ICEQC-R-2006-05 — Completion, Repetition and Late Entry: A Cohort-Based View of Basic Education Quality

The policy outcome is that every child can enter, participate in and complete a full course of primary education of good quality. This formulation contains access, duration and quality. It also places the child, rather than the administrative position, at the centre.[REF-08] [REF-07]

An indicator may represent one component without representing the whole outcome. Final-grade intake records entry to a grade. Survival models retention to a point. An examination pass records a specified assessment decision. A household report of completed primary education records attainment as understood by the survey. None alone establishes all components.[REF-08]

Before calculating completion, the event must be defined. Possible events include first entry to the final grade, attendance for a minimum period, successful completion of required instruction, eligibility to sit a terminal examination, passing that examination, promotion to the next level, or formal award of a certificate.[REF-09]

Systems may use different events under the same label. A published rate should state whether repeaters in the final grade are excluded, whether completion requires examination success, and how learners in private, alternative or non-formal settings are treated.[REF-10]

A cohort measure begins with a population that could experience the event. This may be children of official entry age in a calendar year, new entrants to grade 1 in a school year, all children in a birth cohort, or survey respondents above a stated age.[REF-11]

These populations are not interchangeable. Grade 1 entrants exclude eligible children who did not enrol. A birth cohort includes them. A survey age group may include several school-entry cohorts. Every conclusion should remain within the population observed.[REF-12]

The population of official graduation age is commonly used because it is available from demographic estimates and creates a standard denominator. Gross intake to the last grade divides new entrants to that grade, regardless of age, by the population at the official entrance age for the last grade.[REF-01]

The numerator is therefore not restricted to the denominator’s birth cohort. Late entrants and repeaters from older cohorts, and early entrants from a younger cohort, may all enter the final grade in the same year. A result above 100 per cent can arise without any individual being counted twice in the numerator.[REF-02]

Entry to the last grade is a practical proxy where successful completion data are not consistently available. It indicates that a learner has progressed to the final stage of the cycle, but it may precede drop-out within that grade, examination failure or non-award.[REF-03]

The proxy should be named accurately. If policy requires successful completion, a final-grade entry rate should not be described as a graduation rate without explanation. The distance between entry and successful completion can be material in examination-based systems.[REF-04]

Survival to the final grade estimates the proportion of a pupil cohort expected to reach that grade. It is conditional on entry to grade 1. It does not include children who never enter and should not be presented as the share of all children who complete.[REF-05]

A high survival rate can coexist with low population coverage where few children enter. A high entry rate with low survival indicates a different failure. The two measures should be read together.[REF-06]

Household attainment asks whether a person has completed a specified level, often among an age group old enough to have done so. It can capture persons absent from current school records and is well suited to distributional analysis.

It may lag current policy change because older respondents experienced earlier conditions. Reported level also requires a reliable mapping of national programmes and respondent understanding. The age group and reference definition determine whether the result represents timely, recent or eventual completion.

Transition to lower secondary education measures movement after primary education. It is not the same as primary completion. Some completers may not continue; some systems may permit transition through alternative rules; and the timing between cycles may differ.

The 2005 Global Education Digest shows why the distinction matters by reporting transition from primary to lower secondary separately from indicators of primary completion. Policy analysis should retain that separation while examining the pathway between them.[REF-03]

The 2006 global development monitoring report states that universal primary completion is meaningful only if it signifies command of a relevant level of knowledge and skills. This is a boundary on interpretation, not a reason to abandon completion measurement.[REF-07]

Learning evidence and completion evidence have different denominators. An assessment may cover only enrolled pupils at a given grade, excluding those who already left. Reporting average performance among survivors without the survival rate can overstate the education received by the original child population.

Every use of the word completion should be accompanied by the indicator definition or a direct reference to it. The public table should identify level, cycle length, official age, numerator, denominator, data source, reference period and whether the result is gross, net, cohort-based or household-based.

Where a familiar short label is retained, the technical note should prevent the label from acquiring a broader meaning than the statistic supports.

An entry cohort consists of new entrants to the first grade in a defined school year. It excludes first-grade repeaters. Accurate separation is essential because total grade 1 enrolment can be much larger than new entry where repetition is high.

The cohort may contain children of several ages. Reporting only its size conceals late and early entry. Age should be recorded at a consistent reference date and grouped in relation to the official age.

A birth cohort includes all children born during a defined period, whether they enter school or not. It is the appropriate conceptual population for universal completion but is difficult to follow directly where civil registration, migration or longitudinal learner records are incomplete.

Household surveys and population censuses can approximate the education status of birth cohorts at later ages. Administrative records can describe enrolled members. Reconciliation is required to understand the whole cohort.

Promotion occurs when a learner moves to the next grade under the applicable rules. The promotion rate usually relates promoted pupils in year t+1 to enrolment in a grade in year t, after accounting for the data system’s definitions.

It is a flow estimate, not necessarily an individually linked outcome. Migration, transfers, re-entry, school coverage changes and reporting errors can affect the relationship between years.

A repeater is a pupil enrolled in the same grade for a second or further year, according to the statistical definition. The percentage of repeaters describes repeaters as a share of total enrolment in that grade. A repetition rate may instead relate repeaters in the following year to the prior-year grade population.

These measures use different denominators and should not share one label. A system should specify whether a pupil returning after interruption is classified as a repeater, re-entrant or new entrant.

Drop-out is often derived as the residual after promotion and repetition. Any error in those components enters the drop-out estimate. The term may include permanent departure, temporary interruption, unrecorded transfer, migration or movement to provision outside the reporting system.

For policy purposes, the distinction matters. A child who returns after one year has experienced lost time but not permanent exclusion. Longitudinal or household evidence is needed to determine the subsequent pathway.

Re-entry describes return after an interruption. It can occur in the same grade, a later grade following assessment, an alternative programme or a different institution. Administrative systems frequently lack a separate category, causing re-entrants to be counted as repeaters or new entrants.

The treatment affects flow rates and age-for-grade. A cohort account should identify re-entry where possible and should state the rule where it cannot.

Some learners progress by more than one grade through acceleration, recognition of prior learning or a return assessment. These flows are small in many systems but can disturb a model that permits only promotion, repetition and drop-out.

An explicit category is preferable where the practice is material. Otherwise the technical note should state how it is incorporated.

Movement between schools is not departure from education. A school-level cohort requires transfer records to avoid treating outgoing pupils as drop-outs and incoming pupils as new local entrants. A national cohort is less affected by internal transfer but remains affected by cross-border movement and incomplete provider coverage.

Stable learner identifiers can improve tracking where legally and administratively feasible. They do not correct inaccurate dates of birth, grade codes or completion decisions.

Death and migration change the population that can be observed. A cohort model may implicitly include these events in attrition unless separate data exist. Their treatment should be stated, particularly in settings with substantial displacement or population movement.

Population denominators may also be based on projections from an earlier census. Sudden movement can therefore affect both numerator coverage and denominator accuracy.

For a closed and consistently measured grade cohort, the prior-year population is allocated among promotion, repetition and departure from the observed system. In practice, transfers, re-entry, migration and data correction mean that the identity may not reconcile exactly.

The difference should not be hidden by forcing the residual to zero. A reconciliation item can show the effect of unmatched records and changing coverage.

Consider a constructed cohort of 1,000 new grade 1 entrants in a six-grade cycle. Assume the flow rates shown in Table 2 and no re-entry. The example demonstrates how annual repetition and departure accumulate; it does not represent an observed system.

At each grade, pupils who are promoted proceed, repeaters remain exposed to the same grade in a later year, and leavers exit the model. A full calculation must permit repeaters to attempt progression again rather than remove them from the cohort.

The declining grade-entry figures in this constructed table are not calculated directly from the first-year flow columns because repeaters from earlier years later rejoin progression. They are separate rounded cohort states chosen to show why a one-year flow table and a longitudinal cohort table should not be merged without a transition model.

A percentage of repeaters in one school year does not show how many individual children ever repeat. The same pupil may appear as a repeater in more than one year or grade. Conversely, a moderate annual rate spread across different pupils can affect a large share of the cohort.

Longitudinal records can report the number of repetition episodes and the share of children with at least one episode. Reconstructed models should state the assumption about repeated repetition.

Cycle duration describes the designed number of school years. Time to completion describes the elapsed time actually required by a learner. Late entry shifts the age of completion; repetition and interruption increase both age and elapsed time after entry.

Median or distributional reporting is preferable to a single average where some children complete much later. The share completing within the official duration, within one additional year and eventually should be distinguished.

Timely completion requires an age or elapsed-time rule stated in advance. Eventual completion requires an observation age sufficiently above the expected completion age to capture delayed learners. Neither term has meaning without that boundary.

The gap between the two shows delay among those who ultimately complete, but it can also reflect incomplete observation of younger cohorts. Comparisons should use the same age windows and cycle mappings.

Attrition should be shown by grade, age and time since entry. A final survival percentage cannot identify whether most loss occurs before grade 2, at a terminal examination, during a transition in language of instruction or after several repetitions.

The location of attrition directs different policy responses. Early-grade loss requires attention to readiness, language, attendance, teaching and family costs; late-cycle loss may involve examinations, distance, opportunity cost or limited places beyond primary school.

Repetition increases the number of pupil-years required to produce a completer. This can enlarge classes, consume teacher time and delay access where capacity is fixed. The relevant measure is not only the annual repeater share but the cumulative excess pupil-years across the cohort.

An efficiency calculation should not treat every additional year as waste without examining learning and circumstances. Its purpose is to show resource and time consequences, not to justify automatic promotion.

Pupil-years per completer divide the total years of enrolment accumulated by a cohort by the number completing. The result can be compared with the official cycle duration. The difference reflects repetition and the pupil-years of learners who leave before completion.

This is an internal-efficiency indicator. It does not value learning, equity or the social benefit received during incomplete schooling and should not be used as a complete measure of education quality.

Reconstruction requires enrolment and repeaters by grade for consecutive years, consistent school coverage, stable grade definitions and a rule for promotion, repetition and drop-out. Changes in cycle structure or automatic-promotion policy create breaks that should be marked.

Where rates fluctuate sharply because of small numbers or reporting changes, a synthetic cohort based on one year can be misleading. A multi-year account and sensitivity range may be more responsible.

A public cohort account should show the initial population, entry, progression, delay, departure and completion, with an explicit unobserved or unmatched category. It should not imply person-level tracking where only aggregate reconstruction exists.

The account becomes more useful when disaggregated, but small cells, unstable estimates and incomplete age records require proportionate interpretation. The objective is to reveal where opportunity is lost, not to create spurious precision.

The official entry age is the age prescribed or normally expected for beginning primary education. Its statistical application requires a reference date. A child may turn the official age before the school year, during it or after the enrolment census, depending on national rules.

Comparative reporting should not infer late entry from age alone until the cut-off rule and school calendar are known. A one-year difference can be produced by the timing convention rather than delayed admission.

New entrants to grade 1 should be distributed by single year of age where records permit. The distribution distinguishes entry at the official age from entry one, two or more years later and from early entry.

An average entry age can conceal a wide distribution. The median, selected percentiles and shares by relation to official age provide a more complete account.

Gross intake relates all new grade 1 entrants, regardless of age, to the population of official entry age. Net intake restricts the numerator to new entrants of official entry age and uses the corresponding population.

A high gross rate and a lower net rate can indicate late or early entry, population error or both. The difference is not itself a late-entry rate because the gross numerator may include several ages and the population estimates may not align with school records.

An entry profile places the age distribution of first entrants beside the population participation of the same ages. It distinguishes children entering late from children already enrolled after earlier entry and from children still outside school.

Household evidence is particularly useful for establishing the status of children not present in administrative registers. Administrative entry counts remain necessary to examine grade placement and institutional flow.[REF-04]

Late entry may reflect distance, fees and indirect costs, insufficient places, disability, language, household work, migration, conflict, health, birth-registration requirements or local judgements about readiness. Different causes can produce the same age distribution.

The indicator identifies a pattern requiring inquiry. It does not establish household preference or institutional failure without contextual evidence.

The right to primary education and the duty to encourage regular attendance require attention to barriers that postpone participation, not only to children permanently outside school. A child admitted several years late may retain a route to completion but faces a shortened period before work, marriage or other pressures intensify.[REF-11]

Policies should examine whether admission rules themselves exclude over-age children or direct them to provision with weaker resources or status.

Readiness should not become an undefined basis for delaying entry. If systems assess readiness, the purpose, method, language, decision authority and available early support should be clear. Socioeconomic or linguistic difference should not be mistaken for inability to learn.

Early childhood provision may support transition, but lack of such provision should not become an additional gate to compulsory primary education.

