Pith. sign in

REVIEW 5 major objections 4 minor 21 references

Quantifying Holistic Review: A Multi-Modal Approach to College Admissions Prediction

T0 review · 5 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proposes CAPS, a modular score that breaks holistic admissions into academic, essay, and extracurricular components and claims these predict admission tiers with over 75% accuracy.

desk verdict A coherent proof-of-concept whose headline R² is likely circular; send to review with strong revision demands. read the letter →

arxiv 2507.15862 v2 pith:WEG3NWIO submitted 2025-07-12 cs.LG cs.CY

Jun-Wei Zeng , Jerry Shen This is my paper · ORCID
classification cs.LGcs.CY
keywords collegeadmissionspredictionholisticreviewinterpretablemachinelearningessayqualityassessmenttransformerembeddingsXGBoostregressionSHAPexplanationsmulti-modalscorefusion
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

College admissions decisions described as holistic are usually made behind closed doors, leaving applicants with little sense of where they stand. This paper argues that the evaluation can be decomposed into three measurable parts—academics, essay quality, and extracurricular impact—and then recombined into a single transparent score. The proposed Comprehensive Applicant Profile Score (CAPS) predicts admission tiers from those parts with over 75% accuracy and predicts essay quality with an R² of 0.80 on a synthetic dataset. If the approach transfers to real applicant pools, applicants could receive concrete, component-level feedback and institutions could audit their own criteria instead of relying on opaque heuristics. The paper is careful to call the results a proof of concept because the dataset is synthetic.

What carries the argument

The load-bearing machinery is the three-module decomposition plus a convex score fusion. SAS normalizes academic inputs with z-scores, derives weights from PCA and expert-defined priors, and maps them to a 0–100 scale. EQI encodes essays with all-MiniLM-L6-v2 embeddings and GPT-4o rubric scores, trains an XGBoost regressor on the concatenated 387-dimensional vector, then applies a sigmoid penalty for prompt misalignment. EIS blends GPT-4o impact scores with a five-tier activity ladder and a thematic coherence adjustment. The fusion module combines logistic-regression coefficients, XGBoost importances, and expert priors with tunable weights (α=0.3, β=0.3, γ=0.4), so the final score is an explicit weighted sum rather than a black box. SHAP explains each EQI prediction by feature contribution.

What would settle it

Give the pipeline a corpus of real application essays with independent human admissions-officer ratings, build the same 387-dimensional features, and test whether the XGBoost regressor still reaches an R² near 0.80 for EQI; then apply the full CAPS score to a university's historical admit/deny records and compare predicted tiers with actual decisions. A large drop in R² or near-chance tier accuracy would show the synthetic results were artifacts of the generation process.

Watch

Extended reading notes

Core claim

The central claim is that holistic review is quantifiable: each applicant can be represented by a Standardized Academic Score (SAS), an Essay Quality Index (EQI), and an Extracurricular Impact Score (EIS), and a weighted fusion of these three explains most of the variation in admission tiers. The EQI module is the technical core: GPT-4o scores each essay on content, language, and structure, MiniLM embeddings add 384 semantic dimensions, and an XGBoost regressor combines the 387 features to predict a continuous quality score, reaching an R² of 0.80 on held-out essays. A sigmoid penalty adjusts for prompt alignment, and SHAP attributions show that the three GPT rubric scores dominate the prediction. The final CAPS score is a normalized convex combination of module weights learned from logistic regression, XGBoost feature importance, and expert priors, with an optional diversity bonus capped at 12 points. The paper interprets these results as evidence that an interpretable, modular score can align with human judgment while giving applicants actionable feedback.

Load-bearing premise

The load-bearing premise is that the synthetic applicant scores and admission tiers faithfully resemble real holistic admissions decisions, because if they do not, the reported accuracy and R² measure the data-generation rules rather than the framework's ability to predict human judgments.

