Activating AI Tutoring for Foundational Learning at Scale in Jordan

Last registered on September 21, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Activating AI Tutoring for Foundational Learning at Scale in Jordan
RCT ID
AEARCTR-0019229
Initial registration date
September 13, 2026

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
September 21, 2026, 9:11 AM EDT

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

Locations

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

Request Information

Primary Investigator

Name
Fei Yuan
Affiliation
World Bank
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Shwetlena Sabarwal
PI Affiliation
World Bank
Contact Investigator
PI Name
Zaina Dawani
PI Affiliation
World Bank
Contact Investigator

Additional Trial Information

Status
In development
Start date
2026-10-05
End date
2027-07-01
Keywords
Behavior, Education
Additional Keywords
Generative AI, EdTech, Foundational learning
JEL code(s)
I21, I25, O33
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Jordan’s Ministry of Education and Ministry of Digital Economy and Entrepreneurship have deployed a national AI learning assistant reaching over 1 million students across 5,791 schools. Administrative data show that although platform access is broad, engagement is shallow. This evaluation is a three-arm, school-level cluster-randomized controlled trial testing whether structured human facilitation can activate more meaningful student and teacher engagement with the AI learning assistant among Grade 4 students in Jordanian public schools. 300 public schools will be randomly assigned to: (i) a control group continuing status-quo AI learning assistant access; (ii) Treatment group 1, in which facilitators deliver structured onboarding and monthly coaching to Grade 4 students to promote active-recall, question-driven engagement with the AI learning assistant or (iii) Treatment group 2, which include the onboarding of the AI learning assistant's teacher-facing application, classroom observation and coaching on how to use Siraj to engage students, and monthly online training sessions. The measured primary outcomes are foundational Arabic literacy (EGRA) and numeracy (EGMA); secondary outcomes include student AI-engagement quality, self-efficacy and growth mindset, teacher data literacy.
External Link(s)

Registration Citation

Citation
Dawani, Zaina, Shwetlena Sabarwal and Fei Yuan. 2026. "Activating AI Tutoring for Foundational Learning at Scale in Jordan." AEA RCT Registry. September 21. https://doi.org/10.1257/rct.19229-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
Control (100 schools): Grade 4 students continue status-quo access to AI learning assistant, via personal devices or school computers; teachers receive only basic platform orientation.
Treatment 1 (100 schools): Student engagement activation — facilitators deliver one structured onboarding module teaching Grade 4 students active recall, question formulation, response verification, and avoidance of passive answer retrieval when using AI learning assistant, followed by monthly school visits to provide follow-up coaching.
Treatment 2 (100 schools): Teacher engagement activation — Grade 4 teachers receive the onboarding of the AI learning assistant's teacher-facing application, classroom observations and coaching on how to use Siraj to engage students, and monthly online training sessions.
Intervention Start Date
2026-10-12
Intervention End Date
2027-02-28

Primary Outcomes

Primary Outcomes (end points)
Foundational Arabic literacy — reading fluency and comprehension (EGRA), Grade 3 students
Foundational numeracy — math proficiency (EGMA), Grade 3 students
Primary Outcomes (explanation)
EGRA and EGMA are validated early learning assessments including some composite scores. We will validate the instruments in the Jordanian contexts and follow their practices to construct composite test scores.

Secondary Outcomes

Secondary Outcomes (end points)
Student self-efficacy and growth mindset (Bandura’s Children’s Self-Efficacy Scale; Dweck Mindset Instrument, adapted for Jordan), Student AI literacy (instrument to be developed), Student and teacher AI engagement — frequency, depth, and quality of Siraj interactions (backend usage data), Student attendance and school engagement (school administrative records; classroom observation), Teacher knowledge of student skills/learning gaps (teacher survey), Equity: narrowing of literacy/numeracy learning gaps between refugee and non-refugee Grade 3 students
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Three-arm, school-level cluster-randomized controlled trial in Jordanian public schools with Grade 4 students. 300 eligible schools, drawn from the Ministry of Education’s full school list and stratified by refugee enrollment share and geographic cluster, are randomly allocated to 100 schools per arm: (1) Control - status quo with home access to an AI learning tool developed by the Ministry of Education; (2) Treatment 1 — facilitator-led structured onboarding and coaching to activate student engagement; (3) Treatment 2 — teacher coaching on how to use AI tool to engage students and monthly online training sessions.
Experimental Design Details
Not available
Randomization Method
Randomization is conducted in office by the research team using a computer-generated random sequence, applied to the full list of eligible Grade 4 public schools in three governorates -Amman, Zarqa, Irbid-, supplied by Jordan’s Ministry of Education, stratified by school-level refugee enrollment share (and geographically clustered for cost-effectiveness).
Randomization Unit
School
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
300 schools (100 Control; 100 Treatment 1; 100 Treatment 2)
Sample size: planned number of observations
Approximately 15,000 Grade 3 students and 750 Grade 3 teachers across 300 schools
Sample size (or number of clusters) by treatment arms
Control: 100 schools; ~5,000 students; ~250 teachers
Treatment 1: 100 schools; ~5,000 students; ~250 teachers
Treatment 2: 100 schools; ~5,000 students; ~250 teachers
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Student learning outcomes (EGRA/EGMA): 300 school clusters (100/arm), ~50 students/cluster, intra-cluster correlation = 0.15, significance level = 0.05, assumed attrition = 10%, assumed take-up = 60%. Minimum detectable effect (MDE) = 0.162 SD at 80% power; 0.187 SD at 90% power (15,000 students total). Teacher outcomes (knowledge/mindset): 300 school clusters (100/arm), ~2.5 teachers/cluster, intra-cluster correlation = 0.05, significance level = 0.05, assumed attrition = 10%, assumed take-up = 70%. MDE = 0.26 SD at 80% power; 0.30 SD at 90% power (750 teachers total).
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number
Analysis Plan