Prompt Engineering As Mediator on the Relationship Between Artificial Intelligence (AI) Writing Assistant Usage and Research Writing Competence of Learners

ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 Prompt Engineering As Mediator on the Relationship Between Artificial Intelligence (AI) Writing Assistant Usage and Research Writing Competence of Learners Dave John M. Villarosa Abstract. This study investigated the mediating effect of prompt engineering on the relationship between the use of Artificial Intelligence (AI) writing assistants and research writing competence among senior high school learners in Cluster 2 Public Secondary Schools in Davao City, using Baron and Kenny’s (1986) mediation framework within a non-experimental, descriptive-correlational design. A total of 293 learners were selected through stratified random sampling from a population of 1,150. Results revealed that AI writing assistant usage was moderately extensive, with ethical and appropriate application being the most evident dimension, while frequent tool engagement was the least observed. Research writing competence was also moderately extensive, with logical organization being most evident and synthesis of the literature review the least demonstrated. Prompt engineering showed similar moderate extensiveness, with creativity in prompt variation being frequently observed, whereas tailoring prompts for information accuracy was the weakest. Correlation analysis confirmed strong and significant positive relationships among all variables, supporting the interconnected role of AI use, writing competence, and prompt construction. Mediation analysis demonstrated that prompt engineering significantly and positively mediates the effect of AI usage on research writing competence. The results support the Technology Acceptance Model (TAM) and SelfRegulated Learning Theory, while partially challenging the Cognitive Load Theory due to the potential complexity of prompts. KEY WORDS 1. AI writing assistant 2. research writing competence 3. prompt engineering 4. mediation 5. digital literacy Date Received: August 15, 2025 — Date Reviewed: September 20, 2025 — Date Published: October 10, 2025 1. Introduction Learners’ inadequate and inconsistent research writing ability has become a growing concern in many schools, as students often struggle with constructing logical arguments, synthesizing information, and applying proper citation practices. These gaps in competence result in weak research outputs that limit their capacity to meet the demands of academic writing. In the researcher’s setting, the issue is more evident due to limited mastery and inconsistent practice of research-related tasks. Such conditions prompted the need to examine factors that can strengthen learners’ academic writing competence, particularly in the digital learning NIJSE (2025) - era. This issue serves as a basis for conducting a study on the mediating effect of prompt engineering on the relationship between the use of Artificial Intelligence (AI) writing assistants and research writing competence among senior high school students. In the United Kingdom, the issue of inadequate and inconsistent research writing ability among learners is often linked to difficulties in developing coherence and critical thinking in written tasks. According to Li and Han (2022), many students face challenges in integrating evidence and structuring arguments, which results in fragmented academic outputs. Percy (2020) also pointed out that poor mastery of citation practices and referencing conventions contributes to weakened academic integrity. Consequently, learners tend to rely heavily on descriptive writing rather than critical synthesis. This pattern reflects a broader issue of limited academic writing competence within schools. In Hawaii, learners face similar struggles, as cultural and linguistic diversity often complicate the development of strong research writing abilities. Zorec (2024) emphasized that the inconsistent application of writing conventions weakens students’ ability to present logical arguments. Similarly, Kawano (2023) observed that many learners rely on surface-level narratives rather than engaging analytically with sources. These challenges lead to inconsistent writing quality across student outputs. Thus, limited exposure to structured research practices results in reduced consistency and depth in learners’ academic writing. In Singapore, despite its high educational standards, problems persist in research writing, particularly in terms of critical evaluation and synthesis. Majid et al. (2020) noted that many students display strong technical writing skills but struggle in constructing arguments that demonstrate higherorder thinking. Likewise, Wong and Ng (2020) explained that an overemphasis on formulaic structures can sometimes hinder creativity and the critical integration of sources. This results in research papers that are grammatically correct yet lack analytical depth. Hence, the issue of inadequate and inconsistent writing ability remains a challenge, even in advanced systems. In the Philippine setting, particularly in Luzon, learners’ research writing difficulties are often tied to limited training and insufficient guidance in handling academic sources. Capulso (2020) observed that many learners rely on unverified references, which undermines the credibility of their work. At the same time, Oestar and Marzo (2022) explained that weak organization of ideas results in poorly structured research outputs. These challenges highlight the lack of proficiency in fundamental academic writing skills among learners. Therefore, the problem of inconsistent research writing persists across many secondary schools in the region. In Mindanao, the challenges of inadequate research writing competence are compounded by limited access to resources and insufficient teacher support. Garao et al. (2023) emphasized that students often submit papers lacking coherence, with minimal integration of supporting evidence. Similarly, Mayasa (2022) highlighted the problem of repetitive and descriptive writing, which reflects a lack of critical analysis skills. These weaknesses hinder the development of well-argued and structured academic papers. As a result, the issue of limited academic writing competence remains significant in this area. In Cluster 2 Public Secondary Schools in Davao City, the problem became more pronounced due to inconsistencies in students’ exposure to research practices. Many learners struggled to sustain logical arguments throughout their writing tasks. Citation errors and reliance on paraphrasing without proper synthesis further weakened the overall quality of their research papers. These recurring issues reflected the poor and inconsistent development of academic writing competence among local learners. Hence, there was a pressing need to strengthen instructional approaches to research writing in the district, highlighting 2ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - the importance of addressing the problem within the researcher’s setting. Given these persistent problems and issues in students’ inadequate and inconsistent research writing abilities, it became important to explore the deeper factors that may influence their academic competence. In particular, the growing use of Artificial Intelligence (AI) writing assistants in education highlights both opportunities and challenges in supporting student learning. However, despite the increasing integration of these tools, several research gaps remain unaddressed. A temporal gap exists because most discussions on AI writing assistants are relatively recent, and there is still limited focus on how their use affects academic writing over time. A methodological gap was also noted, as many earlier studies relied on qualitative accounts or descriptive approaches, leaving a limited body of quantitative research that measures the direct and indirect relationships among AI usage, research writing competence, and prompt engineering. An empirical gap was further identified, as prompt engineering, although increasingly important in guiding AI tools to produce accurate and relevant outputs, has not been fully examined as a mediating factor. These gaps highlighted the need for a correlational quantitative study to provide systematic evidence on how AI usage and prompt engineering shape students’ research writing competence. Given these gaps, it became necessary to examine the issue within the local context to ensure relevance and applicability. In Cluster 2 Public Secondary Schools in Davao City, the urgency of conducting this study was evident, as students continued to show weak and inconsistent research writing competence despite exposure to digital tools. Many learners were unable to sustain logical arguments, synthesize sources, or apply proper citation, which further impacted the quality of their academic outputs. Addressing this concern required a closer examination of how AI writing assistants, when supported by effective prompt engineering, could potentially enhance students’ research writing performance. Thus, the study was timely and socially relevant, as its findings could guide teachers, administrators, and policymakers in maximizing digital innovations for improved academic practices in the district. 