AI evidence extraction

The R5E pattern: can artificial intelligence enhance programming skills development?

Yousri Attia Mohamed Abouelenein, Ayat Fawzy Ahmed Ghazala, Eman Mahdy Mohamed Mahdy, Mohamed Hassan Ragab Khalaf · Education and Information Technologies · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

6/10

Relevance

4/4

Quality (LMQS)

Evidence

Citations

5.19

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1007/s10639-025-13616-3

Methodology & findings

Study design

Extended One Group Pretest Post-test Design with two experimental groups (quasi-experimental).

Sample

N = 70, 5 groups

Primary method

SPSS Version 25 at 95% confidence level (α=0.05). Analyses included: descriptive statistics (mean, standard deviation), independent samples t-tests to compare pre-test scores between groups, paired samples t-tests to compare pre- and post-test scores within each group, General Linear Model (GLM) Two-Way ANOVA with repeated measures to examine interactions and main effects, Eta squared (η²) for effect sizes, and Pearson's correlation coefficient for associations between continuous data. Cooper's equation used for inter-rater reliability of observation card.

Main result

The study found that "substantial and statistically significant differences (p < 0.001) emerged between the groups on all variables post-intervention." The R5E pattern group (G1) demonstrated "considerably larger effect sizes (η² > 0.96), suggesting a more pronounced impact on learning outcomes compared to G2 (unstructured ChatGPT use), which also demonstrated large effect sizes (η² > 0.96)." The results demonstrate that "the magnitude of pre-to post-intervention improvement differed significantly between G1 (R5E Pattern) and G2 (unstructured ChatGPT utilization), thereby providing further support for the enhanced effectiveness of the structured R5E approach."

Reports effect sizes and confidence intervals.

Research paradigm

Positivist/Empirical - quasi-experimental design with quantitative outcome measurement

Author conclusions

The authors conclude: "Key research findings Our study at --------University revealed demonstrable improvements in student programming skills through the implementation of the R5E pattern coupled with ChatGPT. This positive impact was observed in both cognitive understanding and practical application. The findings highlight the importance of well-structured pedagogical frameworks when leveraging AI tools in education." They further state: "This research demonstrates the significant potential of integrating the R5E pattern with ChatGPT to enhance programming education, leading to improved student outcomes in both cognitive and practical skills. The structured approach presented offers a replicable model for maximizing the benefits of AI in education." Additionally: "the current findings suggest the potential for developing the R5E pattern for using ChatGPT into a comprehensive and established model for utilizing AI in education."

Risk of bias

Small sample size (n=70 total, n=35 per group) limiting generalizability; Exclusively male participants due to non-coeducational university setting; No random assignment to conditions (groups assigned but context-dependent); Immediate post-test administration without delayed follow-up, introducing recall bias; Instructor training differences between groups potentially introducing expectancy effects; Lack of blinding to treatment condition; Single institution study limiting external validity; Selection bias: Small non-random sample of 70 participants from 102 eligible students; Attrition: Not explicitly stated if any participants dropped out; Confounding: Instructor effect - different instructors for each group may introduce bias; Temporal confound: Post-test immediately after treatment may reflect test-taking recall rather than learning retention; Gender bias: All participants male due to institutional segregation; Location bias: Single institution in Saudi Arabia, single geographic region (2024 academic year); Hawthorne effect: Awareness of participation in research may influence behavior; Instructor allegiance bias: Instructor trained in R5E pattern may differentially favor Group 1; Selection bias: All-male sample due to institutional gender segregation; non-coeducational environment limits generalizability to female students; Small sample size (n=70 total, n=35 per group) restricts generalizability; Lack of delayed post-test assessment: Post-test administered immediately after treatment may inflate results due to recall effects rather than retention; Potential instructor bias: Different instructors trained for each group (R5E vs. conventional); instructor expectancy effects not controlled; Attrition not explicitly reported; unclear if all 70 participants completed all procedures; No randomization mentioned for group assignment beyond initial 'random sample' selection; Confounding variables from concurrent academic demands (midterm/final exams, course projects) acknowledged but not controlled; Hawthorne effect: Structured R5E approach may create observer effects where students perform better due to increased attention

Limitations

The authors state: "This study focuses only on the design and implementation of the third and fourth chapters of the 'Programming and Applications' course for second-level students in the Information Technology program." They also note: "The limited duration of the post-test administration and its occurrence immediately after the experimental treatment may have introduced a confounding effect on the results
Therefore, the results might be influenced by students' recall of learned responses." Additionally: "Limitations related to the small sample size should also be acknowledged
This restricts the generalizability of the findings to the case in the current study, making it difficult to generalize the results to all students at the university where the research was conducted
Furthermore, the nature of the sample, being exclusively male, may limit the generalizability of the findings to female students at the same university."

Open questions raised

The authors identify the need for: (1) expansion of sample size and diversity to improve generalizability; (2) delayed assessment methods to address recall bias from immediate post-testing; (3) longitudinal studies using mixed-methods approaches to understand long-term impact on student learning, knowledge retention, and skill transferability; (4) broader investigation of R5E pattern application across diverse academic disciplines beyond programming; (5) exploration of the R5E pattern's effectiveness with female students and in coeducational settings; (6) development of comprehensive plans to expand AI application use in university education while activating principles of learning theories.
Long-term retention effects: Post-test conducted immediately after treatment; delayed assessment methods needed
Generalizability across contexts: Sample limited to single institution, single gender, single geographic region
Broader disciplinary application: Study limited to programming; findings transferability to other academic disciplines unclear
Female student outcomes: Gender-segregated institutional setting prevented female participant inclusion
Larger sample sizes: Need for expanded sample diversity to increase statistical power

Data: No datasets reported as publicly available; no repository links provided.Code: No code repositories mentioned.Extracted from: pdfAgreement 56%

Explore related topics

AI Literacy Training Academic Writing Experimental