Evidence-Based Education Plan

Ava Kunz

A phased learning architecture from birth to fifteen, synthesized from peer-reviewed research across six academic domains
Explore the Plan
Methodology
Six Agents. One Synthesis.
Six independent Claude Sonnet research agents — each with web search access to live academic databases — reviewed the original education plan from distinct disciplinary lenses. No agent saw another's findings. A synthesis agent then reconciled agreements, flagged conflicts, and produced this revised plan.
◐
Screen Time & Early Media
AAP/WHO guidelines, video deficit effect, co-viewing meta-analyses, educational app evidence for ages 0–5.
◆
Mastery-Based Learning
Bloom's framework, Slavin–Kulik debate, IXL ESSA Tier 1 RCT, adaptive platform meta-analyses, mastery threshold research.
◈
Intrinsic Motivation
Deci & Ryan SDT (128 studies), gamification meta-analyses, reward undermining effects, drill vs. game-based learning.
◎
Physical Play & Manipulatives
Montessori RCTs, block play longitudinal studies, spatial-to-math pathways (N=16,338), tactile exploration evidence.
◇
Early Programming & CT
ScratchJr validation studies, CT→STEM transfer meta-analysis (g=0.85), blocks-before-text progression evidence.
◉
Parental Involvement
Co-use learning effects, JME language development, helicopter parenting risks, self-regulated learning scaffolding.
Evidence Base
Confidence Assessment
Each domain rated by research agents based on quality, consistency, and replication depth of available peer-reviewed evidence.
Phase 0 — Zero Screens
HIGH
20/22 orgs · d≈−0.50 video deficit
Phase 1 — Co-Participated
MED–HI
Co-use validated · no long-term app RCTs
Phase 2 — Core Mastery
MED–HI
d=0.94 elementary · Slavin debate open
Phase 3 — Programming
MEDIUM
CT→STEM g=0.85 · no longitudinal data
Motivation
MED–HI
SDT robust (128 studies)
Parental Role
MEDIUM
Co-use g=0.198 · helicopter risk
Key Effect Sizes from the Research Base
Architecture
Plan Trajectory in Three Dimensions
Each phase plotted across three axes that define the plan's structure: digital tool exposure, parental control, and child independence. Drag to rotate. The arc shows the designed handoff from parent-directed to child-owned learning.
Phase Trajectory — Parental Control × Digital Exposure × Independence
X-axis: digital tool hours/week · Y-axis: parental control level · Z-axis: child independence. Sphere size = tool count in phase. Drag to orbit.
Phase 0 (0–2yr) Phase 1 (2–4yr) Phase 2 (5–7yr) Phase 3 (8–11yr) Phase 4 (12–15yr) Trajectory path
0–24 mo
Build the Substrate
2–4 yr
First Tools
5–7 yr
Core Mastery
8–11 yr
Depth & Independence
12–15 yr
Acceleration
0

Build the Substrate

AGES 0–24 MONTHS
Spatial reasoning, language density, tactile exploration, caregiver attunement. This is infrastructure — measurable gains may not surface until middle school.
📵
Zero Passive Screens
Evidence-Backed

Exception: Video calls with family — these preserve turn-taking and social contingency. Background TV off during active infant time; research documents vocabulary reduction from background media even when the infant isn't watching.
🧱
Physical Tools
Evidence-Mixed

Lovevery kits for Montessori-inspired sensory exploration. Home kits are adapted — value is manipulative quality + structured adult engagement, not Montessori fidelity per se.

Board books from birth. Wooden blocks and stacking/nesting toys from ~9 months.
LoveveryBoard BooksWooden Blocks
🗣️
What the Parent Does
Evidence-Backed

Narrate everything. Follow infant attention rather than directing it. Name spatial relationships explicitly — "on top of," "inside," "around." Spatial language from parents predicts children's spatial thinking independently of general language input.

Monitor your own phone use during play time.
📊
Longitudinal Evidence
Evidence-Backed

Spatial construction at age 5 directly predicts math reasoning at age 17 (N=16,338, Millennium Cohort). Preschool block performance correlated with high school math in a 16-year follow-up. Cross-modal tactile skills at 12 months predict IQ and language a decade later.