Age reporting depends on birth records or another credible method. Where documentation is incomplete, exclusion for lack of a certificate can conflict with the education objective. An accessible provisional procedure may be necessary while the record is regularised.

Statisticians should distinguish recorded age from verified date of birth. Digit preference, default dates and retrospective estimation can create artificial concentrations in age distributions.

Early entry can arise from parental demand, flexible admission, inaccurate age, composite classes or limited early childhood provision. It may contribute to gross ratios above 100 per cent and can complicate learning if the curriculum assumes an older developmental stage.

The policy response requires evidence of learning, participation and wellbeing. A low official-age intake rate should not automatically be attributed to exclusion if substantial early entry is verified, though early entry does not compensate for children who remain outside school.

Age-for-grade compares a learner’s age with the age expected for the grade under uninterrupted progression from official entry. A tolerance rule is required because birthdays and school-year timing can create normal variation.

The measure is a stock outcome. It reflects the combined history of entry, repetition, interruption, grade change and possibly age error. It does not identify which event caused the difference.

Over-age enrolment is the share or number of pupils older than the expected range for their grade. It should be reported by single grade because cumulative delay and grade-specific bottlenecks produce different patterns.

A system may show high over-age enrolment in grade 1 because of late entry, a sharp increase after grade 1 because of repetition, or gradual accumulation. Each pattern requires a different inquiry.

An age-grade distortion index can summarise the distance between actual and expected age, but a single score conceals direction and distribution. Reporting the shares on age, one year over age, two or more years over age and under age is more transparent.

Where age heaping is visible, broad categories may be more reliable than fine estimates. The selected tolerance should be constant across the comparison or clearly identified.

Late entrants can repeat, and repeaters can experience interruption. Their effects are cumulative. A child entering two years late and repeating twice may reach the final grade four years above the official age even with otherwise continuous attendance.

Cross-sectional analysis should therefore examine entry age within later grade-age distributions where records permit. Assuming all over-age pupils are repeaters can misdirect policy.

As children grow older, the opportunity cost of attendance may increase and the social fit of remaining in a class with much younger pupils may weaken. Distance, safety, work and family responsibilities can have different effects by sex and location.

An association between over-age status and leaving does not by itself prove causation. Poverty, school access and learning difficulty may cause both. Longitudinal and contextual evidence should be used.

Wide age ranges can affect teaching, materials, peer relations and safeguarding. Multi-age teaching may be intentional and effective where staff are prepared; an unplanned range created by delay and repetition presents a different condition.

Classroom information should be connected to the age distribution without treating older pupils as the problem. The system response should address the barriers and progression decisions that produced the distribution.

Accelerated programmes may provide over-age children with an age-appropriate route through equivalent learning. Their quality depends on curriculum, instructional time, trained teachers, assessment, transition and recognition within the formal system.

Counting participants as enrolled is not sufficient. Monitoring should establish completion, learning and successful movement into further education where that is the intended route.

Consider a six-grade system with official entry at age 6 and an enrolment census near the beginning of the year. In a constructed intake of 1,000 pupils, 610 are recorded as age 6, 90 as younger, 210 as age 7, 70 as age 8 and 20 as age 9 or older. The official-age share among entrants is 61 per cent and the late-entry share, using any age above 6, is 30 per cent.

These percentages describe entrants, not all 6-year-olds or all children who should have entered. A net intake rate would require the population aged 6, while a population late-entry measure would require observation of children outside school and their later status.

Validation compares dates of birth across admission forms, registers, examinations and household sources where lawful. It examines impossible ages, default dates, abrupt heaping, missingness by school and inconsistent ageing between years.

A correction protocol should preserve the original value, evidence, authorised change and date. Changing ages to make a cohort appear on grade is a serious data-integrity failure.

Public reporting should show the full entry-age distribution and the share of the eligible age population not yet entered, where evidence permits. It should separate administrative entrants from household attendance and explain the school-calendar reference.

The policy account should identify groups and locations with persistent delay, causes supported by evidence, and action on admission barriers, places, cost, language, disability and alternative routes. It should not frame older entrants as statistical inconvenience.

Formal repetition occurs when an authorised decision requires a learner to remain in the same grade for another year. The decision may follow assessment, attendance, examination or another rule. Its existence should be supported by an individual record.

The statistical category should not depend on whether the school uses the word failure. A learner retained in the same grade is a repeater for flow analysis if the applicable definition is met.

Families or learners may request another year because of interruption, illness, language transition or concern about readiness. Systems should state whether such cases are counted with formal repeaters and whether the educational plan differs.

Voluntary classification does not remove the need to examine whether the request resulted from inadequate support or an inaccessible next grade.

Hidden repetition occurs where a learner is recorded as a new entrant, transfer or re-entry despite having studied the same grade previously. It may arise from weak identifiers, movement between schools, incentives attached to low repetition or deliberate misreporting.

Age patterns and matched records can indicate possible undercounting, but they do not prove the status of each child. Data review should protect learners from blame for institutional reporting weaknesses.

Automatic or social promotion reduces formal retention but does not by itself improve learning. Its quality depends on formative assessment, targeted support, curriculum alignment and the capacity of the next grade to address unfinished learning.

A low repetition rate under automatic promotion should be published with learning, attendance and progression evidence. Conversely, the absence of automatic promotion does not establish that repetition is instructionally effective.

Where promotion depends on an examination, repetition may reflect the pass standard, examination design, opportunity to learn, language, attendance or capacity in the next grade. A change in the examination can alter rates without a change in underlying learning.

Trend reporting should mark changes in pass rules, permitted attempts and grade classification. Examination failure and grade repetition should be shown separately if not every unsuccessful candidate repeats.

Some systems require repetition after a threshold of absence. The resulting rate describes attendance conditions as well as academic decision. Illness, transport, work, insecurity and school closure may be material causes.

Intervention should address attendance and lost instruction rather than assume that another identical year will correct the problem.

Children beginning in an unfamiliar language of instruction may be more likely to repeat, particularly in early grades. The indicator should prompt review of language support, teacher preparation, materials and assessment rather than an inference of low capacity.

Where language policy changes, repetition trends should identify affected cohorts and regions. Aggregate national rates can conceal concentrated effects.

Repeated retention can result from inaccessible teaching, assessment or facilities rather than from the learner’s potential. Age-for-grade divergence may then increase social exclusion and the likelihood of leaving.

Data systems should support relevant disaggregation with appropriate confidentiality and definitions. The analysis should examine reasonable support and educational progress, not use disability as an unexplained residual category.

Direct and indirect costs are paid again when a grade is repeated. Households may also lose time for work or care. These costs can turn an initial retention decision into later departure.

Association with poverty may operate through attendance, nutrition, distance, school resources and unequal access to supplementary support. A repetition policy should be assessed for distributional effect.

The overall primary repetition rate is an enrolment-weighted summary. It can hide a severe first-grade bottleneck or a terminal-grade concentration. Counts and rates should be shown for every grade.

The first grade requires particular attention because new entrants and repeaters must be distinguished accurately. Large first-grade enrolment does not necessarily indicate a large incoming cohort.

Within a grade, repetition should be examined by age relative to the expected range. Older pupils may have accumulated prior repetition or entered late and may face greater pressure to leave.

Age-specific rates require enough records and should not imply causal direction. An older learner can repeat because of barriers also associated with the original delay.

The percentage of repeaters is the number of repeaters in a grade and year divided by total enrolment in that grade and year. It describes the composition of current enrolment.

It is not identical to the probability that a prior-year pupil repeats. Changes in grade size and migration can create differences between the two concepts.

A flow repetition rate commonly relates pupils repeating grade g in year t+1 to pupils enrolled in grade g in year t. It requires consecutive years and compatible coverage. The numerator should correspond to the prior-year population.

Where pupils cannot be linked, aggregate counts approximate the flow. The publication should state this and identify major coverage changes.

Misclassification inflates new entrant counts and depresses repetition. In grade 1 it can overstate intake; in the final grade it can overstate gross intake to that grade if repeaters are not removed.

Validation should reconcile each grade’s enrolment with new entrants and repeaters and investigate schools reporting no repeaters despite unchanged large cohorts.

The central educational test is whether an additional year produces learning and sustained participation that could not be achieved through a less disruptive intervention. Evidence should compare the learner’s prior need, support received, later learning and progression.

Simple comparison of repeaters with promoted pupils is biased because the groups differed before the decision. Evaluation should account for prior performance and circumstances.

In a capacity-constrained system, repeaters occupy places alongside new entrants. This can increase class size or restrict first entry. The effect depends on staffing, facilities, grade distribution and whether finance follows enrolment.

Capacity pressure is a reason to improve early learning and support, not to promote children without regard to learning. Both delay and insufficient instruction impose costs.

A policy limiting repetition can cause an immediate statistical decline. The result should be interpreted with promotion, learning, attendance, class size and later drop-out. A break in the series should mark the change.

Implementation may vary across schools. Monitoring should examine actual decision records rather than rely only on the national rule.

Suppose two districts each report that 8 per cent of primary pupils are repeaters. In District A, repetition is concentrated in grade 1, age records are reliable and most retained pupils progress the following year after a structured literacy programme. In District B, repetition is spread across grades, many pupils repeat more than once and follow-up support is not recorded.

The common percentage does not describe a common problem or result. Grade, prior history, intervention and later pathway change its meaning. A comparative report should place those fields beside the rate before drawing a quality conclusion.

An audit should sample the rule applied, assessment evidence, attendance, communication, available support, authorised decision and later outcome. It should examine consistency across schools and groups and identify informal retention not present in the statistical return.

The audit should also examine promoted pupils with very low learning evidence. The objective is an appropriate supported progression decision, not the mechanical reduction or preservation of repetition.

A public release should state the measure, grade pattern, sex and location differences, age relation, policy rules, series breaks and missingness. It should distinguish an observed reporting change from an educational improvement.

The accompanying policy statement should identify what is known about causes and later outcomes, what remains uncertain, and which supports or data improvements will be tested.

Administrative systems observe institutions, classes and registered pupils. Household surveys observe residents selected through a sample and report their education status whether or not they appear in a school return.[REF-04]

The two sources should not be expected to produce identical results. Their value lies partly in the different exclusions and errors they reveal.

Enrolment is an administrative status; attendance is participation during a stated period. A child may be enrolled but absent for much of the year, or attend a form of provision not included in the administrative collection.

Completion analysis should specify which concept is used at every stage. Regular attendance is a substantive concern in its own right.[REF-11]

An annual school census may occur near the beginning or middle of the year. A household survey may ask about attendance during a recent week, current year or previous year. Seasonal work, closure and examination timing can affect the response.

Reconciliation requires alignment of school year, fieldwork date and the period described by the question.

Administrative collections may omit or incompletely cover private, community, religious, alternative or remote provision. Household surveys may capture attendance in these settings but classify programme level imprecisely.

Coverage should be stated by provider type. An increase caused by the addition of previously omitted schools is a statistical improvement, not necessarily enrolment growth.

Household surveys can omit children outside conventional households, including some children in institutions, on the street, displaced or living in insecure areas. Non-response and inaccessible regions can also be concentrated among disadvantaged populations.

Weights correct the sample under stated assumptions; they do not restore groups absent from the sampling frame. Exclusions belong in the principal limitation statement.

Administrative rates depend on population denominators derived from censuses, registration or projections. Age-specific error can materially affect net intake and gross completion measures, particularly for small populations or long intervals since a census.

Revision of the population series can change a rate without any alteration in school records. Releases should identify the denominator version and preserve earlier published values with a revision note.

Funding or staffing incentives may encourage overstatement of enrolment. Duplicate pupils, names of children who left, and attendance below the qualifying threshold can remain on registers.

Validation can compare registers, attendance, transfer records and household evidence. It should not assume fraud where late reporting or weak systems provide a sufficient explanation, but material discrepancies require resolution.

Schools may omit pupils because of delayed admission, missing documents, unofficial fees, language, disability or operation outside the reporting system. Pupils may attend while remaining absent from the official register.

An administrative rate can therefore understate participation as well as overstate it. Household follow-up and local listing can identify the direction.

Reliable repeater status requires a prior grade history. Teachers may infer it from age or familiarity, while transferred pupils may be recorded as new. A changed pupil identifier can break the link.

The collection form should define repetition and re-entry and should include internal checks between enrolment, entrants and repeaters.

Household respondents may report a local programme name, standard, form or year that must be mapped to the national classification. Reform, combined grades and non-formal programmes can make the mapping uncertain.

The published crosswalk should identify ambiguous categories and the rule used. Reclassification should not be concealed as attainment change.

