Self-Determination Theory: Motivating Learners in Online Courses

Most advice on motivating online learners is a list of tactics: add a progress bar, send reminder emails, award a badge. Some of those help and some quietly make things worse, and a list can’t tell you which. Self-determination theory can. Its central question is the one every course builder faces: why do some people keep going when nobody is making them, and others stop at module two?

This entry covers what the theory says, what four decades of studies support (including the uncomfortable results from online courses), and how to design a self-paced course around it.

What the research says

Three needs, and two kinds of motivation

Self-determination theory was developed by the psychologists Edward Deci and Richard Ryan. Its starting point is that the type of motivation matters more than the amount. Doing something because you find it interesting, or because you value what it leads to, is different in kind from doing it to earn a reward, avoid a penalty or escape guilt, even if the effort looks the same from outside.

The theory proposes that people move towards the first, self-endorsed kind when their social setting meets three psychological needs.

• Competence: feeling capable of what’s being asked, and seeing yourself get better.
• Autonomy: feeling that what you’re doing is your own choice and makes sense to you, not that you’re being pushed.
• Relatedness: feeling connected to, and respected by, the people around you.

Ryan and Deci also describe a continuum of extrinsic motivation, from external regulation (rewards and punishments), through introjected regulation (guilt, pride, wanting to look good), to identified regulation (doing something because you value it), with intrinsic motivation, doing it for its own sake, sitting alongside. The practical point is that extrinsic motivation can become more self-endorsed: a learner who starts a compliance course because they must can come to see why it matters (Ryan & Deci, 2000).

Which motivation predicts what

The largest test of whether these distinctions matter in education is a meta-analysis by Joshua Howard and colleagues. They pooled 344 samples, covering 223,209 students, and related each type of motivation to 26 outcomes (Howard et al., 2021).

Intrinsic motivation was related to both success and wellbeing. Personal value (identified regulation) was particularly highly related to persistence, which is the outcome most course builders worry about. Motivation from guilt or ego predicted persistence too, but came with more ill-being. Motivation from rewards or avoiding punishment was not associated with performance or persistence at all, and was associated with lower wellbeing. These are correlations, mostly from questionnaires, so they don’t prove the motivation caused the outcome. But the pattern is consistent across a very large body of work.

What builds self-determined motivation

A companion meta-analysis by Julien Bureau and colleagues asked the next question: what predicts self-determined motivation in the first place? Across 144 studies and more than 79,000 students, teachers’ autonomy support predicted students’ need satisfaction and motivation more strongly than parents’ did. Of the three needs, competence was the strongest predictor of self-determined motivation, followed by autonomy and then relatedness (Bureau et al., 2022). For course design, that ordering is useful: if you can only do one thing well, help learners feel capable.

Experiments show the same levers can be pulled on purpose.

• Rationales. In Jang’s experiment, 136 undergraduates studied a deliberately dull, 20-minute text lesson on statistical correlation. One group was first told why it would be useful to them as future teachers, in non-pressuring language that acknowledged the lesson might not be fun (Jang, 2008). The rationale group engaged more and understood more; factual recall didn’t differ between the groups.
• Choice. Patall, Cooper and Robinson pooled 41 studies and found that giving choices increased intrinsic motivation, effort, task performance and perceived competence. The effect was strongest when people made two to four successive choices, and weaker when choice came with rewards (Patall et al., 2008). One surprise: small choices that didn’t affect the instruction itself had larger effects on intrinsic motivation than choices about the learning content.
• Training teachers. A meta-analysis of 19 studies of programmes that taught teachers and other professionals to be more autonomy-supportive found a weighted effect size of 0.63. The more effective programmes were short (one to three hours), skill-based and taught several techniques together, though the authors caution the small number of studies (Su & Reeve, 2011).
• Whole interventions. A 2024 meta-analysis of 31 school and university intervention studies found the largest effects on perceived autonomy support, a moderate effect on competence, a partial effect on intrinsic motivation and no significant overall effect on relatedness, with high variation between studies (Wang et al., 2024).

Jang’s result is the one to remember for course design, because a rationale costs two sentences.

Rewards: the finding that made the theory famous

The theory began with a counter-intuitive result: paying people to do something they already enjoy can reduce how much they do it afterwards, unpaid. Deci, Koestner and Ryan’s meta-analysis of 128 studies found that expected tangible rewards undermined intrinsic motivation on interesting tasks, measured by whether people kept doing the task in free time (d = -0.34 across tangible rewards). The effect held whether rewards were given for taking part, finishing or performing well, and tended to be worse for children. Positive feedback did the opposite: it increased both free-choice persistence (d = 0.33) and self-reported interest (d = 0.31), though less so for children (Deci et al., 1999).

Two cautions. These were mostly short laboratory experiments with interesting tasks, not multi-week courses, and none involved digital badges. The finding is about rewards that feel like the reason for doing something. Information that tells learners they’re getting better is a different thing, and it supports competence.

The online evidence is thinner and mixed

This is where most articles on the theory stop, and where we’d slow down. Chen and Jang surveyed 262 students in two fully online teacher certificate programmes at a US university (Chen & Jang, 2010). A supportive course climate strongly predicted need satisfaction, and need satisfaction predicted engagement: hours studied and activity in the course site.