Editorial extensions

If this is right

  • If the central claim holds, a single dashboard could show an applicant their academic, essay, and extracurricular scores separately, turning vague admissions feedback into targeted advice.
  • Institutions could inspect the fused weights and SHAP contributions to see which factors drive their decisions, making criteria auditable.
  • The modular design means a university could retrain or re-weight modules on its own historical data instead of adopting a fixed national formula.
  • The reported R² of 0.80 for EQI would justify using embedding-plus-rubric features as a fast proxy for essay quality before human review.
  • The complete failure to classify the rare Tier 4 in the synthetic data shows that rare admission outcomes need explicit imbalance handling, such as resampling or class weights.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper: the real test of CAPS is transfer to non-synthetic data with independently decided admission outcomes; we expect the headline metrics to drop unless the real decisions are as linearly separable as the simulated tiers.
  • Beyond the paper: the EQI module could serve as a feedback instrument even if it never decides admissions, since SHAP-based rationales tell students which essay dimensions to improve.
  • Beyond the paper: because the EQI target and the strongest predictor features both come from the same GPT-4o rubric, the 0.80 R² partly measures the LLM's self-consistency; an independent human-rated essay corpus would separate self-consistency from true essay quality.
  • Beyond the paper: the fusion weights (0.3, 0.3, 0.4) are tuned on synthetic data; institutions adopting CAPS would need a principled way to set them from stated policy rather than from one simulated pool.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 4 minor

Summary. The paper proposes CAPS, a modular framework that combines academic, essay, and extracurricular components into a single interpretable admissions score. SAS is computed from normalized academic features with a blend of PCA-derived and expert weights; EQI is obtained by feeding GPT-4o rubric scores concatenated with MiniLM embeddings into an XGBoost regressor; EIS fuses GPT-4o activity scores with expert tiers and a coherence adjustment. The final CAPS is a convex combination of logistic-regression weights, XGBoost importance weights, and expert priors, optionally adjusted by a diversity bonus. Experiments on a synthetic dataset report an EQI prediction R² of 0.80, classification accuracy over 75%, macro F1 of 0.69, and weighted F1 of 0.74, with SHAP used for interpretability.

Significance. The problem addressed—transparency and interpretability in holistic college admissions—is important and timely, and the modular architecture with SHAP-based explanations is a plausible blueprint for applicant-facing feedback systems. The paper honestly concedes in Section V that the synthetic dataset cannot capture real-world complexity, and it makes a useful engineering contribution by combining LLM rubric scoring, semantic embeddings, and structured feature fusion. However, the quantitative claims are not established because the EQI target is circularly defined and the admission-tier labels are synthetic functions of the same features used for prediction. If the framework were re-evaluated on real admissions data with independently defined targets, the approach could be valuable; as presented, the reported metrics do not provide evidence of alignment with human holistic review.

major comments (5)
  1. [§III-C, §IV] The EQI regression target is never defined independently of the predictor features. In §III-C Step 3 and §IV, the XGBoost regressor is trained on a feature vector formed by concatenating three GPT-4o rubric scores (Content, Language, Structure) with MiniLM embeddings, and it is asked to predict a continuous EQI score in [0,1], yet the paper nowhere states how the EQI target values are computed. If EQI is simply a normalized combination of the same rubric scores, then the R²=0.80 in Eq. (18) measures how well the model recovers its own label-generating function, not agreement with human judgment. The statement that 'a subset of scores was manually verified' provides no counts, no inter-rater reliability, and no evidence that the target is human-aligned.
  2. [§IV] The admission-tier labels used for classification evaluation are synthetic and are generated from the same module scores that serve as features. Each applicant is represented by SAS, EQI, and EIS, and the tiers are drawn from these scores; logistic regression is then trained on those three scores to predict the tiers, and the fused CAPS weights in Eq. (14) are also fit to the same synthetic outcomes. Under this design, high accuracy may only confirm that the label-generating mechanism is recoverable from its inputs, which is not evidence that CAPS predicts real holistic admissions decisions.
  3. [§IV-A] The reported macro F1 of 0.69 is inconsistent with the per-tier results in the same subsection. If the three listed classes are 'Tier 0–2' with F1=1.00, Tier 3 with F1=0.44, and Tier 4 with F1=0.00, the macro average is (1.00+0.44+0.00)/3=0.48, not 0.69. If the macro F1 is computed over five separate tiers, the table must report all five per-class rows. Please provide the full confusion matrix and the exact averaging procedure.
  4. [§IV-A(b)] The XGBoost classifier evaluation reports only training-set performance: 'perfect accuracy on the training set (100%)' with macro and weighted F1 of 1.00. Without held-out test metrics, this result is uninformative about generalization and appears to be an overfitting artifact. The robustness claim based on this result is not supported.
  5. [§IV] The 'Human Validation' step is asserted but not documented: no number of essays reviewed, no agreement statistics, and no scoring protocol are given. Since the paper's central claim of human alignment rests on this validation, the absence of any quantitative human-judgment comparison is a load-bearing gap.
minor comments (4)
  1. [Fig. 2] The diagram title contains the typo 'Comphrehensive'; it should read 'Comprehensive'.
  2. [§IV-A] The multinomial logistic regression results paragraph is duplicated verbatim; one copy should be removed.
  3. [§III-D, §III-C] There are typos in the prompts: 'propmt' in the EIS step and 't he' in the alignment prompt; please proofread the prompt texts.
  4. [References] References [2] and [7] have nearly identical titles; please verify the citation metadata and confirm that the two entries are distinct works.

