Python analysis and visualization scripts for a longitudinal study of undergraduate student success in CS/IT at Kean University. This codebase produced the figures, statistics, and regression analysis for a paper currently in progress.
Project: NSF Building Capacity: Cougar Pathway to Success
NSF Awards: #1928452, #2129795, #2345334
Author: Dahana Moz Ruiz
Institution: Kean University, CS/IT Department
This project analyzes 6 years (2019–2025) of student engagement and outcome data across 1,220 undergraduate CS/IT students to answer these questions:
RQ1 — How did the Cougar Pathway to Success project evolve across iterative intervention cycles during the grant period?
RQ2 — How does student engagement in pathway activities relate to participation in research, internships, and career outcomes?
RQ3 — To what extent is pathway engagement associated with academic success and professional achievement among computing students?
RQ4 — Which pathway interventions are most strongly associated with increased engagement, inclusion, and indicators of retention?
Key finding: Program participation explains 53% of career outcome variance (R² = 0.53, p < .001) compared to only 9% for GPA.
| File | RQ | Description |
|---|---|---|
charts.py |
RQ2, RQ3 | Main visualization pipeline — generates all publication figures including outcome rates by engagement tier, GPA trends, equity analysis, and SI/NSO participation |
activityimpacts.py |
RQ4 | Activity-level impact analysis — ranks RE and EX activities by number of students achieving professional offers, internships, and research outcomes. Generates bar charts and bubble map |
sweetspot.py |
RQ2, RQ3 | Engagement threshold analysis — identifies the "sweet spot" activity count where outcome rates jump significantly. Profiles students who achieved all 3 outcomes |
val.py |
RQ1, RQ2 | Validation and descriptive statistics — computes engagement tiers, demographic breakdowns, outcome counts, GPA summaries, and SI/NSO totals |
demo_race.py |
RQ4 | Demographic data audit — cross-references two student profile sheets to identify and flag missing gender, race, ethnicity, and graduation data |
regression.py |
RQ2, RQ3 | Full statistical analysis — OLS linear regression, logistic regression with odds ratios and 95% CIs, chi-square tests with Cramér's V, and Cohen's d effect sizes across engagement tiers and outcomes |
Students are segmented into 6 tiers based on total RE + EX activity count:
| Tier | Activities | N |
|---|---|---|
| Not Engaged | 0 | 452 |
| Engaged 1 | 1 | 347 |
| Engaged 2 | 2 | 167 |
| Engaged 3 | 3 | 79 |
| Medium | 4–6 | 86 |
| Highly Engaged | 7+ | 89 |
- OLS Linear Regression — Total activities → composite outcome score (R² = 0.53, F(1,671) = 763.05, p < .001)
- OLS Linear Regression — GPA → composite outcome score (R² = 0.09)
- Logistic Regression — Activities + GPA + demographics → each outcome (odds ratios with 95% confidence intervals)
- Chi-Square Tests — Engagement tier × outcome with Cramér's V effect sizes
- Cohen's d — GPA differences between engagement tiers and between outcome achievers vs non-achievers
- Professional Offer — employment or graduate school acceptance
- Internship — paid or unpaid internship placement
- Research Outcome — publication, presentation, or research award
- Graduation — degree completion
- Hispanic/Latino representation rises from 33.3% (Not Engaged) to 46.1% (Highly Engaged), exceeding the 39% institutional baseline
- Female representation rises from 15.5% (Not Engaged) to 34.8% (Highly Engaged), exceeding the 21% CS/IT departmental baseline
Running charts.py produces publication-ready figures including:
- Outcome rates (Professional Offer, Internship, Research Outcome, Graduation) by engagement tier
- GPA trends across engagement tiers
- Hispanic/Latino and female representation by tier
- SI (Supplemental Instruction) participation over time
Running activityimpacts.py produces:
- Horizontal bar charts of top RE and EX activities ranked by outcomes
- Bubble map of activity impact (Professional Offers vs Research Outcomes)
- ROI scatter plot (outcomes per student vs attendance)
- Python — Core language
- Pandas — Data loading, merging, and aggregation
- NumPy — Statistical calculations
- Matplotlib — Publication-quality visualizations
- SciPy — Chi-square tests, t-tests
- Statsmodels — OLS and logistic regression
- OpenPyXL — Excel file reading
pip install pandas numpy matplotlib openpyxl scipy statsmodels
python charts.py
python activityimpacts.py
python sweetspot.py
python val.py
python regression.pyThe analysis uses a longitudinal dataset of 3,191 student activity records across 1,220 unique CS/IT students (2019–2025), merging:
- Institutional student profile data
- Departmental engagement and achievement records
- SI/NSO participation logs
⚠️ Dataset not included in this repository. The dataset contains protected student education records (FERPA) and is not available for public distribution.
"Pathways to Undergraduate Success in Computer Science: Impactful
Interventions to Improve Student Success"
Dahana Moz Ruiz, Daniel Ojeda, Luis Miguel Velazquez Rodriguez,
Ching-Yu Huang, Sarah Hug, Daehan Kwak, Patricia Morreale
In progress
