Programme for International Student Assessment 2025: national report for Scotland

12 Annex A: Technical information on Scotland’s participation in PISA 2025

12.1 Assessment domains and questionnaire participation

PISA 2025 assessed pupils in science, reading and mathematics. Science was the major domain for this cycle, meaning that a larger proportion of assessment time was devoted to science and additional science-related contextual information was collected. PISA 2025 also included an innovative domain, Learning in the Digital World (LDW), which assessed pupils’ ability to learn, problem solve and make judgements in digital environments.

The PISA 2025 assessment used multi-stage adaptive testing in science, reading and mathematics. This means that pupils did not all answer the same questions. Some later groups of questions were selected according to how pupils performed on earlier parts of the assessment, so that pupils could be given questions better matched to their level of performance. This allows PISA to cover a broad range of skills while keeping the test itself to around two hours for each pupil.

In addition to the cognitive assessment, Scotland participated in the following questionnaires:

The Pupil Questionnaire collected information on pupils’ backgrounds, educational experiences, attitudes, wellbeing and aspirations. The School Questionnaire was completed by headteachers or a nominated member of staff and collected information on school organisation, resources, staffing and learning environments.

Scotland also participated in the optional Information and Communication Technology (ICT) Familiarity Questionnaire, which collected additional information on pupils’ access to and use of digital technologies. This questionnaire complements the Learning in the Digital World assessment and supports analysis of pupils’ digital experiences and skills.

12.2 Questionnaire development and quality assurance

PISA questionnaires are developed through a multi-stage international process led by the OECD and its international contractors. Questionnaire content is informed by the PISA questionnaire framework, which is updated each PISA cycle to reflect current educational research, shared policy priorities and the assessment domains being measured.

For PISA 2025, the pupil questionnaire framework comprised 19 thematic modules covering areas such as pupil background, learning experiences, school climate, wellbeing, aspirations and attitudes towards learning. Additional content was developed to support the Learning in the Digital World innovative domain and, as science was the major domain, new questionnaire material was included on topics such as science identity, exposure to science related learning and perceptions of science teaching.

New questionnaire items were reviewed by international experts, tested through pilot interviews with pupils across the OECD, assessed for translation and cultural appropriateness, and reviewed by participating countries and economies. Items were subsequently evaluated through the international field trial before final selection for the main study.

Participating countries and economies were able to adapt questionnaire wording where necessary to reflect national circumstances, provided the intended meaning of questions was preserved. All adaptations were reviewed and approved through OECD quality assurance processes to maintain international comparability.

12.3 Questionnaire response and coverage

Response rates vary between PISA questionnaires. Where questionnaire data cover a lower proportion of the target population, estimates are subject to greater uncertainty. In this report, † indicates that between 50% and 75% of the relevant population is covered by the data and ‡ indicates coverage below 50%.

12.4 Assessment administration

PISA 2025 was administered in participating schools and colleges by trained Test Administrators following standardised procedures developed by the OECD and its international contractors.

Assessment sessions followed a common international format, including approximately two hours of cognitive testing followed by pupil questionnaires. Detailed administration procedures, training materials and assessment scripts were used to ensure consistency across participating countries and economies.

Prior to data collection, participating schools completed technical readiness checks to ensure that local infrastructure met the requirements for computer-based assessment delivery. Where necessary, additional equipment was provided to schools and alternative offline delivery arrangements were available. These procedures were designed to minimise technical disruption and ensure consistent administration across participating schools.

12.5 Automated and human scoring

Responses to many assessment items were scored automatically by the PISA digital assessment platform. Open response items requiring human judgement were coded by trained coders using internationally agreed coding guides and quality assurance procedures. Reliability checks were undertaken throughout the coding process to ensure consistency across participating countries and economies.

12.6 Sample design in Scotland

PISA 2025 used a two-stage sampling design to obtain a representative sample of 15 year old pupils in Scotland.