1.1. Review of Significant Literature— 1.1.1. Artificial Intelligence (AI) Writing Assistant Usage—AI writing assistants use natural language processing to help users generate, edit, and refine written content (Intiser et al., 2023). These tools support error correction, structural organization, and coherence improvement, enhancing students’ academic writing competence (Zhao, 2023). They promote autonomy by reducing dependence on teachers for basic revisions, allowing educators to focus on higher-order thinking skills (Huete-Garc ´ ıa Tarp, 2024; Ippolito et al., 2022). AI writing tools foster inclusivity by assisting students with language barriers and improving academic performance across diverse learners (Gayed et al., 2022; Bibi Atta, 2024). Their adaptive feedback mechanisms strengthen writing independence and confidence, empowering students to produce more coherent, researchaligned papers (Etaat, 2024; Song Song, 2023; Daulay et al., 2024). 1.1.2. Frequent Tool Engagement—Frequent interaction with AI writing assistants develops linguistic accuracy, stylistic awareness, and writing confidence (Intiser et al., 2023; Rad et al., 2024). Students who regularly use these tools gain better mastery of grammar and structure, enabling teachers to emphasize critical analysis and argumentation (Hew et al., 2023; Godwin-Jones, 2022). Routine AI engagement also fosters digital literacy and adaptability, creating a techintegrated learning environment that mirrors professional communication practices (Kim Kim, 2022; Zhang, 2024). 1.1.3. Prompt Customization and Monitoring—Prompt customization involves tailor3ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - ing AI inputs to yield precise, context-relevant outputs. This practice enhances learning engagement, accountability, and goal alignment (Intiser et al., 2023; Sajja et al., 2024). Teachers and students who refine prompts can better guide AI toward task-specific outcomes, promoting personalization and accuracy (Boynagryan Tshngryan, 2024; Sikha et al., 2023). Moreover, refining and monitoring outputs fosters critical evaluation and digital literacy, as students learn to assess and adapt AI-generated information (Zhang et al., 2023; Yeh et al., 2024). 1.1.4. Ethical and Appropriate Application—Ethical use of AI writing assistants ensures academic integrity and responsible authorship (Intiser et al., 2023). Responsible users employ AI as a supplementary aid while maintaining ownership of ideas and proper citation (Khalifa Albadawy, 2024). Balanced use encourages active engagement and critical assessment of feedback, preventing overreliance on automation (Perkins, 2023; Singh et al., 2024). Structured curricular integration of AI ethics reinforces integrity, emphasizing transparency and fairness in tool use (Board, 2019; Kerr, 2020). 1.1.5. Refinement of Final Drafts—AIassisted proofreading enhances coherence, readability, and grammatical precision (Intiser et al., 2023; Singh et al., 2024). Moderate tool use strengthens student accountability and engagement in revision processes, reinforcing understanding of writing conventions (Li et al., 2024; Guo et al., 2022). AI tools also ensure consistent feedback and fairness in evaluation, especially in large classrooms, supporting equitable learning outcomes (Carobene et al., 2024; Lin, 2023). 1.1.6. Research Writing Competence—Research writing competence involves planning, organizing, and presenting arguments effectively (Wang Fan, 2020). It encompasses clarity, coherence, proper citation, and ethical research communication (Meza Gonz ´ alez, 2020). Strong writing skills foster independence, analytical thinking, and collaboration in group research projects (Mulyaningsih et al., 2022; Prosekov et al., 2020). Well-developed research writing abilities also prepare students for advanced academic and professional tasks requiring evidence-based reporting (Keller et al., 2020; Wale Bogale, 2021). 1.1.7. Comprehensive Literature Review— A strong literature review demonstrates the ability to synthesize credible sources and avoid redundancy (Wang Fan, 2020; Astuti et al., 2020). It strengthens research design and fosters critical inquiry (D’Souza, 2021; Camacho et al., 2021). Advanced review skills also prepare students for research-oriented careers requiring evidence-based analysis (Baresh, 2022; Suwartini et al., 2021). 1.1.8. Clear Thesis Formulation—A precise thesis guides the research’s direction and coherence (Wang Fan, 2020). Competence in thesis formulation supports critical thinking and argument development (Sugumlu, 2020; Andhini Sakti, 2021). It also improves collaboration and interdisciplinary integration (Ismiati Pebriantika, 2020; Shokarimova, 2021). 1.1.9. Logical Organization of Content— Logical organization ensures structured argumentation and clarity (Wang Fan, 2020; Hassan et al., 2021). Students with strong organizational skills present knowledge more coherently, improving academic engagement and teamwork (Rojas et al., 2020; Warlizasusi et al., 2023). 1.1.10. Accurate Citation and Referencing—Proper citation fosters academic honesty and respect for intellectual contributions (Wang Fan, 2020; Rababah, 2022). Accurate referencing encourages critical engagement with sources and builds ethical research habits (Andheska et al., 2020; Pentury et al., 2020). 1.1.11. Prompt Engineering—Prompt engineering—the ability to craft effective 4ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - queries—enhances communication with AI systems and supports creative, analytical, and problem-solving skills (Woo et al., 2024; Mzwri Turcs ´ anyi-Szabo, 2025). It refines clarity and focus in research writing (Park Choo, 2024; Heston Khun, 2023) and reduces cognitive overload by improving information retrieval (Federiakin et al., 2024; Kale et al., 2024). Strong prompting skills correlate with clearer language use, efficient research navigation, and responsible AI integration (Lee et al., 2024; Zaghir et al., 2024). These abilities optimize the educational benefits of AI tools, preparing students for technology-driven academic and professional environments (Haugsbaken Hagelia, 2024). 1.2. Synthesis—The use of Artificial Intelligence (AI) writing assistants is increasingly becoming a vital tool in academic writing, as they help students generate, organize, and refine their ideas efficiently. Research writing competence refers to the ability of students to produce well-structured, clear, and properly cited academic work. Prompt engineering is the skill of crafting effective queries and instructions to guide AI systems in generating useful outputs. These three elements work together to support students’ learning processes and improve their overall writing quality. Together, they form an integrated approach that enhances both the process and product of academic writing. The relationship between research writing competence and prompt engineering is significant when using AI writing assistants. When students develop effective prompt engineering skills, they can better direct the AI to produce relevant content that aligns with their research needs. This, in turn, enhances their ability to structure their arguments and logically organize their research writing. As a result, students not only improve their writing competence but also learn to evaluate the information generated by AI tools critically. Hence, prompt engineering acts as a mediating skill that bridges AI assistance and improved research writing outcomes. Overall, the combined usage of AI writing assistants, strong research writing competence, and effective prompt engineering leads to a more refined academic writing process. This synergy enables students to overcome common writing challenges by leveraging technology to generate ideas and organize content more efficiently. Enhanced prompt engineering skills enable students to obtain more accurate and helpful outputs from AI systems, thereby directly influencing the quality of their final drafts. The resulting enhancement in research writing competence can lead to better academic performance and greater confidence in scholarly work. Thus, these interconnected factors pave the way for a modern and effective approach to academic writing that benefits students across various disciplines. 