Source-based writing, a common academic task, requires students to synthesize information from multiple texts. While its benefits are acknowledged, the impact of task characteristics like the number of sources on student performance and self-efficacy remains understudied. This study addresses this gap by comparing the writing performance and self-efficacy of students assigned to either a single-source or a multiple-source writing task. The knowledge-telling/knowledge-transforming model and self-efficacy theory provide the theoretical frameworks for understanding how the number of sources might influence students' cognitive processes and confidence levels. Existing research on source-based writing has examined various task variables such as prompt type, topic familiarity, and genre, but the specific effect of the number of sources on both the quality of the written product and students' self-efficacy remains unclear. This study aims to address this gap by focusing on broader aspects of writing performance, including selecting, integrating, and organizing information, and by directly measuring the impact on students' self-efficacy beliefs. The results will inform pedagogical approaches to source-based writing instruction, helping educators design more effective interventions to support student learning.

This literature review examines the theoretical foundations and previous research related to source-based writing. The knowledge-telling/knowledge-transforming model highlights the difference between shallow information retrieval (knowledge-telling) and deeper engagement with the task involving problem-solving (knowledge-transforming). Self-efficacy theory, focusing on individual beliefs about their ability to perform a task, is another important framework. Previous studies have explored the impact of single versus multiple sources on writing performance, with some suggesting that multiple sources enhance information transformation and understanding. However, these studies often focused on surface linguistic features rather than deeper aspects of writing such as selecting, integrating and organizing information. Additionally, the relationship between the number of sources and self-efficacy in writing has received limited attention. This study aims to address these gaps by investigating both the quality of writing and the impact on students' self-efficacy.

A between-subjects design was employed, randomly assigning 57 first-year university students to either a single-source (n=29) or multiple-source (n=28) condition. Both groups received the same instructions to write an opinion essay on whether the UAE distracted driving law should be changed. The single-source group received one comprehensive text, while the multiple-source group received four sources (a newspaper article, FAQ, graph, and excerpt from TED Talk comments) containing overlapping information. A pre-writing test and a self-efficacy survey (administered before and after the writing task) were used to assess baseline writing ability and self-efficacy beliefs. The writing samples were assessed using a rubric based on four criteria: content, source use, organization, and language use. The self-efficacy survey, adapted from previous studies, measured writing-related abilities, reading-related abilities, and source integration skills using a 5-point Likert scale. A principal component analysis (PCA) was conducted to reduce the 22 items to three subscales. The internal reliability of the survey was confirmed using Cronbach's alpha. Paired-sample t-tests were used to analyze changes in self-efficacy beliefs between pre- and post-writing assessments. Independent samples t-tests were used to compare the overall writing performance and specific aspects between the two groups. Inter-rater reliability for the writing rubric was established through scoring a subset of essays by two raters.

While there was no statistically significant difference in overall writing performance between the single-source and multiple-source groups (p=0.126), significant differences emerged in specific aspects. The multiple-source group significantly outperformed the single-source group in content (t=2.52; p < 0.02; d=0.67), demonstrating a greater reliance on the provided sources and producing more detailed, text-based essays. The multiple-source group also showed a tendency towards greater source use, although this difference was not statistically significant (p=0.538). No significant differences were found in organization or language use between the two groups (p=0.829 and p=0.909, respectively). Regarding source integration attempts, the multiple-source group exhibited significantly more instances of exact copying, borrowed phrases, and paraphrasing than the single-source group. In terms of self-efficacy, the multiple-source group showed a statistically significant increase in reading-related abilities after the writing task (p=0.027), while the single-source group showed a significant decrease (p=0.027). Changes in self-efficacy related to writing abilities and source use skills were not statistically significant for either group.

The findings suggest that while the total number of sources does not guarantee superior overall writing quality within the limited timeframe of the study (45 minutes), providing multiple sources positively influenced content development and source engagement. The multiple-source group's increased reliance on provided sources contrasts with the single-source group's tendency to rely more on prior knowledge. The lack of significant differences in organization and language use suggests that these aspects may be less influenced by the number of sources and more reliant on prior instruction and skill development. The significant increase in reading-related self-efficacy for the multiple-source group, coupled with the decrease in the single-source group, highlights the role of active source engagement in bolstering confidence. The difficulties both groups experienced with paraphrasing and source integration emphasize the need for focused instruction in these skills. The 45-minute timeframe may not have allowed for significant differences in overall writing quality to fully manifest, warranting future research with longer timeframes.

This study demonstrates that while overall writing quality wasn't significantly different between single and multiple source groups in a short timeframe, multiple sources positively affected content, source integration attempts, and reading-related self-efficacy. Students struggled with source integration regardless of the number of sources. Future research should investigate the long-term impact of multiple sources on writing, include a more diverse sample, and explore the writing processes in more detail. Pedagogically, this study emphasizes the importance of pre-writing activities to enhance source comprehension and engagement, explicit instruction in source integration techniques, and the potential benefits of multiple sources in promoting confidence in reading-related skills.

The study's limitations include a non-representative sample (primarily male), a limited timeframe for the writing task (45 minutes), and a single writing task per condition. The focus on written products, rather than writing processes, also limits the depth of analysis. Future studies should address these limitations to enhance the generalizability of the findings.