School records may use documents or estimates; household surveys rely on records or respondent report. Both can show age heaping. The census date used to calculate age may differ.

Age comparisons should inspect missingness, implausible values and concentration on preferred digits before producing precise late-entry rates.

A respondent may interpret completion as attendance through the last year, passage of an examination or receipt of a certificate. The questionnaire and interviewer guidance determine which concept is captured.

Administrative records may separately contain all three events. Household estimates should be described according to the question asked, not retrofitted to the preferred administrative definition.

Household survey estimates are subject to sampling error, especially for small regions, narrow ages and disaggregated groups. Confidence intervals or another uncertainty statement should accompany material comparisons.

Two point estimates should not be described as different where the design cannot support that conclusion. Survey design effects and clustering should be incorporated in variance estimation.

Question wording, recall, interviewer practice, coding, non-response and data processing produce non-sampling error that may exceed sampling error. It is not captured by a confidence interval.

Quality reports should record field procedures, response rates, editing and known problems. Apparent precision to one decimal place does not remove these limitations.

Administrative data are often presented without an error range, but they are not error-free. Coverage, classification, duplication, timing and denominator uncertainty affect the result.

A quality statement can classify fields by verification status and report the size of unresolved discrepancies. The absence of sampling error is not the absence of uncertainty.

School data are commonly assigned to the school location, while household data are assigned to the child’s residence. Cross-boundary travel can create different local rates even when both sources are accurate.

National comparisons are less affected, but boarding, migration and cross-border attendance remain relevant. Local reconciliation needs both residence and service-location views.

The sources should use the same age range, education levels and inclusion rules before rates are compared. A household participation measure that includes secondary attendance among primary-age children will differ from a primary net enrolment rate that excludes it.

The difference should be decomposed by definition before being interpreted as reporting error.

A source gap can point to enrolled non-attenders, attendance outside registered provision, population error, school overcount, household undercoverage or mismatched periods. The size alone does not determine the cause.

Diagnosis should proceed from definitions to coverage, time, geography, age and record validation. Selecting one figure because it is higher or lower is not a statistical method.

The responsible statistical body should document the initial values, harmonised definitions, exclusions, adjustments, residual difference and final interpretation. Adjustments should be reproducible and should not overwrite source data.

Where no resolution is possible, both estimates may be published with their purposes and limits. Uncertainty is more informative than an unsupported single number.

Household data can estimate attainment for successive age groups, providing a retrospective cohort profile. The pattern may reveal delayed completion where attainment continues to rise several years after official graduation age.

It is not a direct longitudinal observation. Differences between ages can reflect cohort change, sampling and migration as well as delayed progression.

Administrative series can reconstruct flows close to the time policy is operating. They offer grade detail and annual frequency but depend on stable coverage and classification.

The publication should label a synthetic or reconstructed cohort and state whether the same individuals are linked. This prevents model output from being read as a tracked history.

Combination can use household evidence for entry and exclusion, administrative evidence for grade flow, and assessment evidence for learning. The components should retain their own populations and uncertainty.

A composite conclusion should explain the bridge between them. Multiplying rates from incompatible years or populations creates an apparently complete estimate without a valid cohort.

Priority improvements include reliable first-entry and repeater fields, dates of birth, stable school and pupil identifiers, complete provider coverage, school-calendar metadata, current population estimates and survey questions aligned to national programmes.

Improvement should focus first on fields that change decisions. Collecting additional variables without correcting basic cohort status increases burden without increasing reliability.

A comparative release should publish source-specific results, definitions, coverage, uncertainty and a reconciliation account. Tables should preserve missing values and footnotes and identify provisional or estimated data.

The narrative should state whether the evidence supports a direction of change, a level comparison, a distributional finding or only a need for further verification. This discipline protects both public accountability and statistical credibility.

Gross intake to the last grade counts pupils entering that grade for the first time and divides them by the population at the official entrance age for the grade. Repeaters in the last grade are excluded from the numerator; pupils of other ages are included.

The measure is widely usable because it requires current grade records rather than a long linked history. Its principal weakness is the mismatch between an all-age numerator and a single-age denominator.

If delayed cohorts, early entrants and pupils who previously repeated converge on the last grade in one year, new entrants can exceed the population of official graduation age. A low population estimate can have the same effect.

Values above 100 per cent should be published as observed, with investigation and explanation. Capping removes information and distorts aggregates.

A net measure restricts final-grade new entrants to the official age and divides by the population of that age. It describes timely entry to the final grade under the selected age rule.

It omits older and younger entrants who may validly complete. Read beside a gross measure and the age distribution, it can distinguish timely progression from total current volume.

Successful completion may require attendance through the final year, satisfaction of course requirements, passage of a terminal assessment or a formal award decision. The governing rule should define the event.

Comparability is weak when one system counts entry to the final grade and another counts examination success. A common heading should not conceal the difference.

Household surveys can report the proportion of persons in a defined age group who completed primary education. An age close to official graduation measures relatively timely completion; an older range captures more delayed completion but refers to earlier schooling conditions.

The selected age group should be justified against cycle length and delay. Wide age bands improve precision but can blur cohort change.

Completion among entrants divides completers by the original entry cohort. It measures internal progression conditional on entry. It can be estimated through linked records or a cohort model.

It must not be described as universal completion because children who never entered are absent from the denominator. Entry coverage should accompany it.

Population completion relates completers to the full relevant child or birth-cohort population. It incorporates non-entry and later attrition and is closest to the universal policy objective.

Its construction is demanding. Numerator and denominator must refer to the same cohort, migration and mortality must be considered, and the observation period must be long enough for delayed completion.

Some national systems use apparent intake or apparent completion measures based on total enrolment in the first or final grade. If repeaters are not removed, the resulting ratio is inflated where retention in those grades is high.

The term and formula should be stated. Apparent measures cannot be substituted for first-entry measures without qualification.

The survival rate to grade g estimates the share of a pupil cohort that reaches that grade, with or without repetition. Reconstructed cohort methods apply observed flow rates to a notional entrant cohort.

Results depend on the model, reference years and assumptions about re-entry and migration. A value should not be reported without the grade and method.

A cohort model can estimate the probability that an entrant eventually completes after one or more repetition episodes. This differs from survival to first entry into the last grade where completion requires an additional event.

The model should specify the maximum duration, transition probabilities and treatment of repeated interruption. Open-ended eventual completion is not observable from a short series.

Promotion, repetition and drop-out rates for a grade and cohort should reconcile to the full observed population, subject to an explicit adjustment. Negative residual drop-out or totals above 100 per cent indicate incompatible counts, coverage change or model failure.

Correction should address the source records. Redistributing the discrepancy mechanically among the three rates makes the table appear consistent while concealing uncertainty.

A period measure describes events in a school year. A cohort measure describes the experience of a group over time. Period improvement can precede full cohort completion because older grades still contain pupils exposed to earlier conditions.

Trend analysis should not infer the eventual result for a new cohort solely from one improved annual promotion rate.

A synthetic cohort applies the grade-flow rates observed in one period to a notional group as if those rates remained constant through its progression. It provides a current summary without waiting for a full cycle.

The price of timeliness is assumption. Where policy, conflict, reporting or enrolment is changing rapidly, the synthetic path may not resemble any actual cohort.

An actual cohort follows the same entrants through records or repeated observations. It captures the sequence experienced but requires stable identifiers, transfer tracking and sufficient time.

Attrition from the data system must be distinguished from departure from education. The longest followed cohorts may also represent conditions no longer current.

Age groups observed in one survey can be arranged to show education status across ages. This provides a population profile but not the same children at successive ages.

Differences may reflect cohort history, migration, mortality and sampling. The method should be called an age or pseudo-cohort profile rather than longitudinal tracking.

Entry coverage and conditional survival can be combined conceptually to estimate population reach. The multiplication is valid only where the measures refer to compatible cohorts, definitions and periods and where the survival denominator is correctly conditional.

Combining one year’s intake with another period’s survival can be useful for a model, but the assumptions and sensitivity should be shown.

Completion by age can be shown cumulatively: by expected graduation age, within one additional year, within two or more years, and by a final observation age. This reveals delayed completion without requiring every school transition.

The approach depends on accurate attainment and age. The last window should not be labelled eventual if a material share may complete later.

The gap between timely and later completion quantifies delay among the population under the chosen windows. It should be expressed in percentage points, with both component rates and age definitions.

The gap is not a measure of repetition alone. Late entry, interruption, re-entry and programme duration all contribute.

The non-entry gap is the share of an eligible cohort that has not entered by a stated age. It is best estimated from population evidence and should distinguish children expected to enter later from those unlikely to enter without intervention where the data support that distinction.

Predictions about future entry require an explicit model. A current status should not be converted into a permanent category without evidence.

The difference between entry and completion can be decomposed into loss before entry, loss after entry and delayed status at observation. Such decomposition clarifies whether expanding places, improving early progression or supporting return is the immediate priority.

Residual and unobserved categories should remain visible. A complete arithmetic allocation is not evidence that every child’s pathway is known.

High repetition in the final grade can raise total enrolment while suppressing the number of new entrants or successful completers. An apparent final-grade ratio may therefore look high despite a bottleneck.

Reporting should separate first entrants, repeaters, examination candidates, passes and awards. The sequence can identify whether congestion arises before or after instruction is completed.

Pass rates commonly use candidates as the denominator. Children who never reached the examination or were not permitted to sit it are excluded. The measure therefore cannot represent completion for the original entrant or child population.

Candidate coverage, absenteeism and repeat attempts should accompany the pass rate. Changes in the examination standard or administration create a series break.

A graduation ratio may divide graduates of any age by the population of typical graduation age. Like gross intake, it can exceed 100 per cent. It differs from a cohort graduation rate based on entrants.

International tables require exact naming. The word rate should not imply an individual probability when the formula is a gross population ratio.

Transition to lower secondary depends on primary completion, available places, admission rules, cost and family decisions. A high transition rate among completers can coexist with low population transition where few complete.

Reporting the complete pathway requires primary entry, completion and transition denominators, not transition alone.[REF-03]

Primary cycles differ in duration. Last-grade entry in a four-year cycle and in a seven-year cycle represents different time exposure and age. Comparison of rates remains possible only for the defined level, not for equal years of schooling.

Reports should show cycle duration and the national programme mapped to the international classification.[REF-05]

Reclassification of grades, extension of compulsory education or movement of a grade between primary and secondary levels can change completion statistics. The series should be split or bridged using a documented method.

Back-casting under the new structure should not be performed unless the historical data support it. Both old and new definitions may need temporary publication.

Countries with different school-year dates may report data labelled with the same calendar year but covering different periods. Adjacent-year flow rates are particularly sensitive to the pairing.

The metadata should state the beginning and end of the school year, census date and year-labelling convention.

Census results can revise age populations and therefore historical ratios. The statistical authority should identify whether a series uses the old or revised denominator and whether past years were recalculated.

Trend breaks caused by revision should not be attributed to schools. Versioned releases preserve accountability for both the original and revised evidence.

Regional results may be population-weighted, country medians or unweighted averages. These have different meanings. A weighted rate can be dominated by large systems; a median describes the middle reporting country.

Coverage should show the countries and share of the relevant population represented. Missing high-exclusion systems can bias the aggregate upward.

Percentage-point change is generally clearer for rates than percentage change, especially near zero or 100. Average annual change should state whether it is linear, compound or based on fitted values.

Starting level, interval and data revisions must be shown. A faster relative increase from a low base does not necessarily produce a higher final level.

The 2006 global development monitoring report estimated that the number of developing countries achieving universal primary completion rose from 37 in 2000 to 50 in 2004, while about one-third remained off track and 18 per cent lacked adequate data.[REF-07]

This evidence supports both recognition of progress and restraint. The missing-data share is part of the result, and an achievement classification does not describe delay, distribution or learning within each system.

A comparative claim should pass six tests: common event, compatible population, aligned period, comparable structure, adequate coverage and stated uncertainty. Failure of a test does not make the underlying data useless; it limits the claim that can be made.

The strongest public comparison is often a set of differentiated findings rather than a single rank.

A national completion rate can improve while particular groups remain excluded or delayed. Distributional analysis asks who enters, who repeats, who leaves, who completes on time and who demonstrates learning.

The public-interest concern is not limited to equal averages. Barriers, treatment and educational results require separate examination.

Completion should be reported separately for girls and boys, with the same definition and denominator. The 2006 global development monitoring report identified substantial completion differences in South Asia and sub-Saharan Africa at the time.[REF-07]

Regional evidence should not be applied automatically to every country or locality. National and subnational patterns may differ in direction and cause.