1.3. Theoretical/Conceptual Framework— This study is anchored on the Technology Acceptance Model (TAM) by Davis (1989), which explains that users’ acceptance of technology is determined by their perceptions of its usefulness and ease of use. It posits that these perceptions directly influence the intention to use and the actual use of technology. This theory is suitable for the study as it helps explain how prompt engineering can affect students’ perceptions of AI writing assistants, making these tools appear more user-friendly and beneficial. When students develop a strong proficiency in prompt engineering, they are likely to perceive AI tools as easier to use and more effective in enhancing their research writing skills. Thus, applying TAM allows the study to investigate how improved prompt engineering mediates the impact of AI writing assistant usage on research writing competence. In support, Self-Regulated Learning Theory, as proposed by Zimmerman (1989), explains that learners actively control their own learning through goal setting, selfmonitoring, and self-reflection. It highlights the role of internal motivation and strategic plan5ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - ning in achieving academic success. This theory is particularly suitable for the study because prompt engineering can be viewed as a selfregulatory strategy that enables students to manage better and direct their use of AI writing assistants. Through effective prompt engineering, students can set clear research goals, monitor their progress, and adjust their writing strategies accordingly. In addition, Cognitive Load Theory, as proposed by Sweller (1988), asserts that the human brain has a limited capacity for processing information, and effective instructional methods should minimize unnecessary cognitive load. The theory emphasizes managing intrinsic, extraneous, and germane cognitive loads to optimize learning outcomes. This theory was appropriate for the study because it provides insight into how prompt engineering can reduce the extraneous cognitive load associated with interacting with AI writing assistants. By streamlining and clarifying the process through well-crafted prompts, students can focus more on generating quality content rather than being overwhelmed by complex interfaces. This study was composed of three variables, as shown in Figure 1. The independent variable is the usage of Artificial Intelligence (AI) Writing Assistants, which involves the practice of utilizing AI-powered tools to generate, edit, or refine written content. The measures of Artificial Intelligence (AI) Writing Assistant Usage are frequent tool engagement or the consistent use of AI writing assistants to create initial drafts or expand on existing content; prompt customization and monitoring or the act of adjusting the inputs or prompts given to the AI writing assistant to yield more accurate results; ethical and appropriate application or the employment of AI writing assistants in a manner that respects academic honesty and avoids plagiarism; and refinement of final drafts or the use of AI writing assistant to polish and proofread text, ensuring coherence and readability. The dependent variable was the research writing competence of learners or the ability to plan, organize, and present research findings effectively, following standard academic conventions. The indicators of research writing competence of students were comprehensive literature review or the student’s ability to find, evaluate, and synthesize various credible sources related to the research topic; clear thesis formulation or the student’s skill in developing a strong, focused statement that guides the overall direction of the research; logical organization of content or the student’s proficiency in arranging ideas and sections in a systematic order that enhances clarity and coherence; and accurate citation and referencing or the student’s practice of properly acknowledging all sources used in the research through recognized citation styles. Lastly, the mediator is prompt engineering, or the skill that involves crafting queries and instructions to effectively communicate with and manipulate outputs from AI and computer systems. 1.4. Statement of the Problem—This study aimed to examine the mediating effect of prompt engineering on the relationship between the usage of Artificial Intelligence (AI) writing assistants and the research writing competence of learners in Cluster 2 Public Secondary Schools in Davao City. Hence, the study sought the answer to the following questions: (1) What is the extent of Artificial Intelligence (AI) writing assistant usage in terms of: (1) frequent tool engagement; (2) Prompt customization and monitoring; (3) ethical and appropriate application; and (4) Refinement of final drafts? 6ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - Fig. 1. Theoretical/Conceptual Framework of the Study 7ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - (2) What is the extent of research writing competence of learners in terms of: (1) Comprehensive literature review; (2) clear thesis formulation; (3) logical Organization of content; and (4) accurate citation and referencing? (3) What is the extent of prompt engineering? (4) I s there a significant relationship among Artificial Intelligence (AI) writing assistant usage, research writing competence of learners, and prompt engineering? (5) D oes prompt engineering significantly mediate the relationship between Artificial Intelligence (AI) writing assistant usage and the research writing competence of learners? 1.5. Hypotheses—The following hypotheses were tested at the 0.05 level of significance: H01: There is no significant relationship among Artificial Intelligence (AI) writing assistant usage, research writing competence of students, and prompt engineering. H02: Prompt engineering does not significantly mediate the relationship between Artificial Intelligence (AI) writing assistant usage and research writing competence of learners. For a more comprehensive understanding, the following terms were defined operationally: Artificial Intelligence (AI) Writing Assistant Usage It was defined conceptually as the practice of using AI-powered tools to generate, edit, or refine written content. In this study, the independent variable was described in terms of the following indicators: frequent tool engagement, prompt customization and monitoring, ethical and appropriate application, and refinement of final drafts. Research Writing Competence It was defined conceptually as the ability to plan, organize, and present research findings effectively, following standard academic conventions. In this study, the dependent variable was described through a comprehensive literature review, a clear thesis formulation, logical organization of content, and accurate citation and referencing. Prompt Engineering This was described as a skill involving crafting queries and instructions to effectively communicate with and manipulate outputs from AI and computer systems. In this study, the term refers to the variable that explains how or why the independent variable influences the dependent variable in a causal relationship. 2. Methodology This section provides a comprehensive overview of the research design, including details on the research respondents, ethical considerations, research instruments, and procedural steps. It also outlines the methods for data collection and analysis, ensuring a clear framework for the study. 2.1. Research Design—In this study, the researcher used a quantitative descriptivecorrelational research design to achieve the objectives stated in the previous chapter. Quantitative research design involved the systematic empirical investigation of observable phenomena via numerical data and statistical techniques. It aimed to quantify relationships between variables and often resulted in outcomes that were generalizable to larger populations (Sukamolson, 2007). This design was appropriate for studying the mediating effect of prompt engineering because it allowed for precise measurement of relationships between variables, such as AI writing assistant usage and research writing competence. By employing quantitative methods, researchers were able to statistically analyze how changes in prompt engineering af8ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - fected the efficacy of AI tools in improving writing skills. The use of such a design also facilitated the testing of hypotheses derived from theoretical frameworks about technology integration in education. Meanwhile, descriptive research methods are aimed at accurately and systematically describing a population, situation, or phenomenon. These methods included surveys, observations, and case studies, which helped to provide a detailed picture of current conditions (Siedlecki, 2020). In the context of this study, descriptive methods were used to gather detailed data about how students used AI writing assistants and their competence in prompt engineering. This information established a baseline of current practices and competencies before examining the relationships between these variables. Descriptive data were crucial for identifying and understanding the variables that were later analyzed through more complex statistical techniques. Further, a correlational research approach explored the relationship between two or more variables without determining a causeand-effect link. It helped to identify patterns and predict relationships among variables based on statistical correlation coefficients (Sullivan, 2024). This approach was suitable for the initial stages of the study, which aimed to explore the extent to which AI writing assistant usage and research writing competence were associated with levels of prompt engineering skill. It provided preliminary insights that informed the development of more detailed mediation models. Moreover, understanding these correlations was fundamental to setting up appropriate mediation analyses to test the hypothesized effects. Furthermore, Mediation Analysis using the Baron and Kenny (1986) Method, including the Sobel z-test, was employed to understand how an independent variable influenced a dependent variable through a mediator. It involved several regression analyses to determine if the mediator influenced the relationship between the independent variable and the dependent variable. The Sobel z-test then tested the significance of the mediation effect (MacKinnon et al., 2007). This analytical approach was particularly relevant for this study as it allowed the researcher to quantitatively assess how prompt engineering (mediator) modified the impact of AI writing assistant usage (independent variable) on students’ research writing competence (dependent variable). It provided a structured method to discern the indirect effects of AI tools, facilitated by students’ ability to engineer prompts effectively. This approach offers a comprehensive understanding of how these elements interact in educational settings. 