Circularity Check

1 steps flagged · score 7.0 of 10

The headline EQI R²=0.80 is not evidence of human-aligned essay evaluation: the EQI target is never defined independently of the GPT-4o rubric scores that are used as predictor features.

  1. fitted input called prediction [Section III-C, Step 3 and Section IV, Model Setup/Feature Construction, Eq. 18]
    "The GPT-4o scores (3 features) are concatenated with MiniLM embeddings (384 features), resulting in a 387-dimensional feature vector. This serves as input to an XGBoost regression model trained to predict continuous EQI scores in the range [0,1]. ... Each essay was represented by: A 384-dimensional sentence embedding vector extracted via MiniLM-L6-v2. The 3 rubric-based scores from GPT-4o. The final feature matrix contained 387 dimensions and was used to train an XGBoost regression model with the goal of predicting a continuous EQI score in the [0,1] range. ... MSE= 0.0316, R2 = 0.7999"

    The paper never defines an independent EQI target. The only essay-quality scores described are the three GPT-4o rubric scores (Content, Language, Structure), and these exact scores are concatenated with MiniLM embeddings to form the XGBoost feature vector. If the continuous EQI label is any function of those rubric scores (e.g., a normalized average), then the regressor's inputs contain the generators of its target, so the reported R2=0.7999 measures recovery of that deterministic mapping, not alignment with human judgment. The sentence about human validation only says a subset of rubric scores was manually verified; no independent EQI labels, counts, or inter-rater reliability are provided.

full rationale

The central quantitative claim is the EQI prediction R²=0.80, presented in the abstract and Section IV as evidence that CAPS reproduces human-like essay evaluation. However, the paper never specifies how the continuous EQI target is constructed. The only essay-quality signal described is the GPT-4o rubric scoring, and the features used by the XGBoost regressor include those same three rubric scores. If EQI is computed from those scores, the R² is a measure of how well the model recovers the label-generating function, not of agreement with human judgment. The paper's own limitations section concedes the synthetic dataset cannot capture real-world complexity, but it does not acknowledge this self-referential target issue. No significant self-citation chain or imported uniqueness theorem appears, so the circularity is concentrated in the EQI regression step. The tier-classification results depend on the synthetic dataset, but because the paper does not explicitly state the tier-generation rule, I do not count that as a separately demonstrated circular step. Overall, the headline result reduces, by construction or by missing independent definition, to a fit of the model to its own input-derived labels.

