Why One-on-One STEM Tutoring Often Creates Dependent Kids Instead of Independent Thinkers

A parent told me her daughter had been working with a private math tutor for 18 months. Her test scores had improved. Her homework completion rate was excellent. But then, on a practice test with no tutor present, she scored 40 points lower. The mom asked the tutor about it. The tutor explained that her daughter was “still building confidence.” Eighteen months in, she was still building confidence.

What the tutor didn’t name — and what the research names clearly — is that the daughter was experiencing tutor-dependent performance. She could solve problems when a tutor was present and could prompt the next step. She couldn’t do it alone. The tutoring had, inadvertently, built a crutch rather than a capability.

This is not a rare outcome. It has a mechanism, it has a name in the research literature, and it’s something parents can change by understanding what good tutoring actually looks like.

What “Productive Struggle” Is and Why It’s the Heart of STEM Learning

Manu Kapur, a learning scientist at ETH Zurich, spent a decade studying what he calls “productive failure” — the learning that happens when students try hard problems before being shown how to solve them, fail in instructive ways, and then receive instruction. His 2016 paper in Educational Psychologist summarizes the core finding: students who struggled first before receiving direct instruction significantly outperformed students who received instruction first on transfer tasks — problems they hadn’t seen before.

The mechanism matters. When a student attempts a problem and gets stuck, their brain activates prior knowledge, constructs partial models of the problem, makes connections between concepts, and generates hypotheses about what might work. That activation creates the cognitive scaffolding that makes subsequent instruction stick. When a student is shown how to solve a problem before they’ve attempted it, the instruction arrives before the mental framework exists to receive it. The procedure gets memorized, but the underlying concept doesn’t embed.

Kapur’s 2012 paper with Kapur & Bielaczyc in the Journal of the Learning Sciences found this effect across multiple content domains — math, physics, chemistry — and across multiple age groups. The productive failure effect isn’t a quirk of one study. It replicates.

This is not an argument against instruction. It’s an argument for sequencing: struggle first, instruction second. Most private tutoring does the opposite.

How One-on-One Tutoring Can Accidentally Bypass the Struggle

The problem isn’t that tutors are unskilled. Many are excellent at explaining concepts clearly. The problem is structural: when a child gets stuck in a private session, the social dynamics of the situation create enormous pressure to help immediately.

The child looks frustrated. The parent is paying by the hour and wants to see progress. The tutor’s instinct is to explain, to scaffold, to help. All of these pressures point in the same direction: solve the child’s confusion as quickly as possible.

Kurt VanLehn at Arizona State University has studied tutoring loops for 20 years. His 2011 review in Educational Psychologist, examining human tutoring, computer tutoring, and self-study across over 200 studies, found that the most effective tutoring interactions were those where the tutor waited — held back assistance long enough for the student to generate an attempt, however partial, before intervening. The key variable wasn’t tutor expertise. It was tutor restraint.

VanLehn called this the “step” problem: at each step in a problem-solving sequence, there’s a choice about whether to give the student the next step or wait for them to generate it. Tutors who give the step too quickly produce students who can follow a worked example. Tutors who wait, prompt with questions, and let the student generate the step produce students who can solve new problems independently.

The irony is that the “good” tutor session — the one where the child makes visible progress through the hour, the one that feels productive — often produces worse long-term learning than the session that feels slower and more frustrating because the tutor kept asking questions instead of explaining.

The Homework Help Trap: When Getting the Right Answer Hurts

Barry Zimmerman’s decades of research on self-regulated learning, summarized in his 2002 paper in Theory Into Practice, identifies a specific failure mode in academic support: when students receive immediate help on errors, they never develop error-monitoring skills. Error monitoring — noticing that something is wrong, diagnosing the source of the error, and correcting it — is a separable cognitive skill from getting the right answer. Tutoring that prioritizes correct answers suppresses error-monitoring development.

For STEM subjects in particular, this has downstream consequences. Engineering is fundamentally an error-detection discipline. At Apple, Samsung, and Texas Instruments, the engineers who stood out weren’t the ones who never made mistakes. They were the ones who noticed mistakes quickly, diagnosed the cause accurately, and didn’t need to be told what to try next. Those skills are built by wrestling with your own errors — not by having them corrected for you.

The homework-help version of tutoring — child brings the assignment, tutor helps them complete it, assignment gets submitted correctly — produces correct homework and suppressed error-monitoring. The child learns the content of that specific assignment, but not the metacognitive habit of catching their own mistakes. Over time, this creates a student who can perform when guided but not when solo.

