Imagine a hiring manager based in Lagos who has two applications to examine. Both of the applicants completed a brief online training course for a data analysis certification. One of the applicants obtained her certification by watching eleven hours of online videos and passing an online exam that she could repeat as many times as needed. The other spent three weeks cleaning an inconveniently written sales file, discovered that there was a branch with ten times as high sales because of a mistake with an extra zero, and explained her conclusions to an assessor during an online meeting. Both applicants acquired the same certificate.

Nigeria is about to decide which of them its universities will issue. On 12 October the National Universities Commission reported on a committee meeting in Abuja, held in late September, on national micro-credential guidelines drafted with the Commonwealth of Learning. The Commission’s executive secretary, Professor Abdullahi Yusufu Ribadu, told the committee that micro-credentials should sit alongside degrees as flexible pathways that tighten the link between universities and work. Assessment, quality assurance, credit structures and recognition are all part of the framework.

Of those four, assessment will decide whether anyone outside the issuing university cares. A micro-credential exists to answer one question for a stranger: what can this person now do? A completion record answers a different question, which is whether the person finished. A university that puts its name on a certificate without evidence of performance is lending its reputation to something it has not checked, and I think the guidelines should say so plainly.

The trust problem is already visible elsewhere. In a US survey of employers reported by Inside Higher Ed, 46 per cent said they did not know what to make of the quality of a non-degree credential on a résumé, and 42 per cent were unsure which skills it showed. A Canadian study by the Future Skills Centre found that only 12 per cent of workers said a micro-credential had helped them get a new job or a promotion. Nigeria is drafting its rules early enough to avoid that confusion, if it chooses to.

What a data analysis certificate should show

Take the short online course in data analysis, an obvious candidate for a professional micro-credential. The student has to understand what the dataset means, and what it does not mean, as well as what is wrong with it, duplicates, incorrectly entered numeric values and so on. Furthermore, the student should be able to analyze a massive amount of data in weeks to form a reasonable suggestion to those who have a limited budget and to justify their reasoning. The quiz can be designed to test the student’s understanding of the pivot table concept, but it will not reveal much more.
So the assessment has to look like the work. Tasks should resemble what an analyst is handed on a Monday, scored against a rubric that learners can read before they start. Employers should help write those tasks. The US survey found that 53 per cent of employers called their involvement in designing non-degree credentials a necessity, while 44 per cent said no college had ever invited them to collaborate. A Nigerian university can do better by asking three or four employers in a field to supply anonymised files and review the marking criteria before the first learner enrols. They know which mistakes cost money.

The online problem

Online delivery makes the obvious objection harder to answer. A take-home analysis can be done by someone else, or by an AI assistant, and proctoring software is expensive and a poor fit for learners on unreliable connections. There are better options. Give each learner a different dataset. Ask for a five-minute recorded walk-through in which the learner explains one choice and an assessor questions it. Keep files small enough to download on a weak connection so that the work can be done offline. None of this is foolproof, and no assessment ever was.

It also costs more. Marking a performance task takes longer than scoring a quiz, and a course serving thousands of learners cannot have a professor read every submission. Structured outputs can be checked automatically, with trained assessors reviewing a sample and every borderline case. Pricing courses to cover assessment will make them less cheap, and that is the right trade.

What the certificate should carry

The credential itself should say what was assessed. One page, readable by an employer in two minutes, can cover the outcomes, the tasks, the standard required, who marked the work and, with the learner’s consent, a link to the work itself. A certificate that cannot say what its holder was asked to do should not carry a university’s name.

Credit raises the same issue. If micro-credentials are going to count towards degrees, the credit has to be earned by assessed performance. Credit awarded for attendance is currency printed without backing, and degrees would absorb the inflation.

The guidelines are still in draft and I have not seen the text, so none of this is a criticism of what they contain. The pilot institutions will learn quickly which assessment methods survive thousands of real learners. The National Open University of Nigeria, one of them, began work on its own institutional policy in early October, and the Commission has said the lessons from pilots will refine the national guidelines before wider implementation.

There is time, then, to ask the questions in the right order: what must a learner be able to show, who will check it, and only then how many people can enrol.