phd survey and data science

Survey and Data Science

Statistics & Data Science

Dietrich college of humanities and social sciences, ph.d. programs, our ph.d. programs enable students to pursue a wide range of research opportunities, including constructing and implementing advanced methods of data analysis to address crucial cross-disciplinary questions, along with developing the fundamental theory that supports these methods..

Unique opportunities for our Ph.D. students include:

• We host four cross-disciplinary joint Ph.D. programs for students who want to specialize in machine learning , public policy , neuroscience , and the link between engineering and policy .
• Our faculty have deep involvement in a range of important, data-rich scientific collaborations, including in the areas of genetics, neuroscience, astronomy, and the social sciences. This allows students to have easy access to both the crucial questions in these fields, and to the data that can provide the answers.
• Students begin work on their Advanced Data Analysis Project in the second semester. This year-long, faculty/student collaboration, distinct from the thesis, provides an immediate intensive research experience.
• Carnegie Mellon is home to the first Machine Learning Department . Many of our faculty maintain joint appointments with this Department and they (and our students) have strong connections to this exciting and growing area of research.

The programs leading to the degree of   Doctor of Philosophy in Statistics   seek to strike a balance between theoretical and applied statistics. The Ph.D. program prepares students for university teaching and research careers, and for industrial and governmental positions involving research in new statistical methods. Four to five years are usually needed to complete all requirements for the Ph.D. degree.

These pages present the requirements for each of our Ph.D. programs.

The page   "Core Ph.D. Requirements"   lays out the requirements for all Ph.D. students, while each of the four joint programs are described under the Joint Ph.D. Degrees pages. Our Ph.D. students can also earn a   Master of Science in Statistics   as an intermediate step towards their ultimate goal.

Joint Ph.D. Programs

Statistics/machine learning, statistics/public policy, statistics/engineering and public policy, statistics/neural computation  .

Doctor of Philosophy in Data Science

Developing future pioneers in data science

The School of Data Science at the University of Virginia is committed to educating the next generation of data science leaders. The Ph.D. in Data Science is designed to impart the skills and knowledge necessary to enable research and discovery in data science methods. Because the end goal is to extract knowledge and enable discovery from complex data, the program also boasts robust applied training that is geared toward interdisciplinary collaboration. Doctoral candidates will master the computational and mathematical foundations of data science, and develop competencies in data engineering, software development, data policy and ethics. 

Doctoral students in our program apprentice with faculty and pursue advanced research in an interdisciplinary, collaborative environment that is often focused on scientific discovery via data science methods. By serving as teaching assistants for the School’s undergraduate and graduate programs, they learn to be adroit educators and hone their critical thinking and communication skills.

LEARNING OUTCOMES

Pursuing a Ph.D. in Data Science will prepare you to become an expert in the field and work at the cutting edge of a new discipline. According to LinkedIn’s most recent Emerging Jobs Report, data science is booming and data scientist is one of the top three fastest growing jobs. A Ph.D. in Data Science from the University of Virginia opens career paths in academia, industry or government. Graduates of our program will:

• Understand data as a generic concept, and how data encodes and captures information
• Be fluent in modern data engineering techniques, and work with complex and large data sets
• Recognize ethical and legal issues relevant to data analytics and their impact on society 
• Develop innovative computational algorithms and novel statistical methods that transform data into knowledge
• Collaborate with research teams from a wide array of scientific fields 
• Effectively communicate methods and results to a variety of audiences and stakeholders
• Recognize the broad applicability of data science methods and models 

Graduates of the Ph.D. in Data Science will have contributed novel methodological research to the field of data science, demonstrated their work has impactful interdisciplinary applications and defended their methods in an open forum.

A Week in the Life: First-Year Ph.D. Student

Ph.D. Student Profile: Jade Preston

Ph.D. Student Profile: Beau LeBlond

Get the latest news.

Subscribe to receive updates from the School of Data Science.

• Prospective Student
• School of Data Science Alumnus
• UVA Affiliate
• Industry Member

_linkedin_partner_id = "1170812"; window._linkedin_data_partner_ids = window._linkedin_data_partner_ids || []; window._linkedin_data_partner_ids.push(_linkedin_partner_id); (function(){var s = document.getElementsByTagName("script")[0]; var b = document.createElement("script"); b.type = "text/javascript";b.async = true; b.src = "https://snap.licdn.com/li.lms-analytics/insight.min.js"; s.parentNode.insertBefore(b, s);})(); School of Education

Ph.d. in educational statistics and research methods.

The Ph.D. in Educational Statistics and Research Methods (ESRM) prepares students interested in education data science, research methods, statistics, causal inference, psychometrics, and evaluation to develop, critically evaluate, and properly use sophisticated quantitative and mixed methodologies to solve important problems in education.

The Ph.D. in ESRM is a STEM-designated degree program .

Many of our ESRM students and faculty are affiliates of the Data Science Institute , where they participate in seminars, courses, and research projects. ESRM students may elect to earn their Master of Data Science or  Masters in Applied Statistics along the way to their PhD in ESRM.