2.2. Ethical Considerations—In this research, ethical considerations played a pivotal role in guiding the conduct of the study to ensure integrity and respect for all respondents. Following a thorough review and subsequent approval from the Ethics Committee, the researcher commenced data gathering activities. This approval was essential as it confirmed that the study upheld the highest ethical standards, protected the welfare of all involved, and ensured that the research processes were transparent and accountable. Social Value This investigation potentially contributed to the educational community by identifying effective methods to enhance learners’ writing skills. By understanding how prompt engineering improved AI writing assistant usage, educators and policymakers were able to develop training programs that benefited both teachers and learners. The findings likely inspired the adoption of technology-driven teaching approaches aimed at elevating academic performance, thereby creating a more competitive and competent generation of students who could contribute positively to the community. Moreover, the study provided insights into how modern digital tools reshaped classroom learning experiences for senior high school students. By focusing on prompt engineering, it highlighted the importance of proper 9ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - idea generation. Moreover, Hew et al. (2023) stressed that frequent AI usage in writing fosters iterative thinking and reflective revision, which are essential for developing strong written communication skills. The individual mean scores for this domain range from 3.09 to 3.38, all within the moderately extensive category. The highest-rated statement is Engaging regularly with the AI writing assistant during academic tasks, with a mean of 3.38, implying that students moderately integrate AI tools into routine schoolwork. Meanwhile, the lowest-rated statement is Relying on the AI writing assistant repeatedly to enhance the overall writing process, with a mean of 3.09, indicating that students are less inclined to depend on the tool for comprehensive writing enhancement. This finding aligns with the observation of Zhang (2024), who noted that while students may experiment with AI tools, they often underutilize them for deeper revision and development purposes. Table 1. Artificial Intelligence (AI) Writing Assistant Usage in Terms of Frequent Tool Engagement No. Statements Mean Descriptive Rating 1 Engaging regularly with the AI writing assistant during academic tasks. 3.38 Moderately Extensive 2 Utilizing the AI tool consistently to support writing drafts. 3.11 Moderately Extensive 3 Accessing the AI writing assistant frequently to generate new ideas. 3.15 Moderately Extensive 4 Interacting with the AI system often to check for grammar and style improvements. 3.17 Moderately Extensive 5 Relying on the AI writing assistant repeatedly to enhance the overall writing process. 3.09 Moderately Extensive Mean 3.18 Moderately Extensive 3.1.2. Prompt Customization and Monitoring—In Table 2, the mean for Artificial Intelligence (AI) Writing Assistant usage in terms of prompt customization and monitoring is 3.38, interpreted as moderately extensive. This indicates that students sometimes engage in customizing prompts and monitoring AI outputs, though these practices are not consistently performed across academic writing tasks. The moderately extensive rating reflects a developing ability to direct AI-generated content through intentional prompt engineering and active oversight. As highlighted by Boynagryan and Tshngryan (2024), students who take initiative in refining AI prompts and reviewing outputs tend to produce more accurate, personalized, and academically relevant texts. Similarly, Sajja et al. (2024) emphasized the importance of prompt customization in fostering digital literacy and critical thinking, particularly when students are guided in iteratively improving AI-generated content. The individual mean scores for this domain range from 3.17 to 3.55, indicating levels from moderately extensive to extensive. The highest-rated statement is Customizing prompts effectively to tailor the AI responses to specific writing needs, with a mean of 3.55, suggesting that students are frequently able to modify inputs to receive more relevant and useful AI outputs. Conversely, the lowest-rated statement is Adjusting prompt settings based on feedback to better guide the AI’s assistance, with a mean of 3.17, indicating a less consistent practice of refining prompts based on reflective feedback. This mirrors the findings of Sikha et al. (2023), who observed that while learners may initially 16 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - customize prompts, they often lack strategies for systematically adjusting them based on output quality. Table 2. Artificial Intelligence (AI) Writing Assistant Usage in Terms of Prompt Customization and Monitoring No. Statements Mean Descriptive Rating 1 Customizing prompts effectively to tailor the AI responses to specific writing needs. 3.55 Extensive 2 Modifying input prompts continually to improve the quality of generated text. 3.41 Extensive 3 Monitoring the outputs from the AI tool to ensure they remain relevant and accurate. 3.53 Extensive 4 Adjusting prompt settings based on feedback to better guide the AI’s assistance. 3.17 Moderately Extensive 5 Reviewing AI responses regularly to refine subsequent prompt configurations. 3.23 Moderately Extensive Mean 3.38 Moderately Extensive 3.1.3. Ethical and Appropriate Application—In Table 3, the mean for Artificial Intelligence (AI) Writing Assistant usage, in terms of ethical and appropriate application, was 3.42, which was interpreted as extensive. This indicates that students oftentimes apply the AI tool in a manner aligned with ethical standards and institutional policies. The extensive rating reflects a generally strong awareness among students of responsible AI usage, particularly in maintaining academic honesty and observing guidelines on plagiarism and citation. Table 3. Artificial Intelligence (AI) Writing Assistant Usage in Terms of Ethical and Appropriate Application No. Statements Mean Descriptive Rating 1 Using the AI writing assistant in an ethical manner that upholds academic integrity. 3.63 Extensive 2 Applying the AI tool appropriately without violating plagiarism policies. 3.23 Moderately Extensive 3 Following institutional guidelines when integrating AI-generated content into assignments. 3.60 Extensive 4 Integrating AI assistance responsibly to complement traditional writing methods. 3.39 Moderately Extensive 5 Observing proper citation practices when referencing AIproduced material in academic work. 3.25 Moderately Extensive Mean 3.42 Extensive As discussed by Singh et al. (2024), ethical engagement with AI tools enhances students’ academic integrity and fosters a sense of accountability in digital learning environments. Adding more, Khalifa and Albadawy (2024) emphasized that institutions must promote responsible AI practices by educating students about the boundaries and expectations when using generative technologies for academic purposes. The individual mean scores for this do17 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - main range from 3.23 to 3.63, spanning ratings from moderately extensive to extensive. The highest-rated statement was Using the AI writing assistant in an ethical manner that upholds academic integrity, with a mean of 3.63, suggesting that students were highly conscious of maintaining ethical standards when employing AI tools. On the other hand, the lowest-rated statement is Applying the AI tool appropriately without violating plagiarism policies, with a mean of 3.23, indicating some uncertainty or inconsistency in fully understanding or adhering to plagiarism rules. This aligns with the findings of Perkins (2023), who noted that while students are generally aware of academic integrity, practical application, especially regarding the originality of AI-generated text, remains a challenge. 