Roelieke Pea’s 2004 analysis in Journal of the Learning Sciences of scaffolding theory in educational settings draws a sharp distinction between “fading” scaffolds (where the support is deliberately reduced as competence grows) and “permanent” scaffolds (where the support level stays constant). Good tutoring is a fading scaffold — it becomes less necessary over time. Tutoring that stays at a constant level of support, session after session, is functioning as a permanent scaffold. That means the student’s independence is not growing.

What Good Tutoring Looks Like vs. What Creates Dependence

The table below distinguishes tutoring behaviors that build independent thinking from those that create dependence.

Tutor Behavior	Effect on Independence	Research Support
Shows how to solve the problem before student attempts	Creates imitation without understanding	Kapur (2016); VanLehn (2011)
Waits 60+ seconds before responding to student confusion	Builds problem-generation skills	VanLehn (2011) wait-time research
Corrects errors immediately	Suppresses error-monitoring development	Zimmerman (2002) self-regulated learning
Prompts with questions: “What do you know about this type of problem?”	Activates prior knowledge; builds retrieval	Kapur & Bielaczyc (2012)
Completes all homework during the session	Prevents at-home independent practice	Pea (2004) scaffolding fading
Reduces scaffolding each session as competence grows	Builds independence over time	Pea (2004); Zimmerman (2002)
Asks student to explain back what they just understood	Activates elaborative encoding; deeper retention	Chi & Wylie (2014) self-explanation effect
Uses the same worked example every time	Limits generalization to new problem types	Kapur (2016) transfer research

The pattern is consistent: the tutor behaviors that feel most helpful in the moment — quick explanation, immediate correction, solving the problem together — are the ones that most interfere with independent thinking. The behaviors that feel slower and more frustrating build the skills that transfer to unseen problems.

How to Brief a Tutor to Teach Thinking, Not Answers

Parents can directly influence how tutoring sessions run by having a clear, specific conversation with the tutor before the engagement begins. Here’s what to ask for:

Ask for explicit “struggle time” before explanation

Tell the tutor: “When my child is stuck, please wait at least 60 seconds before jumping in. Let them try something, even if it’s wrong.” Most tutors will honor this if asked explicitly. Without being asked, the social dynamics of the session will push them toward faster intervention.

Ask the tutor to work backward, not forward

When explaining a concept, ask the tutor to ask the student “what do you think the answer is?” first, and build backward from the student’s attempt. This requires the student to generate a hypothesis before receiving instruction — the sequencing that Kapur’s research supports.

Ask for explicit fading over 8 weeks

Request that the tutor provide less scaffolding each week on problem types the student has seen before. A student who needed step-by-step guidance on quadratic equations in week 2 should be solving them without guidance by week 6. If the scaffolding level hasn’t changed after 8 weeks, the sessions are maintaining performance rather than building capability.

Separate homework completion from session goals

Make it clear to the tutor that getting the homework done is not the session goal. The session goal is building a skill the child can demonstrate without the tutor present. Homework can be done independently, after the session, as a test of the skill built.

For an analysis of the cost structure of different tutoring formats and what the research says about outcomes, see our article on the real price of private engineering tutoring in 2026.

The Alternatives to Tutoring That Build Independent Problem-Solvers

Several formats have stronger track records than private tutoring for building autonomous problem-solvers:

Problem-sets with delayed feedback: Give the child a problem set, let them work through it alone (with struggle), then review the errors together. The struggle period activates the cognitive processes that make correction meaningful. This is the pattern in Kapur’s productive failure research.

Collaborative peer work: Explaining a concept to another person who doesn’t understand it is among the most powerful learning conditions documented in cognitive science. The person explaining has to retrieve, organize, and communicate knowledge — all of which strengthen encoding. Private tutoring eliminates this by removing peers from the equation.

Project-based contexts where errors have natural consequences: When a circuit doesn’t work, the circuit doesn’t work — no tutor intervention needed. The feedback is embedded in the task. This is why hands-on building projects produce different cognitive outcomes than worksheet-based tutoring.

For a comparison of small-group versus private STEM learning formats and the peer learning research, see our article on small group STEM classes vs. one-on-one tutoring.