⚠ Expectation Calibration

Measurable academic gains from this phase will not appear in kindergarten assessments. Multiple longitudinal studies document "sleeper effects" emerging in middle school or adolescence. Phase 0 is infrastructure, not short-term performance.

1

First Tools, Scaffolded Closely

AGES 2–4 YEARS
Phonological awareness, numeracy foundations, cause-and-effect reasoning, early frustration tolerance. Not "effort → satisfaction habit loops" — this cognitive architecture is not accessible before age 4–5.
📱
Screen Allocation
Evidence-Backed

20–30 min/day maximum, co-participated. Well under AAP's 1-hour ceiling. The AAP hard line is at 18 months, not 24 — the plan's threshold is more conservative than the evidence requires.
Khan Academy KidsStarfall
👥
Co-Participation Protocol
Evidence-Backed

Passive co-viewing produces minimal benefit. Instructive mediation is what research validates:

Before: Name what you'll do together
During: Ask predictive questions, connect screen to real objects, follow child's attention
After: Bridge one element to a physical counterpart within 30 min

This is neurological — infants were 19× more likely to transfer learning in high-quality interactive dyads.
🧩
Physical Complements
Evidence-Backed

Magnetic building sets, puzzles, playdough, outdoor play. Minimum 2:1 ratio physical-to-screen. Physical spatial training far more effective than digital sessions (Surrey meta-analysis, 3,700 participants).
🤖
Unplugged CT Pre-Exposure (Age 3+)
Judgment Call

Sequencing games (picture cards for daily routines), Simon Says as algorithm practice, Robot Turtles or similar board games. Low-cost, low-risk, directional evidence supports unplugged vocabulary bridges before digital coding.

App Evaluation Criteria

  1. Requires active response, not passive watching?
  2. Provides immediate, specific, informational feedback?
  3. Progresses in difficulty based on performance?
  4. Ad-free and free of streak mechanics?

If an app fails more than two criteria, do not use it.

2

Core Literacy, Numeracy & First Coding

AGES 5–7 YEARS
Reading fluency, mathematical reasoning, spatial-geometric foundations, early computational thinking. The developmental window with the strongest mastery learning effects (d=0.94 at elementary level).
📐
Academic Stack
Evidence-Mixed

Beast Academy as primary math, 2–3×/week — conceptual-first design addresses rote-drill risk.
IXL as supplement: 10–15 min, 3–4×/week. Not the daily backbone.
Khan Academy for concept explanations, mastery-gated progression.
Beast AcademyIXL (supplement)Khan Academy
💻
Coding Starts Here
Evidence-Backed

ScratchJr targets ages 5–7. Starting at 8 enters at the validated ceiling.

Unplugged activities first (1–2 wks) → ScratchJr 15–20 min, 2×/week, parent present for first 2–3 months.

If-then conditionals are genuinely difficult at this age. ScratchJr omits them by design. Celebrate sequence and repetition mastery.
🎯
Contingent Scaffolding
Evidence-Backed

Ask "What have you tried?" before offering suggestions. Offer the minimum useful hint.

Dashboard protocol: Stagnation across 3+ sessions on same concept → side conversation using physical materials. Single-problem struggle is not the signal to intervene.
🏃
Physical Track
Evidence-Backed

Continue building sets, tangrams, pattern blocks. Don't phase these out — spatial benefits persist and compound. Total screen time under 1 hr/day, active content prioritized.
3

Depth, Independence & Self-Regulation

AGES 8–11 YEARS
Deep mathematical reasoning, independent habits, Scratch-level coding, and explicit self-regulated learning skills. The most underspecified phase in the original plan.
🧮
Math Stack
Evidence-Mixed

IXL continues (10–15 min, 3–4×/wk). Beast Academy → AoPS Pre-Algebra at ~10–11. Brilliant.org for exploratory STEM at 10+, introduced alongside parent initially.
IXLBeast → AoPSBrilliant.org
🐍
Coding Progression
Evidence-Backed

ScratchJr → Scratch at 8–9. Scratch through age 11.

Code quality: Named variables, no duplicated code, custom blocks. Monthly code review on readability — bad Scratch habits impede Python transition.