3.1.4. Refinement of Final Drafts—On Table 4, the mean for Artificial Intelligence (AI) Writing Assistant usage in terms of refinement of final drafts is 3.22, interpreted as moderately extensive. This suggests that students sometimes utilize the AI writing assistant to revise and enhance their final outputs, but such practices have not yet been consistently integrated into their writing process. The moderately extensive rating points to a growing, yet incomplete, adoption of AI-assisted revision strategies, where students recognize the potential of the tool but may lack the skills or habits to capitalize on it fully. According to Singh et al. (2024), refining drafts with AI can improve writing quality, particularly in grammar, coherence, and organization, when students are guided to interpret and apply AI feedback effectively. Likewise, Guo et al. (2022) highlighted that iterative revision supported by AI fosters self-editing skills and metacognitive awareness, especially when students are trained to engage critically with automated suggestions. The individual mean scores for this domain range from 3.15 to 3.39, all within the moderately extensive category. The highest-rated statement is Revising final drafts using the suggestions provided by the AI writing assistant, with a mean of 3.39, indicating that students moderately value the tool’s input during final revisions. Conversely, the lowest-rated statement is ”Enhancing overall clarity and coherence of final drafts through systematic AI-supported revisions,” with a mean of 3.15. This suggests that students less frequently apply the AI tool in a structured, comprehensive manner during the final stages of editing. This echoes the findings of Li et al. (2024), who noted that while students are open to using AI for quick fixes, they often overlook its potential for more strategic revisions. Table 4. Artificial Intelligence (AI) Writing Assistant Usage in Terms of Refinement of Final Drafts No. Statements Mean Descriptive Rating 1 Revising final drafts using the suggestions provided by the AI writing assistant. 3.39 Moderately Extensive 2 Editing written work based on iterative feedback from the AI tool. 3.16 Moderately Extensive 3 Polishing the final version of texts by incorporating AI-generated corrections. 3.21 Moderately Extensive 4 Fine-tuning academic papers with the help of targeted AI recommendations. 3.17 Moderately Extensive 5 Enhancing overall clarity and coherence of final drafts through systematic AI-supported revisions. 3.15 Moderately Extensive Mean 3.22 Moderately Extensive 18 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - 3.1.5. Summary on Artificial Intelligence (AI) Writing Assistant Usage among Senior High School Learners—On Table 5, the overall mean for Artificial Intelligence (AI) Writing Assistant usage among senior high school learners in Cluster 2 Public Secondary Schools in Davao City is 3.30, interpreted as moderately extensive. This indicates that while learners sometimes utilize AI tools across various aspects of the writing process, such engagement is not yet deeply embedded in their academic routines. The moderately extensive rating reflects an emerging familiarity with AI writing technologies, suggesting that students are in the early stages of integrating these tools into their learning practices. Table 5. Summary of Artificial Intelligence (AI) Writing Assistant Usage among Senior High School Students in Cluster 2 Public Secondary Schools in Davao City No. Statements Mean Descriptive Rating 1 Frequent Tool Engagement 3.18 Moderately Extensive 2 Prompt Customization and Monitoring 3.38 Moderately Extensive 3 Ethical and Appropriate Application 3.42 Extensive 4 Refinement of Final Drafts 3.22 Moderately Extensive Overall Mean 3.30 Moderately Extensive As emphasized by Ippolito et al. (2022), moderate usage levels often reflect a need for structured exposure, digital literacy training, and teacher support to guide students in maximizing AI functionalities. Similarly, Zhao (2023) observed that students benefited more from AI writing tools when they were taught not just how to use them, but when and why to apply them at different stages of the writing process. Among the domains assessed, Ethical and Appropriate Application received the highest mean of 3.42, interpreted as extensive. In contrast, Frequent Tool Engagement registered the lowest mean at 3.18, still moderately extensive. This pattern indicates that although students recognize the ethical boundaries of AI usage, they may require further encouragement and guidance to use these tools more frequently and effectively. Echoing the findings of HueteGarc ´ ıa and Tarp (2024), ethical understanding often precedes functional mastery, underscoring the importance of developing both dimensions concurrently. 3.2. Research Writing Competence—Research writing competence of learners in this study is measured in terms of comprehensive literature review; clear thesis formulation; logical organization of content; and accurate citation and referencing. The extent of this variable and its domains are presented in Tables 6-10. 3.2.1. Comprehensive Literature Review— On Table 6, the mean for the research writing competence of students in terms of comprehensive literature review was 3.29, interpreted as moderately extensive. This suggests that students sometimes demonstrate the ability to conduct a thorough literature review, but such competence was not yet consistently evident in their academic writing. The moderately extensive rating suggests a developing skill set in sourcing, analyzing, and integrating research findings, indicating a need for further instruction and guided practice. According to D’Souza (2021), students with moderate literature review competence may understand basic research synthesis techniques but often struggle with higher-order tasks such as identifying gaps and establishing thematic connections. In support, Camacho et al. (2021) noted that scaffolded interventions such as annotated bibliographies and peer-reviewed critiques can significantly enhance students’ ability to pro19 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - duce well-structured and critical literature reviews. The individual mean scores for this domain range from 3.16 to 3.55, spanning ratings from moderately extensive to extensive. The highest-rated statement was Highlighting gaps in existing research that warrant further investigation, with a mean of 3.55, indicating that students oftentimes recognize the importance of contributing to ongoing scholarly discourse. Conversely, the lowest-rated statement is Evaluating a wide range of sources to gather relevant research information, with a mean of 3.16, suggesting that students sometimes struggle in identifying diverse and credible sources for their reviews. This finding supports the observations of Baresh (2022), who reported that many senior high school students rely heavily on limited or familiar references, often neglecting breadth and quality in their source selection. Table 6. Research Writing Competence of Students in Terms of Comprehensive Literature Review No. Statements Mean Descriptive Rating 1 Evaluating a wide range of sources to gather relevant research information. 3.16 Moderately Extensive 2 Identifying key themes and trends across various scholarly articles. 3.21 Moderately Extensive 3 Summarizing findings from multiple studies in a clear manner. 3.25 Moderately Extensive 4 Synthesizing the literature to create an integrated overview of the topic. 3.29 Extensive 5 Highlighting gaps in existing research that warrant further investigation. 3.55 Extensive Mean 3.29 Moderately Extensive 3.2.2. Clear Thesis Formulation—On Table 7, the mean for the research writing competence of students in terms of clear thesis formulation was 3.36, interpreted as moderately extensive. This suggests that students sometimes demonstrate the ability to formulate a clear and coherent thesis statement, yet this skill was not consistently evident across all aspects of their research writing. The moderately extensive rating implies that while students possess a basic understanding of thesis development, many still require guided practice in articulating focused research arguments. According to Wahyuni et al. (2020), the capacity to craft a well-defined thesis statement was essential for structuring research papers, as it directs the logical flow of content and supports coherence. Likewise, Sugumlu (2020) emphasized that clear thesis formulation enhances readers’ comprehension and sets academic tone, making it vital for teachers to reinforce thesis development through workshops, modeling, and continuous feedback. The individual mean scores for this domain range from 3.16 to 3.68, covering both moderately extensive and extensive levels. The highestrated statement is Formulating a precise and focused thesis statement that guides the study, with a mean of 3.68, indicating that students oftentimes succeed in identifying and expressing their primary research focus. In contrast, the lowest-rated statement was ”Establishing a strong thesis that serves as the foundation for the research,” with a mean of 3.16, indicating some difficulty among students in constructing a thesis that unifies the overall study direction. This observation was consistent with the findings of Andhini and Sakti (2021), who noted that while learners can often state research goals, they struggle to integrate those goals into a compelling and cohesive thesis framework 20 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - Table 7. Research Writing Competence of Students in Terms of Clear Thesis Formulation No. Statements Mean Descriptive Rating 1 Formulating a precise and focused thesis statement that guides the study. 3.68 Extensive 2 Stating the research objectives clearly within the thesis. 3.50 Extensive 3 Defining the scope of the study in a concise manner. 3.18 Moderately Extensive 4 Articulating the central argument effectively and directly. 