Thesis Completion Rates by University: 50+ Statistics & Data (2026)

Thesis completion rates statistics by university reveal a sobering reality: approximately 40–50% of doctoral candidates in the United States never finish their PhD, and globally only 56.6% of students enrolled in doctoral programs complete their degree within ten years of entry (Council of Graduate Schools, 2022). These figures — sourced from the NSF Survey of Earned Doctorates, OECD Education at a Glance, and institutional data published by universities including Duke, Stanford, and Cornell — represent one of the most significant efficiency challenges in modern higher education. For students currently navigating a thesis or dissertation, understanding where the attrition happens and why is the first step toward beating the odds.

This roundup aggregates more than 50 data points from authoritative sources across PhD attrition, time-to-completion, master’s programme success rates, mental health burden, plagiarism detection, AI tool adoption, supervision quality, and demographic gaps. Every statistic is cited with source and year.

1. PhD Attrition Rates Globally

The most cited single figure in doctoral education research is the 40–50% attrition rate for US PhD programmes — a statistic that has remained stubbornly stable for three decades despite institutional interventions. The Council of Graduate Schools (CGS) 10-year longitudinal study of 49,000 students across 30 institutions and 54 disciplines found a 10-year cumulative completion rate of 56.6%, implying 43.4% non-completion.

Video: Andy Stapleton — Unglamorous Truths About Writing a Thesis | PhD, Masters, Bachelors

Key discipline-level findings from the CGS baseline study:

• STEM fields report the highest 10-year completion rates: mathematics and physical sciences reach 64%, life sciences 60%.
• Humanities record the lowest: completion sits at 49% after 10 years, partly driven by longer time-to-degree expectations.
• Social sciences sit at 55%, engineering at 61%.
• Most attrition occurs in years 1–2 (coursework phase) and around the dissertation proposal stage (years 3–4).
• Students who pass their qualifying exam have a completion probability of approximately 80% (CGS, 2022).

At the institutional level, Duke University’s published PhD completion statistics show rates ranging from 72% (humanities) to 91% (engineering and physical sciences) across a 10-year window. Stanford’s IRDS data reports a median 5-year completion rate of 55% rising to 87% by year 10. Cornell Graduate School data shows 79% overall completion for cohorts entering from 2009–2013.

For the most current US doctoral degree data by field, see the NSF Survey of Earned Doctorates 2024 — the definitive annual census of all 58,131 research doctorates granted in the US.

Also see: PhD Dropout Rates by Country: Statistics and Data (2026) for a full country-by-country breakdown of attrition.

2. Time-to-Completion by Country and Field

Time-to-degree is a key performance metric for doctoral programmes and a strong predictor of completion itself — the longer a student takes, the greater the risk of dropout. The NSF Survey of Earned Doctorates 2024 (covering all 58,131 research doctorates granted in the US in 2024) provides the most granular current data.

• US median time-to-degree (all fields, 2024): 5.9 years (NSF SED 2024)
• Humanities: 9.1 years median (NSF SED 2024)
• Education: 12.6 years median — the longest of any broad field (NSF SED 2024)
• Physical and earth sciences: 5.4 years median (NSF SED 2024)
• Engineering: 5.1 years median (NSF SED 2024)
• Life sciences: 5.8 years median (NSF SED 2024)
• UK: 3–4 years full-time (HESA 2023), with a maximum funded period of 4 years under UKRI guidelines
• Australia: 3.5 years average (Australian Government DESE 2023)
• OECD average: ~4.5 years across member countries (OECD EAG 2025)

The NSF SED 2024 also found that 45% of US doctoral recipients took longer than 6 years to complete, and 21% took longer than 8 years. Longer programmes strongly correlate with funding gaps: students who lose stipend support in their final two years take an average of 14 additional months to complete (CGS, 2022).

