# SYN-01 — Synthetic Study for Stage 2 Practice

> TRAINING DOCUMENT — This is a fictional study created for Notebook analysis practice. It is not a real publication and must not be cited as research evidence.

## Record information
Study ID: SYN-01  
Year: 2024 (fictional)  
Title: Short AI-generated hints for novice programming practice  

## Abstract
This fictional pilot examined whether an AI-supported hint tool helped first-year computing students during introductory Python exercises. Forty-eight students from one university used the tool for six weeks. Students were assigned to either an AI-hint group or a conventional help-page group. The AI tool provided a short hint after a student requested help; it did not provide a complete solution. The report states that the AI-hint group had a higher mean score on an immediate programming test (78/100) than the comparison group (69/100). The authors also report that the groups completed a similar number of exercises. The study was conducted in one course and did not assess retention beyond the course.

## 1. Introduction
Novice programmers often need timely feedback while practising syntax and problem-solving. This fictional study explored a hint-based tool that responded to a learner's specific question. The tool was designed to encourage another attempt rather than produce a full answer.

## 2. Method
The pilot included 48 first-year computing students enrolled in one introductory Python course. Participants were assigned to an AI-hint group (n=24) or a help-page group (n=24). The intervention lasted six weeks. Both groups completed the same weekly exercises. The AI-hint group could request a short contextual hint. The comparison group could consult a static help page. The report does not describe allocation concealment, and it does not state whether the test was marked blind to group.

## 3. Findings
The immediate programming test mean was 78/100 for the AI-hint group and 69/100 for the help-page group. The report describes this difference as statistically significant, but it does not provide a confidence interval. The number of exercises completed was similar between groups. In an end-of-course questionnaire, 17 of 24 students in the AI-hint group said the hints helped them continue when stuck. The questionnaire instrument is not described in detail.

## 4. Limitations
The study involved one course at one university, a small sample and a six-week intervention. The report did not evaluate longer-term retention or transfer to a new programming language. The hint system sometimes suggested a next debugging step that did not match the student's actual error; the frequency of this issue was not reported.

## 5. Conclusion
The fictional authors suggest that short, on-demand AI hints may support immediate programming practice. They caution that results may not generalize beyond the course and that hint accuracy needs further evaluation.