Two million learners missing from KEMIS: Gov’t should not confuse data errors with ghost pupils

The reported disappearance of approximately two million learners from the Kenya Education Management Information System (KEMIS) has raised serious questions about the accuracy of school enrolment records, the effectiveness of digital registration and the capacity of institutions to manage learner information.

However, as the government intensifies efforts to account for the missing records, education stakeholders must be careful not to equate learners who are absent from the digital database with children who do not exist in schools.

The distinction is critical. A learner can be physically present in a classroom, appear in the school's admission register, attend lessons and undertake assessments, yet remain uncaptured in the national database because of administrative mistakes, inconsistent personal information or technical difficulties.

The emerging debate therefore calls for a comprehensive investigation into the circumstances surrounding the registration gap, rather than an immediate assumption that institutions have enrolled ghost learners.

One of the fundamental challenges is poor coordination of learner information. Some heads of institutions do not adequately connect different sources of data, making it difficult to reconcile records held by schools, parents and relevant government agencies.

The problem often begins with the collection of personal information. Some institutions fail to obtain accurate details from parents and guardians, while others do not sufficiently verify the information before entering it into KEMIS. Errors that could have been identified and corrected at the start then become major obstacles during registration.

Another challenge is the way some institutions handle digital registration. Peter observes that certain heads of institutions entrust the exercise almost entirely to cybercafé operators without adequate supervision. Although seeking technical assistance is not necessarily wrong, external operators may not have sufficient knowledge of individual learners, school admission records or previous examination details.

The situation is further complicated when administrators delegate registration to teachers and other staff without providing the necessary resources. Some employees are expected to complete the exercise without internet bundles, reliable devices or adequate technical support. Such an approach undermines accountability and makes it difficult for those assigned the responsibility to meet the required deadlines.

Accuracy remains one of the most significant obstacles. Digital registration systems depend on consistency, and even minor differences in personal details can trigger rejection. For instance, entering a learner's name as Peter Yego when the corresponding record reads Yego Peter can create a mismatch, even where both entries refer to the same individual. Differences in spelling, name arrangement and other identification details can complicate verification.

The relationship between KEMIS and Kenya National Examinations Council (KNEC) records presents another challenge. Learner names, spellings and assessment numbers must correspond with the information maintained by the examination body. Where discrepancies occur, institutions may struggle to complete registration despite having the actual learners in their classrooms.

This makes it necessary for education authorities to establish effective mechanisms for reconciling records across the different databases, instead of leaving institutions to struggle with rejected entries.

Time constraints also contribute to the difficulties. Some schools begin registration without adequate preparation, while others lack sufficient personnel and technical capacity to complete the exercise within the stipulated timelines. Where thousands of learner records require verification, even minor inconsistencies can consume considerable time.

The reported registration gap therefore raises a fundamental question: is the challenge entirely about missing learners, or are weaknesses in data collection, institutional coordination and digital verification also responsible?

Protecting public resources without blanket suspicion

The government has a legitimate responsibility to protect public resources and ensure that capitation is based on genuine enrolment. Duplicate records, inflated numbers and deliberate falsification must be investigated and addressed wherever evidence establishes such irregularities.

However, the existence of possible irregularities should not lead to a blanket assumption that every learner whose information is missing from KEMIS is fictitious. A database rejection is not proof that a learner does not exist. It is an indication that the information submitted requires further verification.

The Ministry of Education should undertake a comprehensive reconciliation exercise involving heads of institutions, parents, teachers, education officers and relevant examination authorities. Schools should be allowed to present admission registers, attendance records, assessment documents and other supporting evidence to establish the presence of learners whose details have not been successfully captured. There should also be clear procedures for correcting variations in names, spellings and assessment numbers, particularly where genuine learners are affected by inconsistencies between institutional and examination records.

The government must equally examine the operational challenges facing schools. Reliable internet connectivity, adequate devices, trained personnel and sufficient time are essential for successful digital registration. Expecting institutions to deliver accurate results without these resources risks turning an administrative exercise into an avoidable national crisis.

Heads of institutions must also take responsibility for the quality of information submitted under their supervision. Delegating the exercise does not transfer accountability. School administrators should ensure that learner details are collected accurately, verified against available records and entered by competent personnel.

Beyond the immediate registration exercise, the issue exposes the importance of investing in proper data management across the education sector. Digital systems are only as reliable as the information entered into them. Even the most sophisticated platform cannot guarantee accurate national records if institutions provide incomplete, inconsistent or incorrectly arranged information.

The financial implications cannot be ignored either. Where learner registration determines institutional capitation, genuine pupils whose records remain uncaptured could expose their schools to funding difficulties. Ultimately, children may bear the consequences of mistakes committed during an administrative process over which they have little control.

Establishing the facts

The government must therefore approach the two-million-learner discrepancy as both a data integrity challenge and an education access concern. It must establish how many learners are genuinely absent, how many have not been registered because of administrative weaknesses and how many records are affected by technical or identification mismatches.

KEMIS is intended to strengthen accountability, improve education planning and ensure that every learner is properly accounted for. Its success should not be measured merely by the number of records uploaded but by the accuracy, completeness and fairness of the information captured.

The most important question is not simply how many learners are missing from the system, but why their records are missing and what can be done to correct the situation without disadvantaging genuine pupils.

Before the government declares two million learners missing or suspects schools of having ghost pupils, it must establish whether the children are actually absent or whether human errors, inadequate resources, poor coordination and database inconsistencies have made their records invisible.

A missing learner record is not necessarily a missing learner. Sometimes, the problem is not the child but the data.