3. Master’s Thesis Completion Rates

Master’s completion data is considerably harder to find than doctoral data. The CGS acknowledges that master’s completion and attrition is a significant research gap. What is available suggests wide variation by programme type.

• The CGS STEM Master’s Pilot Study (2019) found completion rates between 48% and 85% across disciplines, with an average of around 68%.
• Research master’s programmes (thesis-based) report lower completion than taught master’s (coursework-only): approximately 61% vs 79% (ResearchGate meta-analysis, 2023).
• In the UK, 83% of full-time master’s students complete their programme within two years (HESA 2023).
• Part-time master’s students complete at a markedly lower rate: 62% within 5 years (HESA 2023).
• Online master’s programmes in the US report completion rates of 40–55%, significantly below campus-based equivalents (US Department of Education IPEDS 2023).
• In Australia, the completion rate for master’s by research within 4 years is 72% (DESE 2023).

The thesis component itself is a major attrition point within master’s programmes. A 2023 analysis in Studies in Higher Education found that 28% of master’s non-completers cited the dissertation or thesis specifically as the reason for withdrawal, second only to financial pressures (34%).

4. Mental Health Statistics Among Thesis Writers

The mental health dimension of thesis writing is now well-documented in the literature and represents both a humanitarian concern and a structural driver of attrition.

• PhD students are 6× more likely to experience anxiety and depression than the general population (Evans et al., Nature Biotechnology, 2018 — the landmark study still widely cited in 2025 policy documents).
• A 2023 survey of 3,500 PhD students across 26 countries by the Nature research family found that 36% reported seeking help for anxiety or depression linked to their PhD, up from 26% in 2019.
• 41% of PhD students scored in the moderate-to-severe range for depression on standardised scales (Levecque et al., Research Policy, 2017; validated by replication studies through 2024).
• Thesis-writing anxiety is measurable and distinct: a 2024 study in Higher Education found that students in the dissertation phase scored 22% higher on academic anxiety inventories than students in coursework phases.
• 76% of PhD students who reported poor supervisor relationships also reported clinically significant anxiety (Nature PhD Student Survey, 2023).
• Students with access to dedicated mental health services through their institution complete at rates 8–12 percentage points higher than peers without such access (Times Higher Education Doctoral Experience Survey, 2024).

See also: PhD Student Burnout: The Hidden Crisis Behind Thesis Mental Health Data (2026) and Student Mental Health and Thesis Writing: 2026 Data and Statistics for in-depth analysis.

5. Plagiarism Detection Rates in Thesis Submissions

Academic integrity data has become considerably richer since Turnitin’s widespread adoption and the introduction of AI-content detection tools from 2023 onward.

• Turnitin’s 2024 Academic Integrity Report found that 22 million student submissions were flagged for potential AI-generated content in the 12 months to June 2024.
• Institutional plagiarism detection rates in thesis submissions (similarity above 20% threshold) average 8–12% across OECD universities (Turnitin Global Report 2023).
• In a survey of 800 UK institutions by JISC (2023), 73% reported an increase in academic misconduct cases since 2022, attributed primarily to AI-assisted writing.
• Self-reported rates differ sharply from detected rates: a 2023 meta-analysis in Assessment & Evaluation in Higher Education found that up to 30% of students reported having engaged in some form of academic dishonesty, versus 8–12% detected.
• STEM theses are flagged at lower average similarity scores (mean: 9%) than humanities theses (mean: 17%), reflecting differences in citation-heavy writing conventions (Turnitin 2023).

For detailed data on plagiarism trends, see: Plagiarism Rates in Universities Statistics 2026 and AI in Academic Writing Statistics 2026.

6. AI Tool Adoption Among Thesis Writers

The introduction of large language models from late 2022 onward has fundamentally altered the academic writing landscape. Reliable adoption data is now emerging from institutional surveys and third-party research.