What to Watch for Over the Next 3 Months

If your child has been in private tutoring, here is how to assess whether it’s building independence or dependence:

Week 4: Give your child a problem at the same difficulty level as what they’ve been working on with the tutor — but don’t let the tutor know in advance. Can they start the problem independently? Can they identify what type of problem it is? If not, the sessions are not yet producing transferable skill.

Month 2 red flags: If your child’s first response to any hard problem is “I need to ask my tutor” — without attempting anything first — the tutor has become a cognitive crutch. This is the behavioral marker of learned helplessness in the academic context.

Month 3 self-check: Can your child describe, in their own words, what they do when they get stuck? Do they have a strategy? (“I check my setup, then I try a simpler version of the problem.”) If your child has no answer to “what do you do when you’re stuck?”, the sessions aren’t building metacognitive skill — only answer-getting.

Frequently Asked Questions

My child’s grades went up with tutoring. Isn’t that the goal?

Grade improvement is a signal worth tracking, but it’s not the same as learning. If your child’s performance drops significantly without the tutor present (on standardized tests, in class without the homework having been tutor-helped), the grade improvement is measuring tutor-supported performance, not independent skill. Both are real, but only one transfers to contexts without a tutor.

How long should I wait before deciding tutoring is creating dependence?

The clearest signal is whether the scaffolding level is decreasing over time. After 8–12 sessions, your child should need noticeably less prompting on problem types they’ve already covered. If the sessions look the same at week 12 as they did at week 2 — same level of guidance, same amount of struggle-avoidance — the sessions are maintaining, not building.

What if my child says they learn better when the tutor explains first?

Children generally prefer lower-effort learning conditions — that’s normal. But preference and effective learning often diverge. Kids prefer watching a worked example to struggling with an unseen problem, for the same reason most people prefer watching someone run to running themselves. The preference is real; it’s not evidence of better learning.

Is this a problem unique to private tutoring, or can it happen in school too?

It can happen in any instructional context where the adult prioritizes correct answers over the student’s problem-solving process. The research on classroom instruction shows similar patterns: teachers who move quickly to provide the answer during whole-class discussion produce students who wait to be told, rather than attempting. The dynamic is the same — the setting just changes.

How do I tell a tutor that I think they’re teaching the wrong way without insulting them?

Frame it as a learning goal, not a criticism. “I want [child’s name] to be able to work through problems independently by [date]. Can you help me design sessions that build toward that?” Most tutors will appreciate having a specific, observable goal rather than a vague “improve in math” objective.

At what age does this dependency risk become most significant?

Research on self-regulated learning (Zimmerman, 2002) suggests the habit of seeking external guidance rather than independent monitoring forms gradually across ages 8–13. These are the years when metacognitive habits — how a student manages confusion, monitors their own understanding, responds to errors — are most formative. Dependency formed in these years is harder to reverse than dependency formed earlier.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

Sources

1. Kapur, M. (2016). “Examining Productive Failure, Productive Success, Unproductive Failure, and Unproductive Success in Learning.” Educational Psychologist, 51(2), pp. 289–299. https://doi.org/10.1080/00461520.2016.1155457
2. Kapur, M., & Bielaczyc, K. (2012). “Designing for Productive Failure.” Journal of the Learning Sciences, 21(1), pp. 45–83. https://doi.org/10.1080/10508406.2011.591717
3. VanLehn, K. (2011). “The Relative Effectiveness of Human Tutoring, Intelligent Tutoring Systems, and Other Tutoring Systems.” Educational Psychologist, 46(4), pp. 197–221. https://doi.org/10.1080/00461520.2011.611369
4. Zimmerman, B.J. (2002). “Becoming a Self-Regulated Learner: An Overview.” Theory Into Practice, 41(2), pp. 64–70. https://doi.org/10.1207/s15430421tip4102_2
5. Pea, R.D. (2004). “The Social and Technological Dimensions of Scaffolding and Related Theoretical Concepts for Learning, Education, and Human Activity.” Journal of the Learning Sciences, 13(3), pp. 423–451. https://doi.org/10.1207/s15327809jls1303_6
6. Chi, M.T.H., & Wylie, R. (2014). “The ICAP Framework: Linking Cognitive Engagement to Active Learning Outcomes.” Educational Psychologist, 49(4), pp. 219–243. https://doi.org/10.1080/00461520.2014.965823
7. Kraft, M.A., & Falken, G.T. (2021). “A Blueprint for Scaling Tutoring Across Public Schools.” AERA Open, 7(1). https://doi.org/10.1177/23328584211027923