Scratch → Python at 11–12, deliberate scaffolding. Map Scratch concepts to Python equivalents explicitly.
🧠
Self-Regulated Learning Track
Evidence-Backed

Phase 4 depends on this.

Planning: "Write one sentence about what you want to accomplish today."
Self-assessment: "Scale of 1–3, how well do you understand this?"
Error analysis: Categorize errors — conceptual vs. procedural vs. careless.
Weekly review: 10 min — what was learned, what was hardest, what to explore next.
👥
Social & Parental Role
Judgment Call

Parent becomes "learning coach" — active but through questions, not content delivery.

Social component: Math circles, coding clubs, peer groups. Peer instruction shows large effect sizes. The plan cannot remain parent-child dyadic.

⚠ Gender-Stereotype Audit

STEM-background parent designing STEM-heavy curriculum for a daughter: research documents risks of differential praise. PNAS 2025 (N=2,765) found gender stereotypes about CS/engineering diverge from math/science, affecting motivation as early as first grade. Include reflection checkpoints: Equal encouragement for non-STEM interests? Unconsciously weighting STEM praise?

4

Pre-College Acceleration

AGES 12–15 YEARS
Deep independent learning, university-level coursework, real-world project execution, and the full autonomy transfer the prior phases built toward.
🎓
Academic Platforms
Judgment Call

AoPS for competition math. Coursera/edX for university courses. Khan Academy SAT/AP prep. Vaia for spaced repetition. These require the SRL skills from Phase 3.
AoPSCoursera/edXKA SAT/APVaia
🔬
Real-World Projects
Judgment Call

Lab projects leveraging the home engineering lab. Python for data analysis, hardware prototyping, scientific investigation. Self-directed with parent as consultant. Portfolio development for college applications.
🔄
Motivation Bridge
Evidence-Mixed

Transition from IXL's external scaffold to AoPS's self-directed structure needs a motivational bridge. Research shows externally-motivated behaviors diminish when incentives are withdrawn. If intrinsic motivation isn't self-sustaining by now, the gap is in Phase 3 execution.
👨‍👧
Resource Provider
Evidence-Backed

Full authoritative-parenting model: high warmth, high expectations, progressive autonomy. Parent provides resources and asks about progress.

Budget: ~$720–920/yr at peak. Khan Academy free; IXL ~$80–120; AoPS ~$200–400; Brilliant ~$100; Coursera audit free.
Cross-Cutting
Design Principles
Seven evidence-informed principles governing every phase, updated from the original six based on synthesis findings.

✓ Validated

Mastery Over Completion — d=0.52–0.94 across hundreds of studies. But mastery thresholds vary by platform and are not audited.
Physical-Digital Pairing — Combining physical manipulatives with digital tools outperforms either alone (Terry, 1995).
Screen Time as Tool — Active problem-solving only. Minimum 2:1 physical-to-screen ratio in early phases.
Co-Participation Quality — Added in revision. Before/during/after instructive mediation protocol.

↻ Revised

Nuanced Gamification — Avoid streaks and cosmetic rewards. Allow mastery badges communicating demonstrated competence.
Structured Intervention — Stagnation across 3+ sessions = intervention signal. Single-problem struggle is where learning happens.
Self-Regulated Learning — Added in revision. Explicit metacognitive instruction from age 8. Phase 4 depends on it.
Epistemic Honesty
What We Don't Know
Critical gaps and overstated claims identified by the agent synthesis.
🔴
No Long-Term Transfer Data
Transfer evidence for this platform stack tracking to standardized achievement is essentially nonexistent. The Slavin-Kulik debate on criterion-to-norm transfer remains unresolved after 35 years.
🔴
Bloom's 2-Sigma Overstated
A 2020 meta-analysis of 96 tutoring studies found the average effect was 0.37 SD (14 percentile points). None produced a two-sigma effect.
🔴
IXL Claims Overstate Evidence
ESSA Tier 1 evidence rests on one RCT in a single Michigan district. Effect size 0.13 SD — significant but small to moderate.
🟡
Confound: Parental Selection
Montessori benefits concentrate in high-SES families. Most screen-time studies are correlational. The parent's own motivation is a critical, unmeasured variable.