• A 2024 survey by Times Higher Education of 12,000 students in 100 countries found that 58% of graduate students used AI tools in some capacity during thesis writing.
• 32% of PhD students reported using AI for literature review tasks; 28% for drafting and editing; 19% for data analysis support (THE, 2024).
• Only 14% of universities had a fully operationalised AI use policy for postgraduate research by mid-2024 (JISC Higher Education AI Survey, 2024).
• AI tool use correlates with faster draft production: students using AI writing assistance completed first chapter drafts in a median of 6 days fewer than non-users (pilot study, University of Melbourne, 2024).
• Turnitin’s AI detection flagged 6% of doctoral thesis submissions globally as containing >20% AI-generated text in 2024 — a threefold increase from 2023.
• Among students using AI, 67% reported that their primary concern was whether they were using it ethically (THE Student Survey, 2024).

Students looking to use AI ethically should read: What Is the Best AI Tool for Writing a Thesis in 2026? and Best AI Thesis Writing Tools Compared in 2026. Tesify is designed specifically for postgraduate writers seeking AI assistance that stays within institutional integrity boundaries.

7. Impact of Supervision Quality on Completion Rates

Supervision quality is consistently the strongest single institutional predictor of PhD completion across every major longitudinal study. The data below is unambiguous.

The 2024 Times Higher Education Doctoral Experience Survey (n = 14,500) also found that students who rated their supervisor as “excellent” at giving feedback had a 91% reported intention to complete, versus 63% among those who rated supervision as “poor” — a 28-percentage-point gap that closely tracks actual completion disparities.

Structured writing support tools such as Tesify complement supervisory feedback by giving students a consistent writing framework between meetings, reducing the productivity loss that occurs in unsupervised periods.

8. Gender and Demographic Gaps in Thesis Completion

Completion data disaggregated by gender, ethnicity, and first-generation student status reveals significant structural inequalities within doctoral education.

Beyond representation at graduation, completion rate disparities are stark:

• First-generation college students complete PhDs at rates 11–15 percentage points lower than continuing-generation students (CGS, 2022).
• Under-represented minority students in STEM disciplines face a completion gap of approximately 8 percentage points compared to White and Asian peers, after controlling for programme selectivity (NSF NCSES analysis, 2024).
• Women in STEM fields complete at similar rates to men when supervision and funding are equitable — the gap narrows to under 2 percentage points in fully funded programmes (OECD EAG 2025).
• International PhD students in the US complete at a rate of 63% within 10 years, 6.4 percentage points below domestic students (CGS, 2022), driven largely by visa uncertainty and limited access to campus support services.
• Students with a declared disability complete at rates approximately 12 percentage points lower than non-disabled peers across OECD countries (OECD EAG 2025).

For broader higher education demographic data, see: Higher Education Statistics and Trends 2026: The Global Data Roundup.

Methodology Note

All statistics cited in this article are drawn from peer-reviewed research, official government surveys, or reports from established higher education data agencies. Primary sources include:

• NSF Survey of Earned Doctorates (SED) 2024 — annual census of all US research doctorate recipients (ncses.nsf.gov)
• Council of Graduate Schools (CGS) — PhD Completion and Attrition longitudinal study, baseline analysis covering 49,000 students
• OECD Education at a Glance 2025 — cross-national tertiary education statistics across 38 OECD member states
• HESA / Research England (HEFCE successor) — UK postgraduate data from the Higher Education Statistics Agency
• Turnitin Academic Integrity Reports (2023, 2024) — submission-level data on similarity and AI detection
• Nature PhD Student Survey — annual survey of 3,500–6,300 PhD students across 26 countries
• Times Higher Education Doctoral Experience Survey (2024, n = 14,500)
• Australian Government DESE — Higher Education Statistics collection

Where a statistic is drawn from a single institutional dataset (e.g., Duke, Stanford, Cornell), this is noted explicitly. Completion rate definitions vary: some institutions report within-time completion; others report 10-year cumulative rates. Readers should verify the timeframe when comparing figures across systems.

Frequently Asked Questions