3.30 Moderately Extensive 5 Establishing a strong thesis that serves as the foundation for the research. 3.16 Moderately Extensive Mean 3.36 Moderately Extensive 3.2.3. Logical Organization of Content— On Table 8, the mean for the research writing competence of students in terms of logical organization of content was 3.52, interpreted as extensive. This indicates that students oftentimes exhibit the ability to organize their research papers in a coherent and structured manner. The extensive rating suggests a strong level of competence in arranging content from introduction to conclusion, with attention to logical flow, paragraph transitions, and clarity of argumentation. As noted by Hassan et al. (2021), well-organized research writing enhances reader comprehension and reinforces the persuasiveness of the thesis. In addition, Rojas et al. (2020) emphasized that systematic structuring of ideas not only supports academic rigor b u t also fosters students’ confidence in presenting their arguments effectively. The individual mean scores for this domain range from 3.28 to 3.77, with most statements falling under the extensive category. The highest-rated statement was Structuring the content logically from the introduction to the conclusion, with a mean of 3.77, showing that students frequently follow a clear and well-sequenced research structure. Meanwhile, the lowest-rated statement was Connecting paragraphs smoothly to ensure a continuous flow of ideas, with a mean of 3.28, indicating occasional difficulty in maintaining seamless transitions across sections. This finding was in line with the observations of Warlizasusi et al. (2023), who found that students may understand the general organization of research papers b u t struggle with intra-text cohesion and paragraph linkage. 3.2.4. Accurate Citation and Referencing— On Table 9, the mean for the research writing competence of students in terms of accurate citation and referencing was 3.32, interpreted as moderately extensive. This suggests that while students sometimes demonstrate the ability to cite and reference sources properly, these practices were not yet consistently applied across their research papers. The moderately extensive rating reflects a foundational understanding of citation principles, b u t highlights the need for continued reinforcement of proper academic documentation. As discussed by Rababah (2022), accurate citation practices were essential for upholding academic integrity and preventing unintentional plagiarism. Also, Sari et al. (2021) emphasized that instruction on citation styles and ethical research behavior must be embedded within writing activities to build students’ confidence and competence in source attribution. The individual mean scores for this domain range from 3.11 to 3.61, with most items falling within the moderately e x - tensive category. The highest-rated statement 21 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - Table 8. Research Writing Competence of Students in Terms of Logical Organization of Content No. Statements Mean Descriptive Rating 1 Organizing ideas in a systematic and coherent structure throughout the paper. 3.43 Extensive 2 Structuring the content logically from the introduction to the conclusion. 3.77 Extensive 3 Connecting paragraphs smoothly to ensure a continuous flow of ideas. 3.28 Moderately Extensive 4 Grouping related concepts together for better clarity and understanding. 3.71 Extensive 5 Sequencing the arguments in a manner that supports the overall thesis clearly. 3.43 Extensive Mean 3.52 Extensive is Listing all references clearly and accurately in the bibliography, with a mean of 3.61, indicating that students oftentimes complete bibliographic entries with relative accuracy. Conversely, the lowest-rated statement was Attributing information correctly to its original authors to avoid plagiarism, with a mean of 3.11, suggesting that students sometimes struggle with in-text attribution and source acknowledgment. This finding aligns with the insights of Andheska et al. (2020), who noted that many students were aware of citation requirements but encountered difficulties in consistently applying them, especially under time pressure or when using unfamiliar sources. Table 9. Research Writing Competence of Students in Terms of Accurate Citation and Referencing No. Statements Mean Descriptive Rating 1 Citing all sources accurately to maintain academic integrity. 3.57 Extensive 2 Referencing borrowed ideas using the proper citation style consistently. 3.16 Moderately Extensive 3 Attributing information correctly to its original authors to avoid plagiarism. 3.11 Moderately Extensive 4 Following established citation guidelines in all parts of the research paper. 3.13 Moderately Extensive 5 Listing all references clearly and accurately in the bibliography. 3.61 Extensive Mean 3.32 Moderately Extensive 3.2.5. Summary on Research Writing Competence—On Table 10, the overall mean for research writing competence among senior high school learners in Cluster 2 Public Secondary Schools in Davao City was 3.37, interpreted as moderately extensive. This indicates that learners sometimes demonstrate key academic writing skills, though these competencies are not yet consistently or comprehensively manifested in their research outputs. The moderately extensive rating suggests a developing level of proficiency in areas such as thesis formulation, literature synthesis, organization, and citation, underscoring the need for enhanced instruction and guided practice in academic research writing. According to Meza and Gonz ´ alez (2020), moderate competence in research writing reflects the transitional stage where students grasp 22 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - core principles but struggle with consistent application across all sections of a research paper. Likewise, Valizadeh (2020) highlighted that scaffolded feedback, peer collaboration, and iterative revisions are crucial strategies in helping students internalize and improve their research writing abilities. Among the assessed domains, Logical Organization of Content received the highest mean of 3.52, interpreted as extensive, suggesting that students oftentimes succeed in structuring their papers coherently from introduction to conclusion. In contrast, the Comprehensive Literature Review registered the lowest mean at 3.29, which was still within the moderately extensive category. This reveals that students sometimes face challenges in evaluating sources, synthesizing information, and identifying research gaps. This pattern aligns with the findings of Mulyaningsih et al. (2022), who noted that while students can organize ideas effectively, they often require further support in conducting critical literature analysis. Table 10. Summary of Research Writing Competence among Senior High School Learners in Cluster 2 Public Secondary Schools in Davao City No. Statements Mean Descriptive Rating 1 Comprehensive Literature Review 3.29 Moderately Extensive 2 Clear Thesis Formulation 3.36 Moderately Extensive 3 Logical Organization of Content 3.52 Extensive 4 Accurate Citation and Referencing 3.32 Moderately Extensive Overall Mean 3.37 Moderately Extensive 3.3. Prompt Engineering among Senior High School Learners in Cluster 2 Public Secondary Schools in Davao City —On Table 11, the mean for prompt engineering among senior high school learners in Cluster 2 Public Secondary Schools in Davao City is 3.27, interpreted as moderately extensive. This suggests that students sometimes demonstrate the ability to construct and refine prompts when interacting with AI tools. However, their skills in prompt engineering were not yet consistently applied across writing tasks. The moderately extensive rating indicates an emerging understanding of how prompt formulation influences the quality and relevance of AI-generated responses. As highlighted by Park and Choo (2024), prompt engineering was a foundational skill in maximizing the pedagogical potential of AI, particularly in enhancing content specificity and task alignment. In support, Mzwri and Turcs ´ anyiSzabo (2025) emphasized that training students to craft and evaluate prompts fosters deeper engagement, critical thinking, and iterative learning. The individual mean scores for this domain range from 3.11 to 3.64, falling mostly within the moderately extensive category, with two indicators rated as extensive. The highestrated statement was Experimenting with different prompt structures boosts creativity in idea generation, with a mean of 3.64. On the other hand, the lowest-rated statement was Tailoring prompt inputs leads to more accurate information retrieval, with a mean of 3.11. This observation aligns with the findings of Heston and Khun (2023), who noted that while students enjoy experimenting with AI prompts, they may lack the technical knowledge or strategic approach to consistently optimize prompt effectiveness. 23 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - Table 11. Summary of Prompt Engineering among Senior High School Learners in Cluster 2 Public Secondary Schools in Davao City No. Statements Mean Descriptive Rating 1 Crafting clear prompts improves the relevance of AI-generated responses. 3.12 Moderately Extensive 2 Tailoring prompt inputs leads to more accurate information retrieval. 