A gender parity index divides the female value by the male value. A value near one indicates numerical parity for the indicator but says nothing about whether the common level is adequate.

The two underlying rates and their denominators should always accompany the ratio. Ratios become unstable where the denominator rate is very low.

The percentage-point difference is often easier to interpret than a ratio. It shows the absolute distance between group rates. Its sign convention should be stated.

Neither measure identifies the number of children affected. Counts are necessary for planning places, support and outreach.

Equal entry does not guarantee equal completion. Differences can emerge through attendance, repetition, safety, school work, household work, early marriage, pregnancy, costs, treatment or transition rules.

The pathway should be compared grade by grade. A final gap without earlier flow cannot locate the point of divergence.

Fees, materials, transport, meals and foregone work can delay entry and increase interruption. Household surveys can compare completion across a wealth measure, but the measure is relative to the survey population and may not be comparable across countries without careful construction.

The poorest–richest gap should be accompanied by the middle distribution and urban–rural composition. A binary contrast can conceal where most children are located.

Rural, urban and remote categories depend on national definitions. School location and household residence should not be mixed. Distance, population density, teacher deployment and transport influence progression.

Urban averages may conceal informal settlements and migrant populations; rural averages may conceal accessible towns and isolated communities.

Subnational results reveal where national progress is uneven, but small denominators and population projections can make district rates volatile or above 100 per cent. Boundary changes require a geographic crosswalk.

Publication should support action without treating a rate as a complete judgement on local administration. Resource and context evidence are needed.

The relationship between home language and language of instruction can affect entry, learning, repetition and completion. Categories should reflect the learner’s actual language context rather than broad ethnic labels where possible.

Analysis should examine the availability of teaching, materials and assessment support. Language difference is not an inherent educational deficit.

Many systems lack reliable information on disabled children, particularly those not enrolled. School records alone can therefore produce a misleadingly high participation or completion rate among the identified population.

Household identification and school support records may complement each other, using definitions appropriate to the evidence available in 2006. Absence of data should be reported as a protection and planning gap, not as absence of children.

Conflict can close schools, displace families, destroy records and interrupt cohorts. Administrative denominators and school locations can become obsolete quickly. Household sampling may exclude insecure areas.

Results should identify coverage and interruption and avoid precise population rates where the denominator cannot be defended. Alternative and temporary provision needs recognition in the data system.

Household structure, illness and care responsibilities can affect attendance and completion. Categories should be defined without assuming a uniform pathway or cause.

Confidentiality is important in small groups. Analysis should link status to access to support rather than use a label as an explanation.

Work can compete with attendance, study and timely progression. Domestic work, often under-recorded, should be considered alongside economic activity. Contemporary household studies show the importance of analysing these relationships by sex.[REF-04]

Cross-sectional association does not establish whether work caused school difficulty or followed it. Time use and school conditions strengthen interpretation.

Mobility can break school records and place learners outside fixed catchment systems. Attendance may be seasonal or distributed across providers. A conventional annual enrolment status can misclassify continuity.

Tracking should follow the learner where feasible and recognise approved flexible provision. The completion event and record custodian must remain clear.

Where the next grade is offered only at a distant school, completion can fall even if early-grade participation is high. Grade availability should therefore accompany cohort flow.

Infrastructure counts should distinguish a school offering the full cycle from a site offering only initial grades. A nominal school place is not access to completion.

Late entrants should be compared with on-time entrants on promotion, repetition, attendance, completion age and learning, while accounting for background differences. Their lower completion, if observed, does not establish that age itself is the sole cause.

Policy should remove entry barriers and provide age-appropriate support rather than exclude older children to improve an age profile.

Repeaters are selected by prior performance, attendance or rules. Direct comparison with promoted pupils can therefore exaggerate or understate the effect of repetition.

The relevant evidence includes pre-decision status, support during the repeated year and later progression. Multiple repeaters require separate attention.

Disadvantages interact. A rural girl from a poor household who enters late may face a pathway not visible in any one average. Intersectional analysis can identify compounded patterns, but sample sizes fall as categories are combined.

Publication should prioritise policy-relevant combinations supported by sufficient evidence. Suppressed or pooled results should be identified.

Relative gaps can widen while absolute gaps narrow, or the reverse. Near universal completion, the remaining excluded share may be small in percentage terms but concentrated and difficult to reach.

Reports should choose measures according to the question and present enough information to avoid a one-sided conclusion.

The age distribution of completers shows whether groups reach the same result at different times. Equal eventual completion can coexist with substantial inequality in delay and its associated cost.

Median age and cumulative completion windows should be reported by group where age data are credible.

Total years spent in primary education can be distributed by completion status and group. Excess years may reveal repeated exposure without successful completion, while very short duration identifies early loss.

Interpretation should retain educational experience and learning. Efficiency terminology should not imply that a child’s time is expendable.

Missing age, disability, language or location data may be concentrated in particular schools or populations. An apparently favourable rate based only on complete records can be biased.

Each disaggregation should show missingness and whether the denominator is administrative, survey-based or estimated.

Rates for small groups fluctuate with a few events. Multi-year pooling can improve stability but may conceal recent change. Counts and uncertainty should accompany percentages.

Confidentiality may require aggregation. The chosen protection should preserve enough information for public action.

If girls and boys each have a completion rate of 50 per cent, parity has been reached but half of both groups remain without the measured outcome. A parity statement without the level is incomplete.

The same principle applies to every group comparison. Equality in deprivation is not fulfilment of the education commitment.

Both groups can improve while the faster-improving group pulls further ahead. Reporting only the national average or only the gap gives an incomplete account.

Trend tables should show group levels, counts, absolute gap and, where useful, parity ratio for the same years.

Targeted support can address a concentrated barrier, while universal improvements in teaching, places and records protect the entire cohort. Indicator evidence should determine the balance.

Targeting based on an unreliable category can exclude eligible children or stigmatise recipients. Eligibility, reach and completion effect require monitoring.

An equity finding should state population, group definition, indicator, year, difference, uncertainty and material context. It should avoid attributing cause unless the design supports it.

Action should correspond to the barrier observed: entry, attendance, instruction, progression, examination, cost, safety, support or data coverage.

Availability, accessibility, acceptability and adaptability provide a framework for interpreting the cohort pathway. A place that cannot be reached, instruction that does not provide an acceptable opportunity, or a system unable to adapt to a learner’s circumstances can appear in completion statistics through different mechanisms.[REF-12]

The framework prevents the indicator from being treated as the policy objective itself.

No national completion statement should be released without the most material available disaggregation and a note on excluded populations. Where the data cannot support a group estimate, the gap in evidence should be named and addressed.

Public reporting should neither conceal inequality behind an average nor overstate fragile small-sample differences.

Completion establishes that a learner remained through the defined cycle and satisfied the stated administrative or academic event. It does not by itself establish the depth, breadth or durability of learning.

The quality imperative places outcomes within a wider relationship among learner characteristics, context, inputs and teaching processes. Completion should therefore be joined to, but not replaced by, learning evidence.[REF-02]

Before interpreting assessment results, the system should establish whether the intended curriculum, instructional time, teachers and materials were available. A learner cannot fairly be held responsible for content that was not taught or for prolonged institutional closure.

Opportunity evidence also helps explain repetition and examination patterns without reducing them to pupil ability.

The intended curriculum states what learners should encounter. The implemented curriculum reflects what schools and teachers actually cover. Completion against the first cannot be assumed to show exposure to the second.

School records, teacher reports, observation and pupil work can identify material gaps. Comparative interpretation should note curriculum differences.

The attained curriculum is the knowledge, skills, values and dispositions learners demonstrate. Assessments sample part of it under defined conditions. No assessment covers the complete purpose of basic education.

A public statement should identify the domain, population, administration, standard and limitations rather than refer generally to “quality scores”.

An assessment administered in the final grade observes pupils who survived and attended on the assessment day. It may exclude earlier leavers, absent pupils and children never enrolled.

The result should not be generalised to the original child population unless the design and adjustment support that inference.

Coverage includes eligible schools, sampled schools, participating schools, eligible pupils and participating pupils. Exclusion and non-response can be related to disadvantage and performance.

Participation rates and exclusion rules are part of the finding. A precise mean based on incomplete coverage may be less informative than a qualified range.

Where completion requires an assessment standard, that standard should be documented and applied consistently. Changes in cut-score, examination form or grading can alter completion independently of underlying learning.

Trend releases should identify breaks and, where possible, provide a bridge study. They should not preserve an artificial continuous series.

Grade-level promotion criteria influence repetition and eventual completion. Standards may be national, local or teacher-based and may include subject performance, attendance or conduct.

Variation can be educationally justified, but its scope and consequences should be monitored. Unequal criteria can produce geographic or group differences unrelated to actual learning.

A minimum competency statement should identify the content and performance expected, the evidence supporting the threshold and the consequences of the decision. It should not be defined solely by the proportion passing.

Norm-referenced rank and criterion-referenced attainment answer different questions. Public language should distinguish them.

The 2006 Education for All monitoring report places literacy at the centre of basic education and lifelong participation. Completion without functional reading and writing raises a material quality concern, while literacy acquisition outside formal completion also remains socially valuable.[REF-01]

School completion and literacy status should be related carefully because age, language, assessment method and use context differ.

Numeracy is likewise fundamental, but basic education includes more than reading and mathematics. Health, civic understanding, practical knowledge, values and social development may form part of the intended outcome.

An indicator set should be selective and defensible. Adding many weak measures does not produce a complete quality account.

Assessment before and after repetition can show change, but maturation, additional instruction and test familiarity may contribute. The relevant comparison is with credible alternative support, not with no learning at all.

Reporting should include later promotion and completion. A short-term score gain followed by departure is not a complete result.

Promotion rates can improve while some pupils move forward without prerequisite learning. Early identification and support in the next grade are therefore central to reforms limiting repetition.

The response should not recreate retention through informal exclusion, unrecorded withdrawal or denial of examination access.

An assessment administered in a language not sufficiently accessible to the learner may combine subject knowledge with language proficiency. Interpretation should reflect the intended construct and language policy.

Translation can alter difficulty and meaning. Version preparation and comparability should be documented.

High-stakes examinations may narrow instruction, concentrate resources on likely candidates or encourage exclusion of lower-performing pupils. A rising pass rate may then coexist with lower candidate coverage.

The public account should show who reached, registered for, sat and passed the examination.

Schools may require a preliminary test or attendance threshold before allowing a pupil to sit the terminal assessment. Such selection changes the denominator and can move apparent failure outside the published pass rate.

Eligibility rules and excluded counts should be disclosed. A pass rate without candidate access is not an adequate completion indicator.

Average performance can rise while the lowest-performing learners make no progress. Reporting should show the distribution and relevant group differences, subject to the assessment design.

Achievement gaps should be read beside entry and survival gaps. A group with low school survival may have a selectively higher mean among the few assessed.

When lower-performing or disadvantaged pupils leave earlier, final-grade results are conditioned on survival. Comparisons across systems with different survival can therefore reverse the apparent quality ordering.

No simple adjustment removes all selection. Reporting assessment coverage and cohort survival together makes the limitation visible.

A composite might multiply population completion by the share of assessed completers meeting a standard. This can provide a rough joint measure only when populations, periods and assessment coverage align.

The product inherits uncertainty from both components and assumes unassessed completers resemble those assessed unless adjusted. Component results should remain visible.

Learning and completion are measured with error. Reporting several decimal places or a narrow rank can exaggerate distinctions that the data cannot support.

Rounding, confidence intervals and qualitative evidence should reflect the strength of the source. Uncertainty should inform decisions rather than prevent them.

School completion results reflect intake, transfers, local population, grade coverage and system rules. They should not be used for sanction or reward without data-quality review and contextual evidence.

Incentives attached to a narrow rate can encourage exclusion or misclassification. Balanced review should examine learner records and educational practice.

At system level, the relationship among entry, flow, age and learning can reveal structural bottlenecks. It should guide resource and policy decisions while preserving local variation.

The system remains accountable for enabling schools to improve. Publication should not transfer responsibility to teachers or families without evidence.

A quality finding states the completion result, learning evidence, opportunity conditions, distribution and limitations. It distinguishes observation from explanation and explanation from policy choice.

Where learning evidence is absent, the finding should say that completion quality cannot be fully assessed. It should not assume either satisfactory or poor learning.

The appropriate objective is not maximum promotion or minimum repetition in isolation. It is sustained participation, timely supported progression, successful completion and meaningful learning for the full child population.

Indicators serve this objective when they expose where opportunity is being lost and whether action improves the learner’s pathway.