In the first stage, schools and colleges containing eligible pupils were selected from across Scotland using a stratified sampling approach. Schools were selected with probability proportional to size, meaning that schools with larger numbers of eligible pupils were more likely to be selected. The sample was designed to ensure representation across different types of institutions and geographical areas.

In the second stage, eligible pupils were randomly selected within participating institutions. In institutions with fewer than 40 eligible pupils, all eligible pupils were selected to participate.

PISA technical standards require participating countries and economies to achieve high levels of population coverage and participation. The expected minimum weighted response rates are 85 per cent for schools and 80 per cent for pupils. The overall exclusion rate, combining school level and within-school exclusions, is expected to remain below 5 per cent of the target population.

12.7 Participation in Scotland

The PISA 2025 assessment was administered in Scotland between September and October 2025.

A total of 115 institutions (schools and colleges) were included in the final Scottish sample. This comprised 107 institutions from the main sample and eight institutions from replacement samples. Table 12.2 presents the final participation figures for Scotland.

The weighted participation rate is not calculated simply by dividing participating pupils by the initial sampled total, because eligibility, exclusions and survey weights are taken into account.

The weighted school participation rate was above the OECD threshold of 85 per cent and the weighted pupil participation rate was above the OECD threshold of 80 per cent. Scotland’s data therefore met the OECD requirements for international reporting without any additional caveat relating to participation rates.

Participation rates were similar to those achieved in recent PISA cycles.

12.8 Further education colleges

Scotland’s PISA 2025 sample included eligible 15 year old learners attending further education colleges. This reflects the OECD definition of the target population, which is based on age rather than stage of education.

One participating college was included in the final Scottish dataset.

12.9 Weighting of results

PISA applies survey weights so that participating pupils represent the wider population of eligible 15 year olds in Scotland. The weights take account of pupils’ probabilities of being selected and are adjusted for school and pupil non-response. This means that pupils do not necessarily contribute equally to national estimates. All national estimates presented in this report use the final OECD survey weights.

12.10 Exclusions

OECD technical standards permit a limited number of exclusions from the assessment. Exclusions may occur at either the school level or the pupil level within participating schools.

School-level exclusions relate to institutions excluded from the sampling frame before pupil selection takes place. In Scotland, these primarily comprised special schools and schools providing Gaelic medium education (GME) where assessment materials were not available in the language of instruction (Gàidhlig).

Within-school exclusions relate to individual pupils attending participating institutions who met internationally agreed exclusion criteria. These include pupils with complex additional support needs that prevented meaningful participation, pupils with insufficient experience of the assessment language, and pupils who were no longer enrolled in the sampled institution at the time of assessment.

The overall exclusion rate for Scotland was 7.89 per cent, compared with 6.57 per cent in 2022. While this exceeds the OECD guideline stating that exclusions should not normally exceed 5 per cent of the target population, Scotland’s data were reviewed through the OECD’s international quality assurance processes and accepted for inclusion in the international database.

The increase in the overall exclusion rate was driven primarily by a rise in within-school exclusions. Table 12.4 shows the composition of within-school exclusions in 2022 and 2025.

A large proportion of pupil-level exclusions in Scotland related to additional support needs (ASN). Scotland records a wider range of additional support needs than many education systems internationally, which may contribute to higher exclusion rates than those reported elsewhere. However, because excluded pupils do not participate in the assessment, it is not possible to determine precisely how their inclusion would have affected Scotland’s results.

Although all exclusion categories increased between the 2022 and 2025 PISA cycles, the largest increase was among pupils who were no longer enrolled in the sampled institution at the time of assessment. This largely reflects the earlier timing of PISA 2025 data collection, which required pupil sampling to take place before the end of summer term. Between sampling and assessment, a higher proportion of pupils moved to another educational institution, most commonly from secondary schools to further education colleges, and were therefore no longer eligible to participate through their originally sampled institution.

