Two directions of feedback
Most writing on feedback in education concerns comments passing from teacher to learner. This article looks at the opposite direction: what students tell course teams about the course. For an online or open course, which may have no classroom where a teacher can sense confusion, this is one of the few direct ways to learn what is working.
The research on student opinion is large, mostly from conventional higher education, and generally cautious in tone. Its main message is that student feedback is valuable information but a flawed measure, and works best when read alongside other evidence.
Common ways of collecting student feedback
End-of-course surveys. The most widespread method. They are cheap and easy to compare across courses but arrive too late to help the people who answer them.
Mid-course check-ins. A few short questions part-way through, such as “What is most confusing so far?”, let the team respond while the course is still running. Many practitioners prefer these for improvement purposes.
Short pulse questions. A one-line rating at the end of a lesson shows which parts cause difficulty.
Open comments and forums. Free text and discussion threads show reasons that ratings cannot. They are harder to summarize.
Interviews and focus groups. Small in number but rich in detail. They suit smaller courses or targeted questions.
Behavioral signals. Where learners stop, replay or skip may reveal problems no one reports. These are discussed in the article on learning analytics.
What surveys are good for
Used sensibly, surveys can reveal:
- Unclear instructions or confusing wording.
- Technical problems such as video that will not play.
- Pace that is too fast or too slow for many learners.
- Accessibility barriers, if the survey asks about them.
- Areas where learners feel unsupported.
These are practical, fairly concrete matters that learners are well placed to report. A single comment can lead to a quick fix, and repeated comments are a strong signal.
Known limits
Research on course evaluations has repeatedly pointed to weaknesses. The main ones apply equally online.
Who responds. Online courses often have low response rates. People who finish, and people with strong views, tend to be over-represented, while those who quietly left are rarely heard from. Because they may have had the most important problems, ignoring them can mislead. A rating of “4.5 out of 5” from a small, self-selected group says little about everyone who enrolled.
Satisfaction is not learning. Studies in higher education have generally found only a weak or inconsistent relationship between how learners rate a course and how much they learn. Courses that feel hard can be rated lower even when they teach well, and pleasant courses can be rated highly without much learning. Treat ratings as measuring experience, not outcomes, and see wider discussions of course evaluation and measuring effectiveness for other approaches.
Bias. Research on ratings in conventional teaching has reported that factors unrelated to teaching quality, such as instructor characteristics or the subject, can affect scores. Evidence on this is not uniform, but it is a reason to be careful when comparing instructors or using ratings for staffing decisions.
Question wording. Vague or leading questions produce vague or biased answers. “Did you enjoy the excellent content?” invites agreement.
Averages hide detail. A mean score can conceal two groups with opposite views. Looking at the spread, and reading comments, adds context.
Designing better surveys
A few modest principles are widely recommended.
- Ask fewer, more specific questions. A long form lowers completion. Questions about particular lessons or activities tend to be more useful than general ones about satisfaction.
- Use neutral wording and offer a balanced scale.
- Include an open question. “What one change would help most?” often draws the most useful responses.
- Ask at several points, not only at the end.
- Ask about the course’s goals. If the stated learning objectives were clear, learners should be able to say whether they understood them.
- Say what will happen. People respond more when they believe the answers matter.
Consider also sending a short, polite question to people who dropped out, with an easy way to decline. Their reasons for leaving, including cost, time or the course itself, are often more informative than the views of finishers. Relating this to wider debates on completion rates is useful, but a low completion rate does not on its own show that a course is poor, since many open learners never meant to finish.
Closing the loop
Collecting feedback and doing nothing visible breeds cynicism. Teams that tell learners what they changed, even in a short note, tend to find that future learners trust the process more. This is the course-level version of the cycle discussed in research on feedback loops.
Not every comment should be acted on. Learners may disagree with each other, and some requests would harm the course’s purpose. A useful practice is to look for repeated themes, check them against other data, and make changes one at a time where possible so their effect can be seen.
Privacy and fairness
Surveys can capture personal information. Keep them anonymous where practice allows, collect only what is needed, and be careful that small groups or free-text comments do not identify a person. The principles behind personal data sharing apply to course teams as well as to learners. For legal requirements in your own situation, consult official guidance, since rules differ between countries and institutions.
Putting it in perspective
Student feedback is best treated as one input among several: surveys, comments, behavioral data and some evidence of what people actually learned. No single source is reliable on its own, and correlations between ratings and other measures should be read with care, remembering that a link between two things does not show that one causes the other. Used humbly, student voices can still catch problems that course teams would never notice themselves.