← Back Vanta AI

The research behind the lessons

Every mission in Vanta AI is designed from peer-reviewed learning science. Here is a sample of the studies our curriculum decisions cite — the full annotated bibliography is in progress.

Sentance, Waite & Kallia (2019) — PRIMM

Students aged 11–14 learn programming better when they read and investigate code before writing it (n=673, significant post-test gains). Our coding missions start with reading and predicting, not a blank editor.

Lytle et al. (2019) — Use-Modify-Create

A grade-6 study (n=160) showing that starting from working programs and modifying them flattens the first-day difficulty spike. Our rover missions follow this arc.

Weintrop & Wilensky (2017, 2019) — blocks before text

Block-based coding produces better early learning — but the advantage disappears without an explicit bridge to text. We teach both, with a designed bridge in between.

Tucker et al. (2024); Brod et al. (2018) — predict first

Locking in a prediction before seeing a result measurably improves learning. That's why Vanta asks you to "call it" before every run.

Vartiainen et al. (2021); Sanusi et al. (2023) — learning ML by training

Hands-on example-feeding builds genuine understanding of training data's role, and is the most-endorsed K-12 machine learning pedagogy in systematic review. Our classifier missions are built on it.

Wineburg & McGrew (2019) — provenance first

Professional fact-checkers don't stare harder at content — they trace where it came from. Our media missions teach source-tracing over "spot the fake."

Iqbal et al. (2025) — explaining beats recall

A middle-school filter-bubble study showing gains in students' ability to explain personalization even when multiple-choice scores didn't move. Our feed missions assess by explanation, not recall.

Grootens-Wiegers et al. (2017) — the adolescent reward system

Adolescents have a highly responsive reward system and a still-maturing control system. It's one reason our milestones are personal and we build no leaderboards.

Druga & Ko (2021) — mechanism over magic

Hands-on classification work shifts kids toward mechanistic, accurate accounts of machine behavior. Our copy never says an AI "wants" anything.

Questions about our methodology? Write to info@vantaai.com.