12.11 Non-response bias analysis

Although Scotland met the OECD student participation standard for PISA 2025, an additional non-response bias analysis was undertaken. This was because Scotland’s weighted pupil participation rate was close to the 80 per cent threshold (the expected minimum weighted response rate for pupils) and similar to the rate achieved in PISA 2022.

The non-response bias analysis compared pupils who participated in PISA with the full eligible sample, using national administrative data held by the Scottish Government. The variables used included Achievement of Curriculum for Excellence Levels (ACEL) in reading, writing, listening and talking, and numeracy, as well as general literacy, free school meal status, Scottish Index of Multiple Deprivation (SIMD), and additional support needs (ASN).

The non-response bias analysis found statistically significant differences between the full eligible sample and the responding pupils across all variables included in the analysis. Responding pupils had slightly higher prior attainment and were slightly less likely to be eligible for free school meals or to be recorded as having additional support needs. However, the differences were small in scale. The analysis therefore suggests some non-response bias in the achieved sample. However, the differences were small, and Scotland’s data were accepted for international reporting.

These non-response patterns were broadly similar to those observed in the analysis undertaken for PISA 2022 in Scotland. The differences in 2025 were small in scale and did not identify evidence that would prevent meaningful comparison with previous Scottish PISA cycles.

Positive differences indicate that responding pupils had a higher average value than the full eligible sample. For free school meal eligibility and additional support needs, lower values among responding pupils indicate that these groups were slightly under-represented among participants.

The existence of statistically significant differences between participating pupils and the full eligible sample is not unexpected in a study of this scale and is one of the reasons why weighting adjustments are applied.

12.12 Data quality and validation

A range of quality assurance procedures were applied throughout the study to ensure that Scotland’s data met OECD technical standards. These procedures included:

• independent verification of the sampling approach
• monitoring of assessment administration
• validation of participation data
• checks on response rates and exclusions
• international verification of submitted datasets

As part of the OECD’s international quality assurance arrangements, independent quality monitors observed assessment administration in a sample of participating institutions to assess compliance with international procedures, assessment security and administration standards. Evidence from these visits formed part of the OECD’s wider quality assurance and data adjudication process.

Following data collection, Scotland’s data were reviewed through the OECD’s data adjudication process. This process considers whether each country or adjudicated region has implemented PISA in line with the technical standards, including standards on sampling, exclusions, participation, assessment administration and data processing.

The PISA Adjudication Group noted that Scotland’s overall exclusion rate was 7.9 per cent, above the OECD guideline of 5 per cent. This was part of a wider international pattern, with around one quarter of adjudicated entities exceeding the exclusion-rate standard in PISA 2025. The Adjudication Group recommended that Scotland’s results be fully included in international reports and used for international and trend comparisons. It also invited national authorities and international contractors to review ways to contain within-school exclusions in future cycles.

12.13 Scaling and reporting of results

PISA pupils do not all answer the same questions. Each pupil is allocated a different combination of assessment material, so that the overall study can cover a wide range of content while keeping the assessment manageable for individual pupils.

Because pupils complete different questions, PISA results are not calculated as simple raw scores. Instead, the OECD uses statistical modelling to place responses on a common proficiency scale. This takes account of both the questions pupils answer and the difficulty of those questions. It allows results to be compared across pupils, countries and PISA cycles.

PISA results are designed to describe performance at population level, for example average performance in Scotland or differences between groups of pupils. They are not designed to provide individual pupil scores. For this reason, the OECD generates a set of “plausible values” for each pupil. These are not individual test marks, but statistical values used to estimate group level results while allowing for the uncertainty involved in measuring performance from a sample of questions.

When comparing results over time, PISA also takes account of additional uncertainty known as linking error. This reflects the fact that each PISA cycle includes some changes to assessment content, frameworks or delivery methods. Linking error is included in trend comparisons to ensure that changes over time are interpreted with caution.

How to access the background or source data

Aggregate data are provided within this publication or its annexes.

Data are held by Scottish Government and may be made available on request, subject to consideration of legal and ethical factors. Please contact school.stats@gov.scot for further information.