Parent nudges that work: text messages, not dashboards

The best evidence on getting parents involved in a child's learning is about short, specific text messages, not dashboards. In randomized trials, texts naming one missing assignment or one thing to try at home moved math grades by 0.09 to 0.21 standard deviations, and sending them by default, with an opt-out, mattered more than anything in the message. Peabody (peabody.kids) is free through the end of 2026. This post sets out those studies, what they do not show, and what Peabody gives a grown-up today.

The studies, one at a time

Missing work, Los Angeles. Peter Bergman ran a randomized trial with 462 students in which parents got text messages when their child missed an assignment. After four months, missed classes fell by 28% and math grades rose by 0.21 standard deviations; there was no effect on English. The working paper is on NBER. A larger multi-school version, published in the Journal of Human Resources in 2025, found a smaller +0.09 SD on math grade point average and a 2.7 point rise in pass rates, with effects two to three times larger for students who were most at risk.

Grades and absences, West Virginia. Bergman and Eric Chan sent parents in 22 schools automated weekly texts about grades, absences and missing assignments. Course failures fell by 39% and attendance rose by 17%. Test scores did not move. The Inter-American Development Bank paper is worth reading for that last line: a nudge changes what parents do about the next assignment, and the next assignment is not the same as a test.

The default, Washington DC. Bergman and Todd Rogers offered roughly 7,000 parents the same texting service in three ways. Signing up on a website: 0.8% did. Replying to a simple text: about 8 to 11%. Being enrolled automatically with a way to opt out: 96.5% stayed in. Only the automatically enrolled group showed better grades and fewer failures, as Education Week reported. The content of the message was the same in every arm. The default was the intervention.

One activity a week, preschool. Ready4K sent preschool parents three texts a week for eight months: a fact about a literacy skill, a small activity to do with it, and a follow-up. Parents did more of the activities (+0.15 to +0.29 SD on involvement measures) and children's early literacy rose by 0.11 SD, as Brian York, Susanna Loeb and Christopher Doss reported in the Journal of Human Resources in 2019. One caution: the lead author went on to found the company that sells the program. A separate trial of an opt-out text program for mothers by Lisa Gennetian and colleagues kept 88.7% of them enrolled for 26 weeks.

Why a dashboard is not the same thing

Learning apps mostly offer parents a dashboard: minutes, points, skills practised, a chart. There is very little evidence that dashboards change what a parent does. A 2022 review by Susnjak and colleagues found most learning dashboards stop at describing what happened and do not suggest what to do. The one direct study of a parent-facing visualization the research for this post found, published in 2026, leaned on an older rule from Ioana Jivet's work: information has to be personally meaningful and actionable before anyone acts on it.

The products that come closest to that rule show a trend against an expectation and one recommendation with a dose. DreamBox's family dashboard draws a trend line against an aim line and suggests five lessons a week. Amira draws a progress line with a predicted growth line. Accountability numbers like minutes or experience points tell a parent the app was open. They do not say what to do at dinner.

What the trials have in common

There is a matching finding from family literacy. Sénéchal and Young's 2008 meta-analysis of 16 home reading interventions (1,340 families, kindergarten to grade 3) found that parents reading to children had a small effect (d = 0.18, not significant), parents listening to children read had a larger one (d = 0.52), and parents taught to practise one specific skill with their child had the largest (d = 1.15). The specific action beats the general encouragement. "Read with your child" is a slogan; "listen to page three and, when they stop at a word, cover the ending and ask for the first sound" is a nudge.

What this research does not show

• Grades are not learning. The biggest effects are on missed assignments, course failures and grades. Where test scores were measured, they often did not move.
• These are school messages. Most of the trials text parents about schoolwork set by a teacher. A learning app at home is a different setting, and no trial has tested an app's weekly note the same way.
• Effects shrink as programs grow. The Los Angeles result was 0.21 SD; the multi-school version was 0.09. That is normal for education research and still useful.
• Vendors and authors overlap. The Ready4K study's lead author founded the company. That does not make it wrong; it means it should be replicated.

What Peabody gives a grown-up today

Peabody's grown-up view leans toward plain words and one thing to try, not a score. A free family account shows which story words are sticking (right on first read on three different days) and which are still being practised, a times-tables grid where each square is one fact, the week's number answers and finished parts, and a tip from Peabody when a part looks tricky (12 or more answers and under 60% right over two weeks). A "Tonight" card suggests the evening's story and words. Nothing on it compares a child to anyone else, and nothing is a grade.

Peabody also has two short emails, and here it falls short of the evidence in two ways:

First, both are opt-in, which is the arm that reached 0.8% of parents in the Washington DC trial. Second, Peabody's production email has not yet been confirmed as delivering, so no grown-up should count on receiving either until that is fixed. The research says the fix is a message on by default with an easy off switch, carrying one fact and one two-minute action. That is the plan, and it is not built.

Until then, you can make your own nudge. Once a week, open the grown-up view, pick the one word or fact that is still being learnt, and spend two minutes on it at dinner. Start free or open the library to see where your child is.

Questions parents ask

Do text messages from school really help?

In randomized trials, yes, for the things they target. Texts about missing work and grades cut course failures and raised math grades by about 0.1 to 0.2 standard deviations. They moved test scores much less, or not at all.

Why does it matter whether I opt in or am signed up automatically?

Because almost nobody opts in. In the Washington DC trial, 0.8% of parents signed up by web form, while 96.5% stayed when they were enrolled by default. Only the default group's children improved.

What is the best thing to do with a learning app's report?

Pick one item and act on it for two minutes. Studies of home reading found that teaching one specific skill had a much larger effect than general reading together, so one word, one fact, one question at dinner is the right size.

Does Peabody send me a weekly email?

There is a Sunday note and a story nudge, both off until a grown-up turns them on, and production email has not yet been confirmed as delivering. For now, check the grown-up view yourself: it shows which words are sticking, a times-tables grid and one tip when a part looks tricky.