3.11 Moderately Extensive 3 Revising prompt wording enhances clarity in communication with the AI tool. 3.16 Moderately Extensive 4 Adjusting prompts based on feedback increases the effectiveness of the writing process. 3.40 Extensive 5 Experimenting with different prompt structures boosts creativity in idea generation. 3.64 Extensive 6 Testing various prompts helps identify the most efficient query methods. 3.23 Moderately Extensive 7 Customizing prompts refines research questions for better results. 3.27 Moderately Extensive 8 Monitoring prompt performance supports continuous improvement in writing tasks. 3.20 Moderately Extensive Mean 3.27 Moderately Extensive 3.4. Relationship Among Artificial Intelligence (AI) Writing Assistant Usage, Research Writing Competence, and Prompt Engineering of Learners —The results of the analysis on the relationship among Artificial Intelligence (AI) writing assistant usage, research writing competence of students, and prompt engineering were presented. Bivariate correlation analysis using Pearson product-moment correlation was utilized to determine the relationship among the variables mentioned.The findings reveal a strong and significant positive relationship between the use of Artificial Intelligence (AI) Writing Assistants and Research Writing Competence among senior high school learners (r = 0.816, p = 0.000), leading to the rejection of the null hypothesis. This suggests that students who frequently engage with AI writing tools tend to demonstrate higher levels of competence in key aspects of research writing, such as thesis formulation, organization, and citation accuracy. Daulay et al. (2024) noted that AI tools enhance student writing fluency and clarity by providing real-time feedback and structural suggestions. Similarly, Zhao (2024) emphasized that responsible and guided use of AI in academic settings promotes improved organization and accuracy in student research outputs. There was also a strong and significant positive relationship between AI Writing Assistant Usage and Prompt Engineering (r = 0.716, p = 0.000), leading to the rejection of the null hypothesis. This indicates that learners who actively used AI tools were more likely to develop the ability to craft, modify, and evaluate prompts effectively, skills essential for maximizing the value of AI-generated outputs. Lee et al. (2024) asserted that familiarity with AI systems strengthens students’ ability to tailor prompts for more accurate and creative outputs. Likewise, Haugsbaken and Hagelia (2024) found that iterative prompt construction cultivates strategic thinking and enhances learner autonomy in digital environments. Lastly, the results show a strong and significant positive relationship between research writing competence and prompt engineering (r = 0.693, p = 0.000), prompting the rejection of the null hypothesis. This indicates that students who demonstrate stronger research writing skills are also more capable of formu24 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - lating and refining prompts for AI tools. Knoth et al. (2024) explained that students with solid research backgrounds are more likely to adapt and optimize their prompts to achieve targeted writing goals. In support, Zaghir et al. (2024) emphasized that writing competence facilitates more analytical use of AI tools, enabling students to generate contextually appropriate and coherent outputs. Table 12. Relationship Among Artificial Intelligence (AI) Writing Assistant Usage, Research Writing Competence, and Prompt Engineering of Learners Variables Research Writing Competence Prompt Engineering Artificial Intelligence (AI) Writing Assistant Usage 0.816** 0.716** 0.000 0.000 Research Writing Competence 1 0.693** 0.000 Note. *Significant at p<0.05. Legend: Perfect Correlation for r=1.00; Strong Correlation for 0.7≤r<1.00; Moderate Correlation for 0.3≤r<0.7; Weak Correlation for 0.3>r>0.00; No Correlation for r=0.00. 3.5. Mediating Effect of Prompt Engineering on the Relationship Between Artificial Intelligence (AI) Writing Assistant Usage and Research Writing Competence of Learners —The results of the analysis on the mediating effect of prompt engineering on the relationship between Artificial Intelligence (AI) writing assistant usage and research writing competence of learners in Cluster 2 Public Secondary Schools in Davao City are shown in Table 13. The path coefficients table reveals a significant and positive direct relationship between the usage of Artificial Intelligence (AI) writing assistants and research writing competence among senior high school learners, with an estimate of 0.784, a z-value of 14.104, and p < .001. This indicates that greater use of AI tools directly contributes to students’ research writing competence, particularly in areas such as thesis formulation, content organization, and proper citation. In addition, prompt engineering significantly predicts research writing competence (estimate = 0.229, z = 4.763, p < .001), while AI usage strongly predicts prompt engineering (estimate = 0.828, z = 17.581, p < .001). The presence of this mediating variable reveals the layered influence of AI tools; not only do they directly support writing development, but they also indirectly enhance student outcomes through improved prompt engineering strategies. The parameter estimates further support a partial mediating effect. The indirect effect of AI usage on research writing competence via prompt engineering was significantly positive (estimate = 0.190, z = 4.597, p < .001), while the direct effect remains strongly significant at 0.784 (z = 14.104, p < .001). The total effect stands at 0.974 (z = 24.190, p < .001), indicating that while prompt engineering enhances the influence of AI tools on writing competence, the AI tools themselves remain the primary driver of improvement. The ratio index of 0.1950 and the Sobel z-test value of 4.6049 (p = 0.0000026) provide strong statistical confirmation of a partial mediation effect. The favorable ratio indicates that prompt engineering enhances rather than suppresses the direct relationship between AI usage and research competence of learners. Practically, this implies that as students refine their ability to craft effective prompts, they can better direct 25 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 ¨ ¨ NIJSE (2025) - Ismiati, I., & Pebriantika, E. (2020). Designing strategies for university students’ writing skill. Journal of Languages and Language Teaching,8(1), 8–19. Kale, A., Yaduka, T . , Shaikh, T . , Kaur, G., Pinjarkar, L., Agrawal, P . , & Khanuja, H. K. (2024). Unveiling the power of ai prompt engineering: A comprehensive exploration. 2024 10th International Conference on Electrical Energy Systems (ICEES), 1–9. Kawano, K. (2023). Engaging a kanaka ¯ oiwi literature review methodology through research on native hawaiian culture-based education. Review of Educational Research,93(6), 862– 900. Keller, S. D., Fleckenstein, J., Kruger, M., Koller, O., & Rupp, A. A. (2020). English writing skills of students in upper secondary education: Results from an empirical study in switzerland and germany. Journal of Second Language Writing,48, 100700. Kemalsyah, M., Solehuddin, M., Hariyadi, A., Jenuri, J., & Suwarma, D. M. (2022). An analysis of factors that affect efl students’ writing skill. QALAMUNA: Jurnal Pendidikan, Sosial, dan Agama,14(2), 869–876. Kerr, K. (2020). Ethical considerations when using artificial intelligence-based assistive technologies in education. In Ethical use of technology in digital learning environments: Graduate student perspectives. https://pressbooks.openeducationalberta.ca/educationaltechnologyethics/ chapter/ethicalconsiderationswhenusingartificialintelligencebasedassistivetechnologies-in-education/ Khalifa, M., & Albadawy, M. (2024). Using artificial intelligence in academic writing and research: An essential productivity tool. Computer Methods and Programs in Biomedicine Update, 100145. Kim, N. J., & Kim, M. K. (2022). Teacher’s perceptions of using an artificial intelligence-based educational tool for scientific writing. Frontiers in Education,7, 755914. Kim, Y . J., Oh, Y . , Park, S., Cho, S., & Park, H. (2013). Stratified sampling design based on data mining. Healthcare Informatics Research,19(3), 186–195. Knoth, N., Tolzin, A., Janson, A., & Leimeister, J. M. (2024). Ai literacy and its implications for prompt engineering strategies. Computers and Education: Artificial Intelligence,6, 100225. Lee, A. V . Y . , Teo, C. L., & Tan, S. C. (2024). Prompt engineering for knowledge creation: Using chain-of-thought to support students’ improvable ideas. AI,5(3), 1446–1461. Lee, U., Jung, H., Jeon, Y . , Sohn, Y . , Hwang, W . , Moon, J., & Kim, H. (2024). Few-shot is enough: Exploring chatgpt prompt engineering method for automatic question generation in english education. Education and Information Technologies,29(9), 11483–11515. Li, F . , & Han, Y . (2022). Student feedback literacy in l2 disciplinary writing: Insights from international graduate students at a uk university. Assessment Evaluation in Higher Education,47(2), 198–212. Li, J., Zong, H., W u , E., W u , R., Peng, Z., Zhao, J., & Shen, B. (2024). Exploring the potential of artificial intelligence to enhance the writing of english academic papers by non-native english-speaking medical students—the educational