A rate identifies the scale and location of a pattern. Diagnosis asks which institutional, instructional, financial, social or statistical mechanisms produced it. Action before diagnosis can move the indicator while leaving the underlying problem intact.

The sequence should be evidence, explanation, response, implementation and verification.

Low official-age entry with high later entry suggests delayed access; low entry across ages suggests broader exclusion; high administrative entry with low household attendance suggests weak participation or reporting.

Responses differ: places, admission reform, cost reduction, outreach, documentation, transport, language support or data correction.

Large first-grade repetition or departure can reflect difficult transition, language mismatch, class size, teacher allocation, weak attendance, curriculum pace or incorrect classification of entrants.

Review should begin with grade 1 records and classroom conditions. Expanding later grades alone will not correct early loss.

Loss in the middle of the cycle may relate to distance after a satellite school ends, rising cost, household work, declining relevance, teacher absence or accumulated learning difficulty.

Mapping grade availability and pupil movement can distinguish institutional structure from individual departure.

High final-grade repetition, examination exclusion or failure can prevent administrative completion after substantial participation. The examination, candidate rules, curriculum coverage and next-level capacity should be examined together.

Lowering the standard without addressing learning is not a quality solution. Nor is repeated retention without effective support.

High eventual completion with low timely completion indicates that the system retains or recovers learners but imposes substantial delay. The entry-age and repetition contributions should be separated.

Improvement should reduce avoidable delay while protecting routes for older and returning learners.

If attainment remains low even among older age groups, delay alone cannot explain the result. Entry exclusion, persistent departure and limited re-entry require direct attention.

The age at which attainment stops rising can help define the observation window, but cross-sectional evidence cannot locate every cause.

This pattern may reflect late entry, hidden repetition, interruption, inaccurate ages or a recent policy change. It does not prove manipulation.

Record linkage and entry-age evidence should precede a finding. Where the discrepancy remains, both measures should be published with uncertainty.

If repeated retention does not produce improved learning or progression, the decision and support model require review. The evidence should account for pupil selection and later outcomes.

Alternatives may include early diagnosis, additional instruction, revised language support, tutoring, vacation programmes or targeted progression plans.

Rapid growth may follow increased entry several years earlier, lower repetition, improved survival, examination change, catch-up cohorts or denominator revision. Attribution requires a cohort timeline.

Learning, staffing and class size should be monitored to ensure expansion maintains acceptable opportunity.

A fall may reflect conflict, economic stress, population movement, school closure, policy change, examination standards or data correction. The first task is to distinguish real cohort deterioration from statistical break.

Interim protection should focus on attendance, continuity and records while the cause is established.

Entry action can include abolition of unlawful charges, additional places, flexible documentation, accessible facilities, local-language communication, school meals, transport or community outreach. Selection should follow the identified barrier.

Monitoring should show official-age and later entry, attendance after entry and displacement from other providers.

Attendance measures may address safety, health, transport, teacher presence, scheduling, work conflicts or school climate. Registers must be reliable enough to identify repeated absence early.

The intervention should not punish households for absence caused by inaccessible or unavailable provision.

Structured diagnosis and additional teaching in early grades can prevent later repetition. Support should be aligned to the curriculum, available during adequate time and evaluated for learning and progression.

Separating pupils permanently without a route back can create another form of exclusion.

Teaching in or through a language learners understand, with planned transition where required, can strengthen early participation and learning. The design depends on teachers, materials and the linguistic context.

Monitoring should examine learning in the intended domains and later progression, not only initial attendance.

Reform can define limited decision criteria, require support before retention, prohibit repeated repetition, provide review and monitor later outcomes. A general prohibition may be appropriate in some frameworks but still requires learning support.

The transition year should be evaluated for hidden repetition, informal exclusion and changes in assessment practice.

Catch-up programmes can recover lost instruction after absence or closure. The programme should specify learning objectives, duration, staffing, assessment and re-entry to the regular pathway.

Participation without successful reintegration is an incomplete result.

Accelerated learning can enable older children to cover an equivalent curriculum in less calendar time. It requires an age-appropriate pedagogy rather than simple compression of the ordinary timetable.

Completion and transition should be recognised by the formal system, and the quality standard should remain explicit.

Calendar adjustment may improve attendance where agricultural cycles, weather or mobility create predictable conflict. It should preserve required instructional time and support continuity for pupils who move.

Local flexibility requires clear authority and records so that participation is not treated as closure or absence in national data.

Planning should identify whether every community has access to the grades needed for completion. Expanding initial grades without a feasible route to later grades can improve entry while creating future attrition.

Population projections, travel, boarding, transport and teacher supply should be evaluated together.

Teacher shortage or unequal deployment can produce large classes, combined grades, lost instruction and weak support. Counts should be converted into actual presence and teaching assignment.

Policy should monitor the effect on entry, repetition, learning and completion rather than assume that additional posts are filled and effective.

Assessment reform can align promotion and completion decisions with intended learning, improve feedback and reduce one high-stakes barrier. It must protect comparability over time and train those applying the new process.

The reform should be evaluated for candidate access, learning, repetition and distributional effect.

Scholarships, stipends, meals or removal of charges may improve entry and persistence where cost is causal. Coverage, eligibility, payment regularity and unintended exclusion require monitoring.

An association between poverty and non-completion supports inquiry but not automatic attribution of effect to one financial instrument.

Families need accurate information on entry age, entitlement, costs, attendance, progression and return routes. Communication should use accessible languages and channels and should not present repetition or leaving as inevitable for particular children.

Feedback from communities can identify unofficial requirements not visible in national policy.

Clear re-entry rules protect children after interruption and preserve prior learning. Age limits, documentation, assessment, grade placement and transition should be published.

The system should record re-entry separately to evaluate recovery and avoid inflating first-entry or repetition figures.

Every intervention should state the barrier, affected population, mechanism, resources, expected change, time and possible adverse effects. This connects implementation evidence to the completion pathway.

A general aspiration to raise completion cannot be evaluated without this chain.

The baseline should precede material implementation, use the same definitions intended for follow-up and show recent variation. A single unusual year can give a misleading comparison.

Where no credible baseline exists, the limitation should be stated and prospective measurement established.

Training delivered, classrooms built and payments made are process results. Entry, attendance, learning and completion are learner outcomes. Both are needed to understand whether implementation occurred and whether it worked.

Process completion should not be reported as corrected cohort performance.

Outcome improvement after an intervention does not by itself prove effect. Demography, policy, economy and other programmes may have changed. Comparison groups, phased introduction, prior trends or other designs can strengthen inference.

Where the design supports only contribution, the conclusion should use that language.

The same policy can be applied differently across districts and schools. Variation in reach, timing and quality provides evidence about feasibility and mechanism.

Analysis should avoid rewarding non-compliance merely because local outcomes were initially favourable; baseline and context must be considered.

Repetition reduction may increase unsupported promotion; examination reform may narrow curriculum; targeted benefits may exclude unregistered children; school consolidation may increase distance. Monitoring should identify these risks in advance.

An intervention should be revised where harm outweighs the intended improvement, even if the headline rate rises.

A pilot supported by exceptional staff or finance may not perform similarly at scale. Expansion should assess teacher capacity, recurring cost, data burden and institutional ownership.

Sustainability is not continuation at any cost. It is the capacity to maintain effective protection and learning under ordinary conditions.

Quality improvement is demonstrated when a defined population receives a better educational pathway and result under a credible measurement design. A changed administrative ratio is supporting evidence, not the complete judgement.

The review should remain open to revising both policy and measurement where implementation reveals a mistaken assumption.

The cases are constructed to test indicator interpretation. They do not describe named countries, institutions or observed determinations. Each uses only the information stated; missing evidence remains a finding rather than an invitation to invent a cause.

A six-grade system reports gross intake to the last grade of 104 per cent. A household survey shows that 58 per cent of children complete by age 12 and 79 per cent by age 15. First entry is widely distributed between ages 6 and 9, and the administrative record shows substantial repetition in grades 1 and 2.

The gross result does not show universal completion. Its all-age numerator includes delayed pupils from several cohorts and is compared with one official-age population. The household estimates indicate a material delay gap of 21 percentage points between the two stated age windows. They also show that later completion remains below universality.

The first inquiry is not whether to reduce the gross rate to 100. It is whether population and age records are credible, how much delay originates at entry, how much is added by early-grade repetition, and which children have still not completed by age 15. The report should publish all three results with their definitions.

Policy should address official-age access and early learning while maintaining routes for older learners. Excluding over-age pupils from the final grade would improve the age profile by denying opportunity and would therefore be contrary to the purpose of the analysis.

A system changes its rule so that pupils normally progress through the first three grades. Reported repetition falls from 14 per cent to 3 per cent in two years, and survival to grade 4 rises. A grade 4 assessment is introduced, but participation data and the performance of pupils formerly at risk of repetition are not reported.

The administrative changes are consistent with the policy, but they do not establish improved learning. Higher survival is a relevant benefit if pupils remain in meaningful instruction. The missing assessment coverage prevents a conclusion about whether promoted pupils received effective support.

Review should examine teacher guidance, diagnostic assessment, additional instruction, attendance and age-for-grade. It should also test whether schools use informal retention, unrecorded withdrawal or exclusion from the grade 4 assessment. A lower official repetition rate paired with hidden exclusion would not be improvement.

The release may state that formal repetition declined and grade 4 survival increased. It should reserve judgement on completion with quality until assessment coverage and later progression are available.

Administrative returns indicate that 92 per cent of primary-age children are enrolled. A household survey conducted during the school year reports 78 per cent attending during the reference period. The largest gaps occur in remote districts and among older primary-age children.

The values use different concepts. Enrolment can include intermittent or absent pupils; the survey period may also coincide with seasonal work or local closure. Population estimates, provider coverage and geographic assignment may differ. It would be unsound to label either figure correct before reconciliation.

The investigation should align age ranges, school-year dates, residence and school location; examine attendance registers; review survey coverage and confidence intervals; and identify whether administrative names remain active. Older pupils’ age-for-grade and work patterns require particular attention.

Public reporting should present the difference as evidence of weak effective participation if register checks support that interpretation. Completion forecasts based only on enrolment would otherwise overstate the cohort’s exposure to instruction.

Girls and boys have the same net intake rate. By grade 4, girls’ attendance falls and their repetition rises; timely completion is 11 percentage points lower for girls. Household evidence identifies distance, domestic work and safety concerns, while school records show fewer female teachers in the affected districts.

Entry parity is genuine but incomplete. The pathway locates divergence after entry and supports a multi-factor diagnosis. A single national parity index for enrolment would conceal the material completion difference.

Action may include safer access, attendance support, equitable teacher deployment and engagement on domestic workload, but each intervention needs a defined reach and mechanism. Repetition decisions should be audited to determine whether absence or learning evidence drives them and whether additional support is provided.

Monitoring should retain the two entry rates, grade-specific attendance and repetition, timely completion, eventual completion and learning. A rising completion parity ratio accompanied by falling completion for both groups would not be sufficient.

Removal of fees leads to a large intake that includes children from several older ages. Grade 1 classes expand rapidly. In the following year, reported repetition and pupil–teacher ratios rise, while later-grade enrolment has not yet changed materially.

The increase may represent a major improvement in access and a foreseeable capacity challenge. Comparing the new repetition rate with the prior year without recognising the changed cohort composition and class conditions would be incomplete.

The system should secure teachers, rooms, materials, age-appropriate organisation and early language and literacy support. Entry records must distinguish first entrants from repeaters when the large cohort returns. Planning should project its movement through every grade so that congestion is not merely transferred upward.

Evaluation should follow official-age and older entrants separately, while avoiding any lower expectation for older learners. The relevant test is whether expansion becomes sustained participation and learning, not whether the initial enrolment surge is preserved as a headline.

Survival to the final grade is 86 per cent among entrants, but only 68 per cent of registered final-grade pupils sit the examination and 74 per cent of candidates pass. The public report publishes only the candidate pass rate.

The published 74 per cent excludes pupils who did not become candidates and cannot represent completion among entrants or the population. If 100 final-grade pupils were registered under the stated proportions, approximately 68 would sit and about 50 would pass; rounding prevents an exact individual count. The candidate access gap is therefore material.

The system should identify eligibility, registration, absence and exclusion at each step. It should examine curriculum coverage and support without lowering a defensible standard solely to raise completion. Pupils denied access require a review route.

The release should publish survival, candidate coverage and pass rate separately and may present a combined flow only with aligned counts. The example also shows why percentages should not be multiplied casually when their reference populations differ.

Administrative data show repetition below 1 per cent in every grade. Mean recorded age rises by almost two years between grades 2 and 3, and household histories indicate frequent interruption and return. Schools record returning pupils as transfers.