Nothing in the model predicted final grades, which the authors partly put down to uniformly high grades (an average above 92). A later study at a US university compared 495 students in online courses with 519 in face-to-face courses. Online students reported a weaker learning climate, less need satisfaction, weaker motivation, less perceived knowledge transfer and lower grades. Need satisfaction went with higher perceived knowledge transfer; need dissatisfaction went with lower grades (Wang et al., 2019).

How strong is the evidence?

Moderate, for our purposes. The core claims, that the type of motivation matters and that competence, autonomy and relatedness support the better kinds, rest on large meta-analyses and many experiments. The limits: most outcome data are self-reported and correlational; most studies are in schools and universities with a teacher present; and the few studies of online courses link need support to engagement and perceived learning more clearly than to grades. We’d treat the theory as a reliable guide to what keeps people going, and not as a promise of better test scores on its own. Pair it with the principles that drive learning directly, such as effective feedback and formative assessment.

How to apply it in a course

In a classroom, a teacher supports these needs in the moment: noticing a student is lost, giving a reason, offering a choice. In a self-paced course, most of that has to be written in advance. Here’s how we’d do it, using one invented example: a six-module self-paced course teaching small-business owners to do their own bookkeeping.

1. Design for competence first

Bureau’s results put competence at the top, and it’s the need a self-paced course can support best without a teacher.

• Make the goal visible. Open each module with what learners will be able to do by the end, written as learning objectives they can check themselves against: “Reconcile one month of bank transactions against your records.”
• Make the first win early and real. Module one ends with learners categorising ten of their own transactions, not watching a 40-minute overview. A first task they can finish in one sitting shows them the course is doable.
• Pitch the difficulty. Work up from simple cases to messy ones, and keep each lesson’s load manageable; cognitive load is a competence problem as much as a memory one.
• Give feedback that shows progress. Practice questions with explanations, and a short self-check at the end of each module, tell learners they’re getting better. That’s the kind of positive, informational feedback the rewards research found increased interest.

2. Give a reason for every task, especially the dull ones

Jang’s experiment is the easiest finding in this entry to copy. Before the module on setting up a chart of accounts, which nobody finds thrilling, add two sentences: why it matters to them and an honest acknowledgement.

This part is fiddly, and it’s tempting to skip. But a chart of accounts set up for your business is what makes tax time a half-day job instead of a week of sorting receipts.

Note what it doesn’t say: “You must complete this module to receive your certificate.” That’s a controlling reason, and it’s the kind the research links to weaker motivation.

3. Offer a few real choices

Patall’s meta-analysis suggests a handful of choices beats none and beats a menu. In the bookkeeping course, learners could choose which of two case-study businesses to follow, whether to practise on their own records or a sample set, and the order of the two optional modules (payroll or GST). Keep the core sequence fixed where one skill depends on another. Choice shouldn’t mean learners skipping the foundations they need.

4. Watch your language

Self-paced courses are full of controlling phrasing that nobody chose deliberately: “you must”, “you should”, “required”, red warnings about deadlines. Rewrite instructions as information: “Next, try three practice entries” instead of “You must complete all practice entries before continuing.” If there’s a genuine requirement, say why it exists.

5. Build relatedness without forcing it

Relatedness had the weakest effects in the intervention research, and forced social features often fall flat. Cheap things that help: a short welcome video from the instructor, feedback and emails written in a human voice, and an optional place for learners to ask questions and see that others are stuck on the same thing. In cohort courses, small groups work better than one large forum.

6. Use rewards to mark progress, not to drive it

Certificates, badges and completion percentages aren’t banned by the research, but they shouldn’t be the reason the course gives for continuing. Award them for demonstrated skill rather than clicks, describe what each one shows the learner can now do, and leave out leaderboards, which turn competence into comparison.

Motivation design sits alongside, not instead of, the work on completion: our guide to online course completion rates covers the other reasons learners stop.

How the platforms support it

Most of what the theory asks for is in the writing: the rationales, the feedback, the tone. The platform decides how easily you can offer choices, show progress and give learners somewhere to connect. We haven’t yet published our reviews of Thinkific, LearnWorlds, Teachable and Kajabi, and features change often, so check these in your platform’s current documentation, on the plan you’re paying for:

• Flexible sequencing. Can you lock only the lessons that depend on earlier ones, and leave others open so learners can choose their order?
• Optional and alternative paths. Can you mark lessons as optional, or offer two versions of a practice activity and let learners pick?
• Practice with feedback. Can quizzes give an explanation for each answer and allow retakes, so learners can see themselves improve?
• Progress that means something. Does the progress display reflect skills demonstrated, or only lessons opened? Can certificates or badges be tied to passing an assessment?
• A place to connect. Is there a discussion area or community space attached to the course, and can you keep it optional?
• Message tone. Can you edit the automated emails and on-screen messages, so they read as information rather than pressure?

Further reading

• Ryan & Deci (2000). The short, widely cited overview of the theory by its founders. The place to start.
• Howard et al. (2021). The largest meta-analysis of how each type of motivation relates to student outcomes, including persistence.
• Bureau et al. (2022). What predicts self-determined motivation, and why competence matters most.
• Jang (2008). A clean, readable experiment on giving a rationale for a dull task. Open access through the theory’s website.
• Chen & Jang (2010). An honest test of the theory in fully online programmes, with results that didn’t all go the authors’ way.
• Deci, Koestner & Ryan (1999). The rewards meta-analysis, for anyone about to add points and badges to a course.

For the rest of the principles behind good course design, browse our research library.