Assumptions & free parameters 8 free parameters · 6 assumptions · 1 invented entities

The central results depend on numerous hand-set constants and on the assumption that synthetic labels and LLM scores are valid proxies for human holistic review. No external benchmark, code, or dataset is provided, so the ledger is dominated by domain assumptions.

free parameters (8)
  • PCA component weights (alpha_PCA, beta_PCA) = 1.0, 0.5
    Set in Section III-B1 to balance dominant and secondary variance directions in Eq 4.
  • SAS fusion weight alpha_fusion = 0.1
    Eq 6 trades off PCA-derived and manual weights; the low value keeps expert-defined weights dominant.
  • Manual SAS feature weights = GPA 0.40, SAT 0.15, TOEFL 0.10, AP 0.10, Course Difficulty 0.25
    Table II lists expert-assigned weights used in the SAS score, with no empirical basis given.
  • EQI penalty parameters (lambda, k, x0) = Not fully specified, e.g., k=4, x0=0.3
    Eq 10 uses these to shape the sigmoid alignment penalty, but exact values for lambda and x0 are not stated.
  • EIS gamma = 0.5
    Eq 12 balances GPT score and tier score; chosen as the default by the authors.
  • EIS coherence adjustment constants = 0.85, 0.15
    Eq 13 adjusts average activity score by coherence score using hand-set constants.
  • CAPS weight fusion (alpha, beta, gamma) = 0.3, 0.3, 0.4
    Eq 14 fuses logistic, XGBoost, and expert weights; the values were set after preliminary experimentation.
  • Diversity bonus cap = 12 points
    Eq 16 caps the equity bonus at 12 points with no supporting data or policy analysis.
assumptions (6)
  • domain assumption Admissions decisions are decomposable into academic, essay, and extracurricular components.
    The entire CAPS architecture assumes these three modules capture holistic review, stated in Section III.
  • domain assumption Synthetic data distributions approximate real applicant pools.
    Section IV says the data are designed to follow realistic distributions, but no comparison to real admissions data is shown.
  • domain assumption Admission tiers in the synthetic set were generated in a way that reflects real holistic decisions.
    Tiers are derived from the same SAS, EQI, and EIS scores later used as features, making the validation circular.
  • ad hoc to paper GPT-4o scores align with human judgment.
    Only a subset of scores was manually verified by admissions consultants; no inter-rater statistics or full human label set is provided.
  • domain assumption Expert-defined weights reflect admissions heuristics.
    Used to set SAS and CAPS weights, with references to admissions guidance but no direct evidence that the chosen numbers are valid.
  • standard math Standard machine learning assumptions for PCA, logistic regression, and XGBoost hold.
    These are background methods used as tools; the paper does not need to prove their mathematical foundations.
invented entities (1)
  • CAPS composite score and module scores (SAS, EQI, EIS)
    purpose: Provide a single interpretable number summarizing holistic admissions review.
    The scores are defined inside the paper and validated only on synthetic data; no real admissions outcomes are used.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantifying Holistic Review: A Multi-Modal Approach to College Admissions Prediction." pith.science (2026). https://pith.science/paper/WEG3NWIO

@misc{pith2026250715862,
  author       = {Pith},
  title        = {Pith review of: Quantifying Holistic Review: A Multi-Modal Approach to College Admissions Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WEG3NWIO}},
  note         = {Machine review of arXiv:2507.15862}
}
read the original abstract

This paper introduces the Comprehensive Applicant Profile Score (CAPS), a novel multi-modal framework designed to quantitatively model and interpret holistic college admissions evaluations. CAPS decomposes applicant profiles into three interpretable components: academic performance (Standardized Academic Score, SAS), essay quality (Essay Quality Index, EQI), and extracurricular engagement (Extracurricular Impact Score, EIS). Leveraging transformer-based semantic embeddings, LLM scoring, and XGBoost regression, CAPS provides transparent and explainable evaluations aligned with human judgment. Experiments on a synthetic but realistic dataset demonstrate strong performance, achieving an EQI prediction R^2 of 0.80, classification accuracy over 75%, a macro F1 score of 0.69, and a weighted F1 score of 0.74. CAPS addresses key limitations in traditional holistic review -- particularly the opacity, inconsistency, and anxiety faced by applicants -- thus paving the way for more equitable and data-informed admissions practices.

Figures

Figures reproduced from arXiv: 2507.15862 by the authors.