The discrepancy is compatible with a classification problem, but the age pattern may also reflect age error, migration or different school coverage. It is evidence for validation rather than proof of hidden repetition.

The authority should sample individual histories, define interruption and transfer, test dates of birth and reconcile school identifiers. Re-entrants should receive a distinct status where feasible. Historical figures may require a break or quality note rather than an unsupported numerical correction.

Policy should protect return. Reclassifying pupils accurately is necessary for planning, but it should not create an admission barrier for children whose records are incomplete.

Several districts report gross last-grade intake between 110 and 135 per cent following internal displacement. School counts include newly arrived pupils, while district population projections pre-date the movement and allocate families to their former residence.

The numerator and denominator refer to different current populations. The rates cannot support district ranking or a conclusion of unusually strong completion. The school counts remain valuable for staffing and service planning.

The statistical body should produce an interim population range, publish counts, identify receiving and origin districts and use household or registration evidence where available. National totals may be less affected but still require review of cross-border movement and missing populations.

Learners should not be excluded from enrolment to restore a conventional ratio. The appropriate response is denominator revision, transparent uncertainty and resources aligned to actual attendance.

Household attainment suggests that eventual completion increased over successive age cohorts. Administrative reconstruction shows that average pupil-years per completer also increased because multiple repetition became more common.

The first result represents expanded educational attainment; the second identifies additional time and resource costs. They are not contradictory. The system may be retaining children who previously left, while progressing them slowly.

Analysis should separate productive recovery from avoidable repetition. It should compare learning and later progression for repeaters, locate grades of congestion and examine whether support is available during the additional year.

Policy should preserve the gain in eventual completion while reducing delay through earlier entry, attendance and instructional support. An efficiency reform that causes vulnerable learners to leave would improve pupil-years per completer by narrowing the population and would fail the public-interest test.

Final-grade assessment results improve over three years, but survival to the final grade declines and assessment participation among eligible pupils falls. Schools report increased pre-examination selection.

The assessed group has become more selective. The higher mean cannot support a conclusion that the original cohort learned more. It may reflect better teaching among survivors, stronger selection, or both.

The analysis should report entry, survival, examination eligibility, assessment participation and score distribution. Where feasible, earlier-grade assessment or household learning evidence can broaden the population view. Candidate rules require review for fairness and educational purpose.

Improvement is established only if learning rises without excluding pupils from the pathway or the evidence. A system should not exchange completion for a favourable assessed mean.

The cases demonstrate recurring denominator, timing and selection problems. A high gross rate can coexist with delayed and incomplete attainment; a low repetition rate can coexist with weak learning or hidden interruption; a high pass rate can exclude much of the original cohort; and a local rate can become uninterpretable after population movement.

The corrective principle is to restore the pathway. Entry, progression, age, candidate access, completion and learning should be connected through compatible populations and explicit uncertainty. Policy is then judged by the learner result, not by movement in the most favourable indicator.

The national statistical authority or designated education statistics body should maintain controlled definitions, classifications and calculation rules. Programme managers and local administrations should contribute operational knowledge without altering terms independently.

International comparison depends on mapping national data to common concepts while preserving national meaning.[REF-05]

Every completion-related indicator should have a stable identifier, title, purpose, formula, numerator, denominator, unit, level, disaggregation, source, frequency, revision rule and responsible owner.

Superseded definitions remain available with effective dates. A label should not be reused for a materially different formula without a version change.

Schools and providers should know the legal or administrative basis, required population, reference date and confidentiality rule for each return. Instructions should define entrants, repeaters, transfers, re-entrants and completers through examples.

Unclear categories generate systematic variation that cannot be corrected by later calculation.

Every field should serve a defined policy, statistical or accountability purpose. Excessive returns reduce accuracy and divert school capacity.

Core cohort fields deserve priority. New fields should be piloted and tested for consistent interpretation before national use.

Schools should reconcile enrolment by grade with new entrants and repeaters, review age and sex totals, confirm transfers and completion decisions, and explain material change from the prior year.

Validation should occur close enough to collection for source records to be corrected lawfully. It should not create pressure to reproduce expected values.

District or intermediate offices should test completeness, duplicate schools, implausible ratios, abrupt change and internal consistency. They should document corrections and preserve the original submission.

Comparison can identify outliers for review but should not define every outlier as error.

National review reconciles provider coverage, classification, school calendars, population denominators and regional totals. It compares administrative patterns with household and other independent evidence.

Release decisions should record unresolved issues, affected indicators and treatment in aggregation.

Provisional and final status should be visible. Revisions identify the value changed, reason, source, date and effect on prior conclusions.

Major denominator or definition revisions may require republication of the series. Silent replacement weakens trust and makes policy evaluation irreproducible.

Missing should remain distinct from zero, not applicable, suppressed and not collected. Imputation should be identified with method and uncertainty.

An aggregate should show coverage. A complete-looking table produced through undisclosed imputation is not a stronger public record.

Individual pupil information requires protection. Small cells may need suppression, combination or restricted access, particularly for disability, health, displacement or small communities.

Protection should be designed so that totals do not reveal suppressed values. Aggregate publication should still communicate material inequality.

Statisticians verify calculation and metadata; education specialists verify institutional meaning; population experts review denominators; and the responsible authority approves the public conclusion. Review roles and changes should be logged.

Institutional clearance should not be used to remove an unfavourable result supported by evidence.

The narrative should distinguish observed value, change, comparison, possible explanation and established cause. It should identify the population not represented and the consequence of uncertainty.

Terms such as universal, complete, on track and quality should be tied to an explicit rule.

Tables should include units, years, source and symbols. Charts should begin from a scale appropriate to the question and should not magnify small uncertain differences.

Country order may follow region or alphabet rather than rank where comparability is limited. A visual display cannot remove a definitional caveat.

Every population rate should identify the population source, reference year and revision. Gross measures state that pupils of all ages appear in the numerator; net measures state the age rule.

Where movement or an outdated census weakens the denominator, counts and ranges should be prominent.

A source note distinguishes reported, estimated, modelled and survey values. It identifies whether administrative data include private provision and whether household results measure enrolment or attendance.

The source organisation alone is not adequate metadata. The actual collection and reference period matter.

Comparability notes identify differences in cycle length, structure, programme mapping, school calendar, completion event, assessment and coverage. They should appear with the comparison rather than in an inaccessible general disclaimer.

A reader should be able to determine which claims remain valid despite each difference.

A correction notice states the original release, affected value or text, corrected result, reason, date and effect on conclusions. The original remains identifiable.

Corrections should reach users of downloadable tables as well as the narrative report. A revised file without notice is insufficient.

Researchers, schools and the public should have a route to query definitions, provide contrary evidence and report errors. Responses should distinguish explanation from correction and should be logged for recurring issues.

External scrutiny can strengthen the series while final responsibility remains with the publishing body.

Technical decisions should be protected from pressure to meet a target through classification or suppression. At the same time, methods, revisions and limitations should be publicly accountable.

Independence is demonstrated through transparent practice, not through isolation from education expertise or affected communities.

The release should move from data freeze through validation, calculation, analytical review, disclosure review, authorisation and publication. A calendar allows users to distinguish delay from absence.

Late data may be incorporated through the revision policy rather than through uncontrolled changes immediately before publication.

Definitions, source files, code or calculation records, published tables, corrections and methodological decisions should be retained. The archive enables reproduction and interpretation of historical cohort results.

Personal records require separate secure retention and access controls. Public reproducibility does not require disclosure of identifiable children.

The same controlled data should support planning, inspection, resource allocation and public reporting, with uses appropriate to accuracy. Parallel unofficial totals create confusion and opportunities for selective presentation.

Where different operational counts are necessary, their purpose and reconciliation to the statistical series should be documented.

National submission to an international collection should use the agreed classification and disclose deviations. The national public series and international value should be reconcilable through a mapping and revision record.

Differences are not necessarily errors; they may reflect common definitions, timing or later validation. The explanation should be available.

A national programme should produce a coherent account of entry, progression, delay, completion and learning for policy and public accountability. It should integrate existing records before creating new collections and should strengthen the capacity of schools to use accurate information.

The programme is statistical and educational. Its success is measured by better evidence and better decisions, not by the volume of returns.

The programme should answer how many children enter at each age; where promotion, repetition, interruption and departure occur; how long completion takes; which groups face different pathways; and what completers and non-completers have had the opportunity to learn.

Each question requires a defined population and cannot be answered by one common denominator.

The annual return should include schools and sites, grade and programme, total enrolment, first entrants, repeaters, re-entrants, transfers, age, sex, attendance status, completers and relevant teacher and grade-availability fields.

Definitions and reference date should be uniform. The return should distinguish a blank, zero and not applicable value.

Learner-level records support movement and duplicate checks but require identifiers, security, correction and transfer procedures. Aggregate returns impose less personal-data risk but limit cohort tracking and make residual flow more prominent.

The choice should reflect capacity and lawful purpose. A system should not collect identifiable data without the controls necessary to protect and use them reliably.

The school register is the source for attendance and status. It should record admission date, prior school and grade, first-entry status, date of birth and evidence, daily or periodic attendance, grade decision, transfer, departure, re-entry and completion.

Corrections should be authorised and traceable. Summary totals should reconcile with individual entries.

Single-year age populations should be maintained with source, base census, projection method, geographic boundaries and revision. Education statisticians should participate in review where school counts reveal plausible population inconsistencies.

Alternative denominator ranges may be appropriate after rapid movement or long intervals between censuses.

A periodic household module should capture age, current enrolment, recent attendance, age at first entry, current or highest grade, repetition or interruption where recall permits, completion, provider type and relevant background characteristics.

Questions require cognitive and field testing. The survey should identify persons outside the household frame to the extent possible through complementary evidence.

Attendance should be measured over an explicit period and related to the school calendar. The module can distinguish occasional absence, prolonged absence and formal departure.

Thresholds should serve early support and statistical analysis rather than automatic punishment. Absence caused by institutional closure should be separately recorded.

For each school year, an entry cohort file should contain new grade 1 entrants and their age distribution. The system should preserve the original cohort size and later additions due to verified transfer or delayed reporting.

Pupils entering primary education through another grade require a separate status so that first entry is not lost.

The transition file links or reconciles each grade’s prior population with promoted, repeating, transferred, re-entered, departed, unobserved and completed statuses. It records the date of extract and the rule for late decisions.

Unmatched cases are reported and investigated. They are not distributed automatically to make rates sum to 100 per cent.

The completion register identifies the event achieved: final-grade entry, course completion, examination eligibility, examination result and award. Dates and responsible decision bodies are recorded separately.

This separation enables accurate public definitions and protects former learners who later require evidence.

Where lawful and technically credible, assessment records can be linked to cohort status using protected identifiers. The analytical file should contain only information necessary for the research purpose.

Coverage and linkage failure should be reported by relevant group. Unlinked pupils should not silently disappear from the learning result.

The programme should map which grades and programmes each site actually provides, opening and closure dates, location, provider type and travel relationships. A school name alone does not show access to a complete cycle.

Changes in site codes or administrative boundaries should be versioned.

The calendar coordinates school census, attendance extracts, assessment, completion decisions, household fieldwork and population updates. It identifies which year label applies to each source.

Sufficient time should be allowed for validation without delaying information beyond its policy use.

Quality assurance should assess relevance, accuracy, timeliness, accessibility, comparability and coherence. It should document source risk and use independent checks proportionate to the consequence of the indicator.

A quality score should not replace the underlying findings. Different dimensions can move in different directions.

New cohort fields should be piloted in diverse settings, including small, remote, mobile and high-repetition populations. The pilot tests interpretation, burden, missingness, transfer and the ability to produce the intended indicator.

Revision should precede national expansion. A successful electronic form is not sufficient if source records cannot support the fields.

Training should use concrete pupil histories to distinguish first entry, repetition, re-entry and transfer. It should explain why accurate age and status matter for learners and planning.

Assessment of understanding is preferable to attendance at a training session. Guidance and help should remain available during collection.

Funding or performance consequences tied directly to enrolment, repetition or completion can influence reporting. Controls should separate data verification from target pressure and use multiple evidence sources.

The programme should monitor unusual clustering around thresholds and protect staff who report accurate unfavourable results.

Schools and districts should receive their validated data in a usable form: entry-age profile, attendance, grade flow, age-for-grade, completion and comparison with their own history. Contextual peer comparison may be used with caution.

Data extracted without local return weakens correction and improvement.

The annual report should present current period indicators and longer cohort or age profiles. It should distinguish provisional annual flow from outcomes that require several years of observation.

