Publications & Academic Work

MA in Statistics

University of California, Los Angeles (UCLA)

BA in Mathematics/Economics, Minor in Statistics

University of California, Los Angeles (UCLA)

Research Interests

I study student outcomes in community college mathematics and statistics. Two threads run through the work: what happened to course success and equity gaps when California placed students directly into transfer-level statistics under AB 705 and AB 1705, and how AI-assisted course and assessment design changes the cognitive demand of what we ask students and the distribution of their results. I also apply the same quantitative methods to questions in public health and social policy. Every study is reported with its limitations stated plainly.

Keywords: community college mathematics, placement reform, educational equity, AI-assisted course design, assessment, quantitative methods

Rising Success, Persistent Gaps: Student-Level Statistics Outcomes Across California's Community College Placement-Reform Era (2018 to 2025)

Author: Safaa Dabagh

Journal: Community College Review (SAGE)

Submitted: June 2026

Status: Under Review

A student-level analysis of 867 transfer-level introductory statistics enrollments across 17 semesters (Fall 2018 to Fall 2025) at a large California community college, spanning the AB 705 and AB 1705 placement reforms. Course success rose steadily across the reform era (42% to 69%) and for every racial and ethnic group, which contradicts the premise that direct placement depresses outcomes, while a persistent gap of about 24 percentage points between White and Latinx students shows that broadened access alone did not equalize outcomes. The temporal trend is reported as correlational, with semester-clustered standard errors, composition-adjusted robustness checks, and confidence intervals on all subgroup estimates.

AI-Assisted Assessment Design in Mathematics and Statistics: A Multi-Course Bloom's Taxonomy Analysis of Cognitive Demand Shift

Author: Safaa Dabagh

Journal: Teaching Statistics

Status: Revise & Resubmit

A Bloom's-taxonomy analysis of how AI-assisted assessment design shifts the cognitive demand of mathematics and statistics course materials. Assessment items across multiple courses are coded by cognitive level to characterize where AI-supported design raises or lowers the demand placed on students, informing how instructors can use AI tools without eroding the rigor of their assessments.

Raising the Floor, Holding the Bar: Distributional Effects of AI-Assisted Course Design in Two Community College Mathematics Courses

Author: Safaa Dabagh

Repository: EdArXiv / OSF preprint

DOI: 10.35542/osf.io/x9nkw_v1

Status: Preprint Available

A distributional analysis of AI-assisted course design across two community college mathematics courses (College Algebra for STEM and Introductory Statistics). Rather than focusing on average outcomes, the study asks who is affected and where in the grade distribution. The strongest effect is a floor effect in algebra, where the lowest-performing students gained the most, alongside evidence that the validity of an assessment depends on how well it is aligned with the AI-designed instruction.

Economic Disadvantage and Mass Shootings Across U.S. States, 2018 to 2024: A State-Level Panel Analysis

Author: Safaa Dabagh

Repository: Research Square preprint (updated October 2026)

Status: Preprint Available

Mass shootings in the United States roughly doubled after 2019 and stayed high. Using every Gun Violence Archive mass shooting in the 50 states from 2018 to 2024 (3,783 incidents) and state data drawn directly from official sources, this study finds that where poverty and economic disadvantage concentrate, mass shootings concentrate too: each percentage point of poverty is associated with about 15% more mass shootings per million residents, and the 10 highest-poverty states have 1.7 times the rate of the 10 lowest. Poverty cannot be separated from low income or segregation at the state level, so the finding is about economic disadvantage as a whole. State data cannot settle whether gun laws work, and state mental-illness rates are not a guide to where shootings happen. These are associations, not proof of cause. Updated October 2026: the state data were rebuilt from official files and the conclusions revised; one script reproduces every number.

Instructor of Mathematics with Main Focus on Statistics

Santa Monica College (SMC) • 2016-Present

• Math C1000: Introduction to Statistics (formerly Math 54), including sections paired with the Math 54C co-requisite support course
• Math 4/4C: College Algebra for STEM Majors with co-requisite support, and other mathematics classes

West Los Angeles College • Current

• Math C1000: Introduction to Statistics, POPP program section
• Other mathematics and statistics courses

Loyola Marymount University • Current

• Math 205, Applied Statistics in R: Two sections. A first course in statistical reasoning using R, designed for life sciences students

Joyce University of Nursing and Health Sciences • Current

• Faculty-Adjunct: Statistics and data science courses for nursing and health sciences graduate programs

Teaching Approach:

• Hybrid and online course delivery
• Canvas LMS integration and course design
• Focus on educational equity and student success for diverse learners
• Experience across multiple institutional contexts (community colleges, a private university, a nursing and health sciences university)

AI in Education

• Google AI for Educators Completed
• Applied practice: AI-assisted course and assessment design across my statistics and algebra sections (studied in the papers above), the Stats Without Walls tool, and a Canvas AI grading assistant in development for the SMC mathematics department

Programming & Statistical Analysis

• R: Advanced statistical analysis, data visualization (ggplot2), modeling
• Python: Data analysis (pandas, numpy), machine learning (scikit-learn), AI integration
• Statistical Methods: Regression and generalized linear models, panel and count models, quantile and distributional analysis, sensitivity analysis for unobserved confounding, survey analysis

Educational Technology

• LMS: Canvas, course design, online pedagogy
• Web Development: HTML, CSS, JavaScript (Learn Without Walls and Stats Without Walls are built by hand, no frameworks)
• AI Integration: Claude API, Chrome extensions, prompt design for teaching and grading
• Data Platforms: Google Colab, Jupyter notebooks, Observable

• Learn Without Walls: Free open courses, four open textbooks, and the Stats Without Walls tool, accessible to all students regardless of financial means
• Community College Advocacy: Conducting research that centers community college student experiences and needs
• Educational Equity: Focus on understanding and addressing achievement gaps in STEM education