Figure 1
Figure 1. Radar comparison of CAPS, Transformer-only, and tra [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Framework of the CAPS B. SAS Module The Standardized Academic Score (SAS) quantifies an applicant’s academic ability based on multiple metrics: GPA, standardized tests scores (SAT/ACT), English proficiency test scores (TOEFL/IELTS)[16], the number of Advanced Placement (AP) exams scored at 5, and the rigor of high school coursework (Course Difficulty). These input features are standardized using the Z-score nor￾mali… view at source ↗
Figure 3
Figure 3. MiniLM Embedding Pipeline . c) Step 3: XGBoost Regression Model: The GPT-4o scores (3 features) are concatenated with MiniLM embeddings (384 features), resulting in a 387-dimensional feature vector. This serves as input to an XGBoost regression model trained [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: EQI Module Architecture D. EIS Module The Extracurricular Impact Score (EIS) module quan￾tifies the quality, leadership, and thematic coherence of ap￾plicants’ extracurricular activities using a hybrid approach combining LLM-based evaluation and expert-informed tiering…
Figure 5
Figure 5. Figure 5: EIS Module Architecture E. CAPS Score Fusion Module The Comprehensive Applicant Profile Score (CAPS) inte￾grates the outputs from the three core modules: SAS, EQI and EIS —into a unified, interpretable score for holistic evaluation. a) Step 1: Input Standardization.: T…
Figure 6
Figure 6. Figure 6: Predicted vs. true EQI scores on held-out test set. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: SHAP Summary Plot for the EQI XGBoost Model. This [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 15 canonical work pages

  1. [1]

    Top US admissions trends 2023-24: Expert insights & tips - Crimson Education SG,

    A. Patel, “Top US admissions trends 2023-24: Expert insights & tips - Crimson Education SG,”Crimson Education, Jun. 4,

    work page 2023

  2. [2]

    Augmenting holistic review in university admission using natural language processing for essays and recommendation letters,

    A. Haim, S. K. Shaw, and N. Heffernan, “Augmenting holistic review in university admission using natural language processing for essays and recommendation letters,”arXiv preprint arXiv:2306.12345, 2023

    arXiv 2023

  3. [3]

    Admission Prediction in Undergraduate Applications: an Interpretable Deep Learning Approach

    A. Priyadarshini, B. Martinez-Neda, and S. Gago-Masague, “Admission Prediction in Undergraduate Applications: an Interpretable Deep Learn- ing Approach,”arXiv preprint arXiv:2401.11698, 2024

    work page Pith review arXiv 2024

  4. [4]

    Common Data Set,

    Common Data Set, “Common Data Set,”Common Data Set. [Online]. Available: https://commondataset.org/. [Accessed: Jun. 28, 2025]

    work page 2025

  5. [5]

    Evaluating a Learned Admission-Prediction Model as a Replacement for Standardized Tests in College Admissions,

    H. Lee, R. F. Kizilcec, and T. Joachims, “Evaluating a Learned Admission-Prediction Model as a Replacement for Standardized Tests in College Admissions,” inProc. Tenth ACM Conf. on Learning @ Scale, Copenhagen, Denmark, 2023, doi: 10.1145/3573051.3593382

    arXiv 2023

  6. [6]

    J. W. Young and J. L. Kobrin,Differential Validity, Differential Pre- diction, and College Admission Testing: A Comprehensive Review and Analysis, Research Report No. 2001-6. New York, NY , USA: The College Board, 2001

    work page 2001

  7. [7]

    Augment- ing Holistic Review in University Admission using Natural Language Processing for Essays and Recommendation Letters,

    J. Lee, B. Thymes, J. Zhou, T. Joachims, and R. F. Kizilcec, “Augment- ing Holistic Review in University Admission using Natural Language Processing for Essays and Recommendation Letters,”arXiv preprint arXiv:2306.17575, 2023

    arXiv 2023

  8. [8]

    J. S. Kusuma, K. Halim, E. J. P. Pranoto, B. Kanigoro and E. Irwansyah, ”Automated Essay Scoring Using Machine Learning,” 2022 4th Int. Conf. on Cybernetics and Intelligent System (ICORIS), Prapat, Indone- sia, 2022, pp. 1-5, doi: 10.1109/ICORIS56080.2022.10031338

    work page arXiv 2022

Show all 21 references
  1. [9]