The report should include policy response, data-quality limitations and progress on earlier evidence gaps.

Approved researchers may require de-identified or secure access to analyse pathways and programme effects. Governance should define purpose, minimisation, security, publication review for disclosure and destruction or retention.

Access decisions should not depend on whether the proposed analysis is likely to produce favourable findings.

Priority studies include the causes and outcomes of first-grade repetition, effects of entry age, pathways after interruption, candidate exclusion, learning after promotion reform, and the distribution of time to completion.

Design should be selected according to the question. Descriptive monitoring and causal evaluation should not be conflated.

The programme requires sustainable staffing, school support, population expertise, survey capacity, secure systems and publication resources. Cost should be assessed against decisions improved and burdens reduced elsewhere.

Phased development should protect core accuracy rather than produce a nominally complete but unreliable national system.

The recommended sequence is to control definitions; improve first-entry, repeater and age fields; reconcile school coverage and population; establish annual transition files; align household evidence; then link learning where coverage is adequate.

This order protects the cohort foundation. A sophisticated learning composite cannot repair an unknown entry population.

The monitoring programme itself should be reviewed for accuracy, use, burden, disclosure risk and influence on decisions. Users at school, district and public levels should be consulted.

Fields or products that do not support a legitimate use should be revised or removed. Evidence needs may change, but historical continuity should be preserved through versioning.

Completion research should combine regular monitoring with studies designed for specific questions. Annual statistics identify scale and change; cohort studies identify pathways; surveys identify population distribution; classroom and institutional studies identify mechanisms; evaluation tests defined interventions.

No one design should be made responsible for every conclusion. Findings become stronger when different evidence converges under compatible definitions.

Descriptive work establishes who enters, moves, delays and completes, at what ages and under which school structures. Its quality depends on population coverage and clear measures rather than on complex modelling.

Description can establish a material policy problem without establishing its cause. The distinction should remain visible in executive conclusions.

Longitudinal records can examine the order of late entry, absence, repetition, transfer and departure. They permit time-to-event analysis and distinguish multiple episodes from the number of pupils affected.

Loss to follow-up should be reported and investigated. A cohort of continuously observed pupils may become progressively less representative.

Repeated household observation can capture children outside school and changes in work, residence and attendance. It is valuable where administrative identifiers are weak.

Panel attrition, migration and respondent burden require control. Linking to schools should occur only under lawful consent and data protection.

School studies can examine how admission, attendance, assessment and promotion rules are applied. Sampling should include schools with both expected and unusual indicator patterns.

Observation, document review and interviews should be triangulated. Stated policy and daily practice are separate evidence.

Learners and families can explain cost, distance, language, safety, work, illness, treatment and the consequences of repetition or interruption. Their accounts should be selected and analysed systematically rather than used only as illustrative quotations.

Participation should be voluntary and safe. Individual testimony should not be exposed in a way that creates school or community consequence.

Cases may be selected for similar completion with different repetition, similar repetition with different learning, rapid improvement, persistent delay or unusual data discrepancy. The selection logic should be stated before interpretation.

Illustrative selection cannot support prevalence claims. A diverse case set supports mechanism learning, not a global frequency estimate.

Comparison before and after a policy can identify change but remains vulnerable to trends, cohort composition, other reforms and measurement breaks. Several pre-policy periods and an unaffected comparison strengthen the design.

The policy start should reflect actual implementation, not only the formal announcement date.

Where implementation is phased for operational reasons, earlier and later areas may provide a comparison if selection and timing are understood. Ethical and legal obligations should not be delayed solely to create research variation.

Analysis should test whether areas differ in baseline, capacity or exposure to other programmes.

Statistical adjustment can examine associations among entry age, repetition, background and completion. It cannot repair variables measured inconsistently or unobserved selection by itself.

Models should report specification, missing-data treatment, uncertainty and sensitivity. Coefficients should not be translated automatically into causal policy effects.

Time-to-entry, time-to-departure and time-to-completion methods can use censored observations rather than treating every child not yet complete as a permanent non-completer. The event, origin and observation end must be defined.

Competing events such as transfer, interruption and death require appropriate treatment and careful public explanation.

Pupils are situated within classes, schools, districts and systems. Multi-level methods can separate some variation at these levels, subject to sample and model assumptions.

A school-level association does not establish that a school caused an individual result. Institutional evidence remains necessary.

Cross-system analysis should test whether an indicator represents the same event under different cycle lengths, promotion rules and examinations. Mathematical identity does not ensure substantive equivalence.

Where equivalence is limited, comparison may focus on within-system trend or on clearly separated indicator families.

A trend requires stable definition, source and coverage or a documented bridge. Policy and data reforms should be annotated on the series.

Revised historical values should be released as a new version. They should not be combined selectively with old values to produce a preferred direction.

Thresholds for concern can support timely review, but they should reflect indicator reliability, grade, population and consequence. A universal numerical threshold may be inappropriate where structures differ.

Crossing a threshold initiates inquiry; it does not establish a violation or cause.

A benchmark may be an earlier national result, a comparable group, an evidence-based standard or a policy target. Its basis should be named.

Benchmarking should not turn a conditional proxy into the final objective. A completion target remains subject to learning and equity.

Targets can focus action but can also encourage exclusion, reclassification or concentration on pupils near the threshold. The indicator set should include balancing measures such as assessment coverage, attendance and group completion.

Independent validation and public metadata reduce, but do not eliminate, incentive risk.

Entry age, early attendance, grade 1 learning and first repetition can provide early warning before a cohort reaches completion. They permit support while there is still time to change the pathway.

Their predictive value should be tested. A leading association should not become a fixed label on a child.

Completion and attainment are lagging outcomes that confirm whether the pathway delivered its intended result. Their delay makes them unsuitable as the sole basis for current management.

The annual account should therefore combine early signals with completed-cohort evidence and distinguish their time horizons.

Authorities often must act before perfect evidence exists. The decision record should state the risk of action and inaction, evidence strength, reversibility, affected population and review date.

Urgent support can be justified without making an unsupported causal claim. More consequential or irreversible action requires stronger evidence and fair procedure.

Completion evidence can guide teachers, classrooms, transport, learning support and re-entry services. Allocation rules should consider counts and severity, not only rates, and should avoid penalising accurate reporting.

Resources should address the diagnosed stage. Funding a final-grade intervention will not correct delayed entry unless it also reaches the earlier barrier.

Cost analysis may compare expenditure per entrant, per pupil-year and per completer, with distribution and learning. Costs borne by households and the time consequence of delay should be recognised where evidence exists.

Low cost per completer can reflect adequate efficiency or inadequate provision. Cost does not determine quality without the educational result.

Before adoption, a policy should be assessed for legal authority, affected population, evidence, implementation capacity, cost, equity, learning, data requirements and adverse effects. Alternatives and the reason for selection should be recorded.

An indicator target is not a policy design. The mechanism connecting action to the cohort pathway must be explicit.

Implementation review should occur early enough to detect exclusion or failure, while outcome review should allow sufficient time for the relevant grade or cohort result. Different measures therefore have different review dates.

The timetable should not postpone correction of a known data or rights problem until the final outcome is available.

Evidence can be graded by relevance, design, coverage, accuracy, consistency and proximity to the decision. A high formal design with weak implementation data may be less useful than a strong descriptive record for an immediate operational question.

The grade should accompany, not replace, the reasons for confidence and limitation.

Principal calculations and research results should be reproducible from preserved sources under appropriate access. Independent reanalysis can identify coding, assumption and interpretation errors.

Replication does not require public release of identifiable child data. Secure procedures can preserve both scrutiny and confidentiality.

An intervention may show no detectable effect, uncertain effect or adverse effect. These results should be reported and used. Selective publication exaggerates confidence and wastes resources.

A null result should be interpreted against statistical power, implementation and outcome timing; it does not always prove that the mechanism has no value.

A policy effective in one setting may depend on language, staffing, school structure, household conditions or administrative capacity. Transfer requires comparison of these conditions and adaptation of delivery.

The original effect estimate should not be represented as the receiving system’s expected result without a model and uncertainty.

The immediate research priority is to explain the joint distribution of late entry, repetition and completion rather than study each in isolation. Their sequence determines age, exposure and risk.

The strongest programme will connect population evidence to school flow and learning while keeping every denominator and assumption visible.

Research on late entry, repetition and departure concerns decisions with direct consequences for children. The design should minimise burden, avoid labels that follow a learner through school and ensure that participation or refusal does not affect educational treatment.

Questions should be necessary for the stated purpose and appropriate to age and language. Where a child’s account indicates an immediate protection concern, researchers need a lawful and communicated response procedure.

Consent arrangements should reflect the method, age, legal framework and sensitivity of the information. The child’s willingness remains important even where an adult or institution gives formal permission.

Participants should understand purpose, use, confidentiality limits and whether records will be linked. Broad permission for unspecified future use is not a substitute for accountable research governance.

Individual pathways can be identifiable through age, school, grade and rare circumstances even after names are removed. Access, analysis and publication should be designed around this risk.

Small cells, case descriptions and maps require review. Protection should not alter the substantive conclusion or erase a materially excluded population without explanation.

Linking school, assessment and household records can strengthen cohort evidence but increases privacy and error consequences. The responsible authority should define legal basis, fields, matching method, false-match review, retention, access and correction.

Uncertain links should not be treated as confirmed histories. Linkage quality should be reported by group to identify differential exclusion.

Data collected to improve completion should not be used automatically for immigration, policing, debt collection or punitive action unrelated to the educational purpose. Such use can deter enrolment and truthful reporting.

Any additional use requires independent legal and public-interest assessment. Institutional convenience does not establish compatibility.

Communities can identify whether categories, calendars and explanations reflect local conditions. Review should include persons affected by late entry, interruption and alternative routes, not only formal administrators.

Consultation does not transfer responsibility for statistical decisions. The publishing body should explain how material evidence was considered.

Statistical rules cannot anticipate every pathway. Trained professional judgement is needed for unusual re-entry, transfer and completion cases, under documented authority and review.

Judgement should not become unrecorded local discretion that produces unequal treatment. Reasons and classification consequences should be auditable.

A learner or family should be able to correct material personal records and challenge a progression or completion decision through the applicable process. Statistical correction and academic review are distinct but may depend on the same evidence.

Aggregate publication should not wait for every individual dispute, but unresolved cases and their possible effect should be assessed.

Recurring classification errors, repeated late reporting or common learner complaints indicate a system issue. Correcting the current table is insufficient if instructions, incentives or record design remain unchanged.

The annual quality report should record root causes, corrective action and whether the next cycle shows improvement.

Authorities should respond to material findings with a responsible office, action, timetable and follow-up measure. They should state when evidence does not yet justify a particular intervention.

Accountability is strengthened by accurate uncertainty. Overstated certainty can direct resources poorly and make later correction appear to be failure rather than statistical integrity.

The enduring standard is a traceable connection from child population to educational event, from event to indicator, from indicator to finding and from finding to proportionate action. At every link, definition, evidence and limitation remain visible.

This standard enables comparison without erasing national structure and supports public judgement without converting a statistical proxy into a complete account of education quality.

The international evidence available by August 2006 provides substantial information on enrolment, grade flow, last-grade intake, survival and selected population outcomes, but coverage and comparability remain uneven. Many series rely on population projections, aggregate school returns and reconstructed cohorts. Household evidence improves visibility of children outside school while introducing its own sampling, recall and classification limits.

The report therefore does not estimate one global timely-completion rate or one global effect of repetition. Such results would require aligned age histories, comparable completion events and evidence concerning children not observed in school systems. Published regional or country aggregates retain the definitions and coverage of their cited sources.

The relationship between repetition and later learning is also not settled by the descriptive indicators reviewed here. Repetition is selected by prior performance, attendance and institutional rule. A valid estimate of effect requires a credible comparison and evidence on the support received during the additional year. The report consequently treats repetition rate as a diagnostic signal and administrative outcome, not as proof of either quality protection or educational harm in every case.

The immediate evidence priority is consistent identification of first entrants, repeaters and completers by single year of age. Without these fields, late entry and repeated progression cannot be separated, gross final-grade intake remains difficult to interpret, and the population denominator cannot be connected to the cohort pathway.

The second priority is reconciliation of school returns with household attendance and attainment. This should focus on areas with the largest discrepancies and on populations poorly represented in conventional frames. Reconciliation must preserve source distinctions and should lead to improved provider coverage, age records, survey questions and population estimates.

The third priority is to connect completion with learning without selecting away excluded pupils. Assessment participation, eligibility and cohort survival should be published together. Research should follow pupils after promotion, repetition and return and should examine whether additional support changes learning and eventual completion. These priorities are feasible extensions of contemporary statistical practice and do not depend on a single universal data system.