application of chatgpt. BMC Medical Education,24(1), 736. Lin, Z. (2023). Supercharging academic writing with generative ai: Framework, techniques, and caveats. https://arxiv.org/abs/2310.17143 32 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 ´ ´ NIJSE (2025) - MacKinnon, D. P . , Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. Annual Review of Psychology,58(1), 593–614. Majid, S., Foo, S., & Chang, Y . K. (2020). Appraising information literacy skills of students in singapore. Aslib Journal of Information Management,72(3), 379–394. Mayasa, B. A. (2022). Influencing factors in the academic writing of online learners. Psychology and Education: A Multidisciplinary Journal,5(10), 754–772. Meza, P . , & Gonzalez, M. (2020). Construction and validation of the self-efficacy scale for disciplinary academic writing. Cogent Education,7(1), 1830464. https://www.tandfonline. com/doi/full/10.1080/2331186X.2020.1830464 Montaner, S. (2020). Efl written competence through twitter in mobile version in compulsory secondary education. Global Journal of Foreign Language Teaching,10(2), 101–110. Mulyaningsih, I., Rahmat, W . , Maknun, D., & Firdaus, W . (2022). How competence of production, attention, retention, motivation, and innovation can improve students’ scientific writing skills. International Journal of Language Education,6(4), 368–385. https://eric.ed.gov/ ?id=EJ1375313 Mzwri, K., & Turcsanyi-Szabo, M. (2025). The impact of prompt engineering and a generative ai-driven tool on autonomous learning: A case study. Education Sciences,15(2), 199. Oestar, J., & Marzo, C. (2022). Teachers as researchers: Skills and challenges in action research making. International Journal of Theory and Application in Elementary and Secondary School Education,4(2), 95–104. Park, J., & Choo, S. (2024). Generative ai prompt engineering for educators: Practical strategies. Journal of Special Education Technology. https://doi.org/10.1177/01626434241298954 Pentury, H. J., Anggraeni, A. D., & Pratama, D. (2020). Improving students’ 21st century skills through creative writing as a creative media. Deiksis,12(2), 164–178. https://eric.ed.gov/ ?id=EJ1290959 Percy, M. (2020). Facilitating the acquisition of critical writing skills? an exploration of pedagogical practices at a new university in the uk. Journal of Contemporary Issues in Education, 15(2), 104–120. https://westminsterresearch.westminster.ac.uk/item/vv670/facilitatingthe-acquisition-of-critical-writing-skills-an-exploration-of-pedagogical-practices-at-anew-university-in-the-uk Perkins, M. (2023). Academic integrity considerations of ai large language models in the postpandemic era: Chatgpt and beyond. Journal of University Teaching and Learning Practice, 20(2). https://open-publishing.org/journals/index.php/jutlp/article/view/635 Prosekov, A. Y . , Morozova, I. S., & Filatova, E. V . (2020). A case study of developing research competency in university students. European Journal of Contemporary Education,9(3), 592–602. https://eric.ed.gov/?id=EJ1272421 Rababah, L. M. (2022). Contextualization to enhance students’ writing ability. Theory and Practice in Language Studies,12(11), 2316–2321. Rad, H. S., Alipour, R., & Jafarpour, A. (2024). Using artificial intelligence to foster students’ writing feedback literacy, engagement, and outcome: A case of wordtune application. Interactive Learning Environments,32(9), 5020–5040. Rojas, F . A., Villar-Mayuntupa, G., & Roman-Gonzalez, A. (2020). The formative research in the development of reading and writing skills and their impact on the development of indexed 33 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - publications by engineering students. 2020 IEEE World Conference on Engineering Education (EDUNINE), 1–4. https://ieeexplore.ie Sajja, R., Sermet, Y., Cikmaz, M., Cwiertny, D., & Demir, I. (2024). Artificial intelligenceenabled intelligent assistant for personalized and adaptive learning in higher education. Information,15(10), 596. Sari, Y. I., Utomo, D. H., & Astina, I. K. (2021). The effect of problem-based learning on problem solving and scientific writing skills. International Journal of Instruction,14(2), 11–26. https://journal.lppmunindra.ac.id/index.php/Deiksis/article/view/5184 Shokarimova, K. A. (2021). The way of improving reading and writing skills during the lessons. Academic Research in Educational Sciences,2(2), 1032–1040. https://cyberleninka.ru/ article/n/the-way-of-improving-reading-and-writing-skills-during-the-lessons Siedlecki, S. L. (2020). Understanding descriptive research designs and methods. Clinical Nurse Specialist,34(1), 8–12. Sikha, V. K., Siramgari, D., & Korada, L. (2023). Mastering prompt engineering: Optimizing interaction with generative ai agents. Journal of Engineering and Applied Sciences Technology,5(6), 2–8. https://doi.org/10.47363/Jeast/2023(5)E117 Singh, N., Wang, L. L., & Bragg, J. (2024). Figura11y: Ai assistance for writing scientific alt text. Proceedings of the 29th International Conference on Intelligent User Interfaces, 886–906. https://dl.acm.org/doi/abs/10.1145/3640543.3645212 Singh, S., Kumar, R., Maharshi, V., Singh, P. K., Kumari, V., Tiwari, M., & Harsha, D. (2024). Harnessing artificial intelligence for advancing medical manuscript composition: Applications and ethical considerations. Cureus,16(10), e71744. https://pmc.ncbi.nlm.nih.gov/ articles/PMC11569240/ Song, C., & Song, Y. (2023). Enhancing academic writing skills and motivation: Assessing the efficacy of chatgpt in ai-assisted language learning for efl students. Frontiers in Psychology,14, 1260843. https://www.frontiersin.org/journals/psychology/articles/10. 3389/fpsyg.2023.1260843/full Sugumlu, U. (2020). An action research on the improvement of writing skill in teacher training. Educational Policy Analysis and Strategic Research,15(1), 137–162. https://eric.ed.gov/ ?id=EJ1249636 Sukamolson, S. (2007). Fundamentals of quantitative research. Sullivan, S. (2024). Correlational designs. In Research methods in special education (pp. 113– 130). Routledge. Suwartini, Y., Lustyantie, N., Suseno, M., & Supriyati, Y. (2021). Writing ability with metaphorming approach: A systematic review. International Journal of Education,14(2), 149–158. Valizadeh, M. (2020). The flipped pedagogy: Effects on the grammatical competence and writing skill of basic users of english. Journal of Language and Linguistic Studies,16(3), 1100– 1113. Wahyuni, K. A. S., Mantra, I. B. N., & Handayani, N. D. (2020). The use of picture series to improve writing skill of efl learners. International Journal of Applied Science and Sustainable Development,2(2). Wale, B. D., & Bogale, Y. N. (2021). Using inquiry-based writing instruction to develop students’ academic writing skills. Asian-Pacific Journal of Second and Foreign Language Education, 6, 1–16. https://link.springer.com/article/10.1186/s40862-020-00108-9 34 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025 NIJSE (2025) - Wang, L., & Fan, J. (2020). Assessing business english writing: The development and validation of a proficiency scale. Assessing Writing,46, 100490. Warlizasusi, J., Ifnaldi, I., Nuzuar, N., Qodri, A., & Utami, H. S. (2023). Developing academic writing learning module to improve writing skills of graduate program students. Ta’dib, 26(2), 391–404. https://ejournal.uinmybatusangkar.ac.id/ojs/index.php/takdib/article/ view/10496 Wong, C. P., & Ng, D. (2020). The roles of school leaders in developing future-ready learners: The case of singapore. International Journal of Educational Management,35(1), 249–269. Woo, D. J., Wang, D., Yung, T., & Guo, K. (2024). Effects of a prompt engineering intervention on undergraduate students’ ai self-efficacy, ai knowledge and prompt engineering ability: A mixed methods study. https://arxiv.org/abs/2408.07302 Yeh, C., Ramos, G., Ng, R., Huntington, A., & Banks, R. (2024). Ghostwriter: Augmenting collaborative human-ai writing experiences through personalization and agency. https: //arxiv.org/abs/2402.08855 Zaghir, J., Naguib, M., Bjelogrlic, M., N ´ ev ´ eol, A., Tannier, X., & Lovis, C. (2024). Prompt engineering paradigms for medical applications: Scoping review. Journal of Medical Internet Research,26, e60501. Zhang, M. (2024). Enhancing self-regulation and learner engagement in l2 speaking: Exploring the potential of intelligent personal assistants within a learning-oriented feedback framework. BMC Psychology,12(1), 421. Zhang, Z., Gao, J., Dhaliwal, R. S., & Li, T. J. J. (2023). Visar: A human-ai argumentative writing assistant with visual programming and rapid draft prototyping. Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology, 1–30. Zhao, D. (2024). The impact of ai-enhanced natural language processing tools on writing proficiency: An analysis of language precision, content summarization, and creative writing facilitation. Education and Information Technologies, 1–32. Zhao, X. (2023). Leveraging artificial intelligence (ai) technology for english writing: Introducing wordtune as a digital writing assistant for efl writers. RELC Journal,54(3), 890–894. https://journals.sagepub.com/doi/full/10.1177/00336882221094089 Zorec, K. (2024). “we’re all for the same mission”: Faculty mentoring native hawaiian undergraduates in stem research. Journal of Diversity in Higher Education,17(1), 40. 35 ISSN 3028-1261 10.5281/zenodo.17354831/NIJSE.2025