Boosting the quantity and quality of parent-child language use with Chat2Learn-AI

Last registered on July 07, 2025

View Trial History

Pre-Trial

Trial Information

General Information

Title

Boosting the quantity and quality of parent-child language use with Chat2Learn-AI

RCT ID

AEARCTR-0015872

Initial registration date

July 07, 2025

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

July 07, 2025, 3:24 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

United States of America

Region

Primary Investigator

Name

Daniela Bresciani

Affiliation

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Ariel Kalil

PI Affiliation

University of Chicago

Contact Investigator

PI Name

Michelle Michelini

PI Affiliation

University of Chicago

Contact Investigator

PI Name

Linxi Lu

PI Affiliation

University of Chicago

Contact Investigator

Additional Trial Information

Status

In development

Start date

2025-09-15

End date

2026-02-27

Keywords

Behavior, Education

Additional Keywords

Parenting, parent-child talk, early childhood, language use, large language models, artificial intelligence

JEL code(s)

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

This study aims to evaluate the effectiveness of Chat2Learn-AI for boosting the quantity and quality of parent-child language use. Chat2Learn-AI is a messaging program enhanced by large language models (LLMs) that delivers both static and dynamic, on-demand conversation prompts with illustrations to parents, tailored to children’s interests as indicated by parent-recorded child responses to program prompts. The project will consist of a sample of 200 parents of kindergarten children, who will be randomized into the treatment and control groups. Parents in the treatment group will get access to Chat2Learn-AI, while parents in the control group will remain business-as-usual. The study will evaluate the impact of Chat2Learn-AI on 1) the quantity and quality of syntax, sentiment, linguistic diversity, and conversational turns in parent-child conversations, and 2) parental agency and growth mindset. To collect the first set of outcomes, parents in the sample will be asked to complete an audio task by recording themselves having a conversation with their child. To collect secondary outcomes, parents will be asked to complete an endline survey.

External Link(s)

Registration Citation

Citation

Bresciani, Daniela et al. 2025. "Boosting the quantity and quality of parent-child language use with Chat2Learn-AI." AEA RCT Registry. July 07. https://doi.org/10.1257/rct.15872-1.0

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

Parents in the treatment group will get access to Chat2Learn-AI. Chat2Learn-AI is a messaging program enhanced by large language models (LLMs) that delivers both static and dynamic, on-demand conversation prompts with illustrations to parents, tailored to children’s interests as indicated by parent-recorded child responses to program prompts. Below is an example of an interaction with the program:

Chatty-AI : Ask Jamie – If you could choose to have any pet, which pet would you choose?
Mom: Jamie would choose a horse.
Chatty-AI: Awesome! Ask Jamie – If you could go on a trail ride with your horse, where would you go? A forest, beach, mountain, or somewhere else?

Chat2Learn-AI delivers a foundation of three to five identical conversation prompts to all treatment families paired with illustrations each week. In addition, it layers on AI chatbot functionality, allowing the parent and child to continuously engage with the program on a countless range of topics that promote ongoing conversational engagement. Parents and children can spontaneously direct the program to generate prompts tailored to their specific interests to support longer, richer and more engaging conversations.

Intervention (Hidden)

Intervention Start Date

2025-10-06

Intervention End Date

2025-12-26

Primary Outcomes

Primary Outcomes (end points)

Our main outcome is the quantity and quality of parent-child conversations. Parents will be asked to complete an audio-recording task with their children, where they will be asked to record themselves while having a conversation. Using this recording, we will code the quality and quantity of syntax, sentiment, linguistic diversity, and conversational turns.

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

Our secondary outcomes are parent's growth mindset and sense of agency. Parents will be asked to complete an endline survey with growth mindset and agency questions.

Secondary Outcomes (explanation)

Experimental Design

We will partner with 1-2 elementary school networks to recruit 200 low-income parents of kindergarten children (both Spanish and English speaking). We will enter into a data use agreement (DUA) with our partners to enable an opt-out model of consent. With a DUA in place, we will obtain primary parent contact information and basic family demographic information for all participating kindergarten classrooms. Parents will be informed about the study partnership in advance and can opt-out at any time before or after randomization activities.

Parents will be randomized into either the treatment or the control group. Parents in the treatment group will get access to Chat2Learn-AI, while parents in the control group will continue business-as-usual. All participants will be asked to complete a baseline audio task, an endline audio task, and an endline survey.

Experimental Design Details

Randomization Method

Randomization done in office by a computer

Randomization Unit

Parents (individual-level randomization)

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

200 parents

Sample size: planned number of observations

200 parents

Sample size (or number of clusters) by treatment arms

100 parents control, 100 parents treatment

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

To calculate the minimum detectable effect size (MDES) given our sample size of 200, we use PowerUP! . Given a 0.05 significance level, 0.8 power, and an expected R-squared of 0.45 (based on previous projects), power estimates suggest an MDES of 0.295 standard deviations.

Supporting Documents and Materials

IRB