    W. N. Espeland and M. Sauder,Engines of Anxiety: Academic Rankings, Reputation, and Accountability. New York, NY , USA: Russell Sage Foundation, 2016

    2016

  2. [10]

    Krislov and N

    M. Krislov and N. Lemann,Higher Admissions: The Rise, Decline, and Return of Standardized Testing. Princeton, NJ, USA: Princeton University Press, 2024

    2024

  3. [11]

    Au- tomated Essay Scoring Using Transformer Models,

    S. Ludwig, C. Mayer, C. Hansen, K. Eilers, and S. Brandt, “Au- tomated Essay Scoring Using Transformer Models,”arXiv preprint arXiv:2110.06874, 2021

    2021 arXiv

  4. [12]

    Predicting GPA of University Students with Supervised Regression Machine Learning Models,

    L. Fal ´at and T. Piscov ´a, “Predicting GPA of University Students with Supervised Regression Machine Learning Models,”Applied Sciences, vol. 12, no. 17, p. 8403, 2022, doi: 10.3390/app12178403

    2022 doi

  5. [13]

    A Unified Approach to Interpreting Model Predictions,

    S. Lundberg and S. Lee, “A Unified Approach to Interpreting Model Predictions,”arXiv preprint arXiv:1705.07874, 2017

    2017 arXiv

  6. [14]

    Extracurricular involvement in high school is not a level playing field,

    J. J. Park and B. H. Kim, “Extracurricular involvement in high school is not a level playing field,”Brookings Institution, May 17, 2023. [Online]. Available: https://www.brookings.edu/articles/ extracurricular-involvement-in-high-school-is-not-a-level-playing-field/. [Accessed...

    2023

  7. [15]

    Factors in the Admission Decision,

    National Association for College Admission Counseling, “Factors in the Admission Decision,” 2023. [Online]. Available: https://www.nacacnet. org/factors-in-the-admission-decision/. [Accessed: Jul. 10, 2025]

    2023

  8. [16]

    Ikeda et al.,Aligning Scores of Language Proficiency Tests: A Score Concordance Study Between IELTS Academic and TOEFL iBT®, ETS Research Report No

    N. Ikeda et al.,Aligning Scores of Language Proficiency Tests: A Score Concordance Study Between IELTS Academic and TOEFL iBT®, ETS Research Report No. RR-25-02. Princeton, NJ, USA: Educational Testing Service, 2025. [Online]. Available: https://www.ets.org/Media/ Research/pdf...

    2025

  9. [17]

    MINILM: Deep Self-Attention Distillation for Task- Agnostic Compression of Pre-Trained Transformers,

    W. Wang et al., “MINILM: Deep Self-Attention Distillation for Task- Agnostic Compression of Pre-Trained Transformers,”arXiv preprint arXiv:2002.10957, 2020

    2002 arXiv

  10. [18]

    GPT-4o System Card,

    OpenAI, “GPT-4o System Card,”arXiv preprint arXiv:2410.21276, 2024

    2024 arXiv

  11. [19]

    4 Tiers of Extracurricular Activities - The Complete Guide,

    Inspira Advantage, “4 Tiers of Extracurricular Activities - The Complete Guide,” Apr. 11, 2025. [Online]. Available: https://www. inspiraadvantage.com/blog/tiers-of-extracurricular-activities. [Accessed: Jul. 11, 2025]

    2025

  12. [20]

    Admissions by source school,

    University of California, “Admissions by source school,” [Online]. Available: https://www.universityofcalifornia.edu/about-us/ information-center/admissions-source-school. [Accessed: Jul. 11, 2025]

    2025

  13. [2024]

    Available: https://www.crimsoneducation.org/sg/blog/ us-admissions-2023-24-data-trends/

    [Online]. Available: https://www.crimsoneducation.org/sg/blog/ us-admissions-2023-24-data-trends/. [Accessed: Jun. 28, 2025]

    2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.