Finаl Exаm: Stоrybоаrd fоr Improving Student Retention at Westbrook State University The Office of Institutional Research and Student Success Analytics at Westbrook State University, a large public university serving more than 25,000 undergraduate students, has completed an eight-month project to develop recommendations for improving first-year and second-year retention among its first-time freshmen population. The project integrated four previously siloed data sources — transcript records, Pell Grant eligibility files, pre-enrollment academic preparation data, and demographic records — covering 23,379 first-time freshmen across four entry cohorts. Findings were validated through interviews with academic advisors and college deans across eight academic colleges. The team is now ready to present its findings and recommendations to senior leadership, including the Provost, who holds final authority on which retention interventions to fund. The most significant findings include the following: The average number of credit units earned by the end of the second academic year was 45.83 units, with a standard deviation of 19.55, indicating substantial variation in students' academic momentum. Pell-eligible students earned statistically significantly fewer credits in their first two years than non-Pell-eligible peers (p < 0.001), and minority students earned an average of 45.04 units compared to 49.47 units for non-minority students (p < 0.001). First-generation college students accumulated fewer credits than peers whose parents had attended at least some college, with the gap visible across all four entry cohorts. Credit accumulation varied substantially by college: Engineering students averaged 49.35 units by the end of the second year, while University Programs (undeclared majors) averaged only 39.08 units — a 10-unit gap equivalent to roughly one full semester. Clustering analysis identified three distinct academic engagement profiles: (1) high-GPA students with full course loads, (2) students attempting moderate-to-heavy course loads but earning GPAs below 2.5, and (3) students with consistently low unit loads regardless of GPA. A predictive deep learning model trained on freshman-year data was able to forecast sophomore-year credit accumulation with strong accuracy, demonstrating that at-risk students can be identified at the end of the first academic year — well before traditional warning indicators trigger advising outreach. Students in low-credit-accumulation profiles were disproportionately Pell-eligible, minority, or first-generation, meaning the three demographic risk factors tend to compound within the same students rather than operating independently. Based on these findings, the team would like to make the following recommendations: Implement an early-warning predictive analytics system that uses end-of-first-year data to flag students at elevated risk of falling behind, routing flagged students to proactive advising outreach. Launch a targeted "15 to Finish" credit momentum initiative for Pell-eligible, minority, and first-generation students, combining guaranteed course availability, block scheduling, and financial-aid-aware advising. Establish differentiated advising tracks aligned with the three engagement profiles identified through clustering: enrichment opportunities for high-performing students, academic support for struggling moderate-load students, and barrier-removal advising for consistently low-load students. Based on this information, assume that you have created a storyboard outlining the structure of a presentation tailored to the aforementioned audience. This storyboard is presumed to satisfy the following criteria: The protagonist is clearly defined—including sufficient details such as name, background, and academic status—enabling the audience to forge an emotional connection with a specific student rather than merely viewing aggregated statistics. It adheres to a five-act narrative structure (exposition, rising action, climax, falling action, and resolution). Aristotle’s Rhetorical Triangle has been applied—explicitly indicating where ethos, pathos, and logos are introduced within the presentation. The traceability of logos has been demonstrated—each recommendation is mapped directly to the specific findings that substantiate it. Q1. If the presenters removed the student protagonist entirely and delivered the same presentation using only statistical data (e.g., the 45.04 vs. 49.47 unit gap, the p-values, the predictive model accuracy), among ethos, pathos, and logos, which appeal in the Triangle would suffer the greatest loss? [Q1] Q2. What would happen if the presenters featured the student protagonist not just in the introduction and conclusion, but on every slide — including the slides showing the predictive model results and the clustering analysis? Among ethos, pathos, and logos, which appeal in the Triangle would suffer the greatest loss? [Q2] Q3. In the conclusion, the presenters explicitly acknowledge that the predictive model is based on data from four entry cohorts only and that the model's accuracy may vary for transfer students or part-time students. Intuitively, this seems to expose weaknesses in the analysis but it actually increases the credibility rather than diminish it. Among Ethos, Pathos, and Logos, which appeal within the 'Triangle' would stand to benefit the most? [Q3] Q4. Imagine the audience is not the Provost and senior administrators, but the university's faculty senate — the body of professors that ultimately must support the proposed advising and curriculum changes. How should the balance among the three appeals shift? Which appeal becomes more important, and which becomes less important? Among Ethos, Pathos, and Logos, [Q4] becomes heavier, [Q5] becomes lighter