Education authorities should commit to six actions: admit eligible children without avoidable delay; identify and support weak attendance and learning before repetition becomes the default response; retain age-appropriate routes for older and returning learners; publish completion as a defined cohort or population event; show distribution and learning beside the aggregate; and correct the public record when source or denominator revisions alter the result.

Statistical authorities should maintain definitions, population sources, revision histories and reproducible calculations. Schools should receive validated information that helps them act, not only submit returns. Households and learners should be able to understand the entry, progression and completion rules and to correct material records.

International comparison should assist this work through common definitions and transparent metadata. It should not reward a system for excluding difficult-to-serve pupils, suppressing repeaters or selecting examination candidates. The governing test remains whether a fuller share of the child population receives sustained, timely and meaningful basic education.

Completion is the result of entry, attendance, progression, time, support and learning. Each stage can fail, and the same final rate can arise from very different pathways.

A cohort account provides the organising discipline needed to connect these stages without pretending that one data source observes them all.

The first priority is not a more elaborate composite. It is reliable age, first-entry, repeater, grade, attendance, completion and population information, with definitions that remain stable and visible.

Administrative and household evidence should be reconciled and their distinct purposes preserved.

Completion should be interpreted with opportunity to learn and valid learning evidence. Promotion without support and repetition without effective additional learning are both inadequate responses.

The system should protect timely progress while retaining credible routes for late entrants, returning learners and those requiring additional assistance.

National progress must be examined across sex, residence, resources, language, disability and other circumstances relevant to the system. Entry parity does not establish completion parity, and completion parity at a low level is not fulfilment.

Counts, rates, delay and learning together provide a more truthful distributional account.

Every completion claim should identify event, population, period, source and uncertainty. Missing data and incompatible definitions should constrain the conclusion rather than disappear in aggregation.

Public confidence depends on corrections, historical versions and an accessible explanation of how the rate was produced.

The public objective is that every child completes basic education with meaningful learning and without avoidable delay. A completion rate is authoritative only to the extent that its cohort, event and evidence are authoritative.

The proper use of indicators is therefore not to declare success from one number. It is to identify where children are excluded, delayed or moved forward without sufficient learning, and to verify whether action improves the complete educational pathway.

For that purpose, a credible annual account should preserve counts as well as rates, publish timely and later completion separately, identify the grade at which delay accumulates, and disclose which children and providers remain outside the evidence. Improvement should be confirmed across successive observations and should remain visible after population, definition and coverage changes are taken into account. Where the evidence cannot yet establish the result, the responsible conclusion is a defined uncertainty accompanied by a programme to resolve it.

The audit determines whether the national completion account represents relevant populations, measures their pathways consistently and avoids conclusions that conceal exclusion or delay. It applies to data, analysis, publication and policy response.

List children represented in administrative schools, household frames, alternative provision, displaced settings and no recorded provision. Identify groups likely to be missing or misclassified and the evidence used to estimate their importance.

The map should include geographic and provider exclusions, not only demographic categories.

Every group variable has a source, purpose and permitted categories. Administrative and survey categories should not be merged without a crosswalk. Changes over time are versioned.

Definitions should avoid treating a social category as a cause. The analysis examines barriers and institutional conditions.

For each group rate, verify that the denominator covers the population eligible for the outcome and that numerator classification is compatible. Unknown group status is reported separately.

Using enrolled group counts as denominator for population completion converts exclusion before entry into invisibility.

The audit examines entry age, attendance, promotion, repetition, interruption, re-entry, completion age and learning for the same groups where evidence permits. It identifies the first stage at which a material difference appears.

A final gap should not be assigned to entry barriers where divergence begins after several grades.

Group levels, counts, percentage-point gaps and parity ratios are calculated from unrounded values. The audit notes where a relative measure is unstable or an absolute measure conceals proportionate inequality.

No result is released as parity without the underlying levels.

For each group, report completion by the same age windows. Compare both timely levels and the gain between windows. This distinguishes persistent non-completion from delay.

Age and attainment error should be examined by group because data quality can itself be unequal.

Compare assessment participation and cohort survival by group. Where a group has lower survival, the assessed mean among survivors should not be generalised to the original population.

Candidate exclusions, absence and language accessibility are part of the audit.

Select intersections supported by policy relevance and adequate sample or counts. Report the number represented and avoid extensive searching for extreme differences without correction or replication.

Where disclosure or precision prevents publication, the need should still inform secure analysis and data improvement.

Calculate missing group status by school, location, provider and outcome. Test whether records with missing values have different entry, progression or completion patterns.

Complete-case estimates should be labelled and should not be treated as population results without justification.

Severity considers the size and number affected, duration, stage of education, learning consequence, compounding barriers and strength of evidence. A small percentage gap may be severe where it represents complete exclusion or a large population.

Severity should not be reduced because the national average is improving.

For each material finding, the audit identifies the responsible level and policy mechanism: places, cost, documentation, instruction, language, safety, attendance, assessment, re-entry or data coverage. It states whether the response is universal, targeted or both.

The response should include reach and the risk of new exclusion.

Follow-up uses the same group definitions or a documented bridge. It reports whether levels improved, gaps changed and adverse effects emerged. National change and group change are shown together.

Where the intervention reached only part of the eligible group, effect and coverage remain separate.

The public account should state what is observed, how reliable it is, which mechanism is supported and what action follows. It should avoid deficit descriptions of children and communities.

The responsibility of education institutions remains visible even where household circumstances affect participation.

Closure requires corrected data or a documented limitation, publication of the material finding, an assigned response, and a date for outcome review. Lack of sufficient evidence can close a specific statistical claim only after a plan addresses the evidence gap.

All values in this appendix are constructed. They demonstrate formulas and interpretation and are not country, regional or global estimates. Totals are selected for transparent arithmetic; rounding is stated.

Assume 10,800 new entrants of all ages enter the final primary grade and the population of official entrance age is 10,000. The gross intake rate is 108 per cent. If 8,900 of the new entrants are of official age, the corresponding net rate is 89 per cent.

The 19-point numerical difference does not equal the share of late entrants because the gross and net numerators are divided by the same official-age population, not by all entrants. The entrant age distribution is required.

Of the 10,800 new final-grade entrants, suppose 8,900 are official age, 300 are younger and 1,600 are older. The shares are approximately 82.4, 2.8 and 14.8 per cent respectively after rounding.

These shares describe current entrants. They do not show what proportion of the official-age population will eventually enter the grade.

Suppose grade 2 enrolment in year t is 5,000. In t+1, 3,900 of those pupils are represented among new grade 3 entrants, 700 repeat grade 2, 100 are confirmed transfers outside the observed unit and 300 are otherwise unobserved.

Promotion is 78 per cent and repetition 14 per cent. Confirmed transfer is 2 per cent and unobserved status is 6 per cent. Calling the full 8 per cent “drop-out” would merge confirmed transfer with unobserved cases.

If grade 2 enrolment in t+1 is 4,700 and contains 700 repeaters, repeaters are 14.9 per cent of current enrolment. The flow repetition rate remains 14 per cent because its denominator is the prior-year 5,000.

The values are both correct for different questions. They should not be compared as if one were a revised version of the other.

In a survey-weighted population of 20,000 children in an eligible cohort, 12,400 completed by the timely age and another 3,000 completed by the later window. Timely completion is 62 per cent; later cumulative completion is 77 per cent; the delay gap is 15 percentage points.

The remaining 23 per cent includes those who may complete after observation unless the window is demonstrably final.

Suppose timely completion is 66 per cent for girls and 72 per cent for boys. The female-to-male parity index is approximately 0.917 and the female-minus-male gap is minus 6 percentage points.

If both rise to 75 per cent, the parity index becomes one and the gap zero, while one-quarter of each group still lacks timely completion.

A constructed cohort accumulates 6,900 pupil-years and produces 850 completers from an official six-year cycle. Pupil-years per completer are approximately 8.12. The additional 2.12 years relative to cycle duration reflect repetition and years accumulated by non-completers.

The result does not show that every completer took 8.12 years or that all additional instruction lacked value.

From 1,000 original entrants, 800 reach the final grade, 600 sit the terminal examination and 480 pass. Survival to the final grade is 80 per cent; examination participation among final-grade pupils is 75 per cent; pass rate among candidates is 80 per cent; recorded passes equal 48 per cent of original entrants.

The two 80 per cent figures have different denominators and meanings. Publishing only the candidate pass rate would conceal half of the entrant cohort not recorded as passing.

If 9,500 new final-grade entrants are divided by a projected official-age population of 9,000, the gross rate is about 105.6 per cent. If a defensible population range is 9,000 to 10,000, the rate range is 95.0 to 105.6 per cent.

A conclusion of universal completion cannot rest on the point estimate alone, particularly because final-grade entry is a proxy event.

Suppose reporting countries represent 82 per cent of the region’s primary-age population. Their weighted completion rate is 74 per cent. No rate is available for the remaining 18 per cent.

The 74 per cent is a covered-population result, not automatically the regional rate. Imputation would require a method and uncertainty.

Rates should be calculated from source counts. Multiplying rounded component rates can produce a different combined result. In the candidate example, use 480 divided by 1,000 rather than rounded survival, participation and pass percentages if exact counts are available.

For every worked or observed calculation, the reviewer signs off the event, denominator, time, coverage and claim. A correct quotient with an incorrect label fails the review.

The audit covers metadata, source data, calculation, tables, narrative, downloadable files, corrections and policy response. It is performed before publication and repeated after material revision.

The release identifies the frozen source versions and date. Late submissions are logged and handled under the revision policy. Source files are read-only within the release record.

The freeze prevents unexplained value changes during drafting while preserving a controlled route for correction.

An independent reviewer reproduces principal counts, rates, aggregates and constructed cohort results from the frozen sources and documented rules. Differences beyond the rounding tolerance are resolved.

Reproduction includes missing-value and suppression treatment, not only visible numbers.

Every table and indicator has definition, unit, population, level, period, source, status, coverage and limitation. References resolve to the source available by the evidence cut-off.

Titles and short labels are checked against the formula, particularly completion, graduation, drop-out and survival.

The evidence register is reviewed for publication and availability by 9 August 2006. Later revisions, concepts, estimates and events are excluded from the report’s analytical claims.

Where a current archive hosts an historical document, the document’s original issue and content, not the archive’s present interface, determine eligibility.

Every number in narrative text is traced to a table, source or labelled constructed example. Percent and percentage-point language is verified. Denominators and reference years are repeated where ambiguity is possible.

No rank or causal claim is released unless the design supports it.

Visible Markdown tables, structured reader blocks and downloadable CSV files must contain identical labels, row order and cell values. Character encoding, quotation, line breaks and missing symbols are tested.

No orphaned table file remains, and every visible table has one downloadable counterpart.

The review checks for promotional language, false certainty, repeated boilerplate, institutional overclaim, hidden production instructions and language suggesting technology or policy not available at the cut-off.

Normative statements identify the responsible party and evidential basis. Limitations are placed with the relevant conclusion.

The report structure should follow the research problem rather than a standard sequence imposed across the corpus. For this report, the required logic is policy event, cohort movement, age, repetition, source reconciliation, completion measures, distribution, learning, diagnosis, applied cases and statistical governance.

Sections that merely restate earlier conclusions are removed or consolidated.

Exact substantive paragraphs are prohibited. Near-duplicate review compares wording within the report and against previously released corpus reports. Shared controlled terms and source titles are distinguished from repeated analysis.

Material overlap is rewritten to fit the specific indicator question or removed.

Each source is used within its authority and date. Global monitoring figures are not converted into national findings; a regional or comparative report is not treated as a binding rule; a rights instrument is not presented as a statistical method.

Every registered source either appears in the public text or has a stated evidence role retained for audit.

The report does not approve, rank or determine a named education system. Constructed cases are labelled and do not resemble undisclosed country findings. Recommendations remain general and require national legal and statistical application.

Tables with long text, numerical alignment, missing symbols and citations should be sampled in the target reader. Headers, wrapping and horizontal behaviour must preserve meaning. CSV opening is tested in a UTF-8 compliant application.

The visual test does not replace cell-level parity and schema checks.

The authorised reviewer records pass or fail for every gate, unresolved warning, version identifier and release date. A failed temporal, numerical, boundary, duplication or rendering gate prevents release.

Passing the automated audit does not replace manual review of evidence and prose.

Queries, corrections and new source information are logged. A correction is linked to the original release and propagated to narrative, structured tables and downloadable files.

Later evidence may inform a later report; it should not be inserted silently into the historical 2006 version.