

 

 

Free Text

System Admin

Data relating to

Index Sentence

Transcribed

Admission/ Operational

Input value list

System Admin

 

 

 

 

Free Text

System Admin

 

 

 
Back to top