What the term covers
Learning analytics means collecting and interpreting data about learners and their activity in order to understand and improve learning. For an open course team, it usually means answering practical questions: where do people get stuck, which materials are used, and which parts could be clearer?
It is not a magic lens. Data can point to where to look, but it rarely explains why something is happening, and it always leaves out the learning that happens elsewhere. This guide gives general information for course teams and curious learners. It is not legal or technical advice.
Data that teams commonly have
What is available depends on the platform and the course. Typical items include:
- Enrollment and activity. Sign-ups, logins, and which pages or videos were opened.
- Progress. Which units were completed, and in what order.
- Assessment results. Quiz scores, number of attempts, and which questions were missed most.
- Video behavior. Where viewers pause, replay or stop.
- Forum activity. Posts, replies and unanswered questions.
- Survey responses. Collected deliberately, as in research on student feedback surveys.
- Technical details. Device type, browser or connection problems.
Many learning management systems provide a dashboard with some of these. Small teams can get a lot from simple exports and spreadsheet basics.
Questions worth asking
Data is most useful when it starts from a question. Some examples:
- Where do most learners stop? A sharp drop at one point may show a confusing step or a sudden jump in difficulty.
- Which quiz questions are missed by most people? That may mean the wording is unclear rather than the content being hard.
- Which resources are never opened? They may be hard to find or not needed.
Writing down a question before opening a dashboard helps avoid wandering through numbers.
Reading the numbers carefully
Completion rates are not a verdict. In open courses, many people enroll out of curiosity, to look at one topic, or to keep the materials. A low completion rate may reflect that mix rather than poor teaching. It is more informative to ask whether the people who intended to finish were able to. Related ideas appear in discussions of student retention and course completion.
Averages can hide things. A mean quiz score of 70 per cent might reflect everyone scoring around 70, or half scoring near 100 and half near 40. Looking at the distribution, not just the average, reveals more. Charts help, but misleading charts, such as truncated axes, can distort a picture.
Small samples are noisy. With 20 learners, a difference of two or three people can look large. Be cautious about drawing conclusions from small groups, and look for patterns that repeat across several runs.
Activity is not learning. Time on page, clicks and logins measure activity. A learner may leave a video playing in another tab, or learn the material quickly with few clicks. Treat these as hints.
Correlation is not cause. If learners who post in the forum finish more often, it does not follow that forcing everyone to post would raise completion. Motivated learners may both post and finish. This familiar problem is at the heart of correlation versus causation.
Missing data matters. Learners who quietly left, who used blockers, or who studied offline may be absent from the figures. What you see is a partial picture.
From insight to action
When a pattern appears, a sensible sequence is:
- Check that it is real: is it present across several cohorts, and is the sample large enough?
- Look for an explanation from other sources, such as comments, forum posts or a quick conversation with a few learners.
- Make one small change.
- Watch whether the same measure moves, remembering that other things may change at the same time.
Privacy limits
Learner data is personal data, even when it looks dull. Combined, clicks, times and locations can reveal a lot about a person’s habits and circumstances. Some general principles apply for responsible teams:
- Collect only what you need for a stated purpose. If you cannot say why a field is useful, do not gather it.
- Tell learners. A clear, plain-language notice should explain what is collected and why. Learners should not have to guess.
- Prefer aggregated or anonymized results when reporting. Be aware that small groups can be identified even without names.
- Keep data only as long as needed, and store it securely.
- Be careful with children’s data, where stricter rules and expectations commonly apply.
- Be wary of trackers and third-party tools. Web trackers can send learner information to outside companies, so check what any added tool does.
Laws differ by country and by type of institution. In many places data protection rules apply, and the details, including rights of access and deletion, should be checked with official sources or a qualified adviser rather than inferred from a general guide like this one. Learners can also look after themselves by reviewing what they share, understanding their digital footprint and checking privacy settings.
Fairness and open values
Analytics in an open education setting raises its own questions. Open courses aim to widen access, so data should not be used to quietly screen out learners who seem unlikely to succeed. Predictions based on past learners can reproduce past inequalities. Labels such as “at risk” should lead to offers of help, not to being ignored, and humans should review any decision that affects a learner.
Starting small
A team does not need sophisticated tools. A reasonable first step is to pick three questions, such as where people stop, which quiz items are confusing and which resources go unused, find the simplest data that addresses them, and review them after each course run. Combine the numbers with what learners tell you, and treat conclusions as provisional. Data is a helpful companion to judgment, not a substitute for it.