Students will:

• Design research projects, focusing either on advancing research methodologies or on applying advanced methods to education issues
• Develop assessment instruments
• Implement program evaluation
• Understand psychometric theory, as well as technical issues underlying construction and use of tests for selection, placement, and instruction
• Develop skills in advanced statistical modeling using a variety of software
• Examine how these statistical models are applied to areas such as school effectiveness, economic and social stratification, the structure of human abilities, and achievement growth

Doctoral students also present research at conferences, collaborate with faculty on peer-reviewed publications, engage in the work of interdisciplinary research centers through graduate assistantships, and learn in an environment with small class sizes and supportive faculty.

Our graduates accept tenure-track or research faculty positions in research universities as well as research positions in state departments of education, school districts, and organizations such as the Educational Testing Service, Pearson, and Mathematica.

Program Coordinator: Dr. Kenneth Shores

Program Faculty

Program Requirements

• Core Content Courses: Core coursework includes two Proseminars ( EDUC 805 ,  EDUC 806 ) that students take in the first two semesters of their program. These courses introduce the key domains of education research, examined through qualitative and quantitative data collection methods and analyses. Topics include learning and development, curriculum and instruction, school reform, and social contexts of education. These courses also allow students to interact with PhD in Education students across all 6 specializations and PhD in Economic Education students.
• Research Methods Core Courses: Students take 15 credits of research methods core courses, including EDUC 856 : Introduction to Statistical Inference, EDUC 812 : Regression and Structural Equation Modeling, EDUC 865 : Educational Measurement Theory, EDUC 874 : Applied Multivariate Data Analysis, and EDUC 850 : Qualitative Research in Educational Settings. These courses combine sophisticated theory with practical application, beginning with an introduction to statistical inference and extending to structural equation modeling, multiple regression, and the use of applied multivariate data analysis.
• Additional Required Methods Courses: Students take 9 credits of additional required methods courses, including EDUC 826 : Mixed Methods in Social Science Research, EDUC 863 : Program Evaluation in Education, and EDUC 873 : Multilevel Models in Education.
• Elective Methods Courses: Students take 3 or more credits of additional elective methods courses. Topics include Advanced Structural Equation Modeling, Bayesian Analysis and Monte Carlo Simulation, Causal Inference, Data Mining in Education, Experimental and Quasi-Experimental Design, Item Response Theory, Longitudinal Data Analysis, Randomized Field Trials, Secondary Analysis of Large-Scale Survey Data, and Social Network Analysis.
• Education Specialization Courses: Students take 6 additional credits of content courses from a specialization area within the PhD in Education. View specialization courses online .
• Colloquium Series Course: A one-credit course ( EDUC 840 ) is offered each semester in conjunction with the colloquium series , and students complete a minimum of 4 credits of colloquium. The research colloquia introduce students to the foremost thinkers and researchers in the field of education. Guest scholars are invited to share their research findings with doctoral students and faculty in a setting that encourages collegiality and familiarizes students with a number of scholarly presentation styles and content areas.
• Dissertation credits: Nine hours of dissertation credit ( EDUC 969) is required of all PhD students, and additional coursework may be specified by a student’s advisory committee as part of the student’s Individual Program Plan.
• Total credits: A minimum of 55 credit hours is required to complete the program.

Download a sample student schedule for this program or view the schedule of course offerings .

Apprenticeship Activities

All of our PhD students are offered full funding for four years. Funded students participate in a 20-hour a week assistantship where they work closely with one or more UD professors, and have opportunities to learn and practice multiple methodologies, analyze data, and co-author academic papers. All students participate in the Steele Symposium, an annual college research forum; submit a publication to a peer-reviewed journal on which they are a coauthor; present their work at a national conference; and develop skills in university teaching.

Most of our students are in residence for all four years of the program (assistantships typically require residency, though there are exceptions). Students are required to complete at least one year in residence (one continuous academic year with 9 credit hours per semester). Students are strongly encouraged to complete this requirement in the first year.

Examinations

All students must pass an assessment based on the work completed in the Proseminars at the end of the first year. After students successfully pass the First Year Assessment, they may enroll in second-year courses. This First Year Assessment fulfills the University requirement for a qualifying examination.

Students must also pass the Fourth Year Exam in order to proceed to the dissertation. The exam assesses student proficiency in integrating various aspects of research methodology to address substantive issues in education.

Dissertation Proposal

Students complete a written proposal for their capstone dissertation project and defend it orally before their advisory committee.

Dissertation and Defense

Students complete a dissertation, an original work of scholarship, meeting SOE, College, University, and professional requirements. They also complete an oral defense of the work before their advisory committee.

Program Policy Document

Students may download the program policy document for complete information about this degree.

Program Requirements for the Master of Arts in Education

The MA in Education provides a master’s degree option for PhD students in good standing who want to obtain a master’s degree in conjunction with their doctoral degree, or for students in good standing who must leave the doctoral program prematurely because of family, health, or personal reasons. Students will not be admitted directly to the MA program, since the program requirements are embedded within the PhD requirements.

Admission Information

To apply to the PhD in ESRM program, complete the steps of the UD online graduate application process .

Application Requirements

Some application items specific to the PhD in ESRM program include:

• Transcripts of all previous academic work at the undergraduate and graduate (if applicable) level. Applicants may upload unofficial copies of their transcripts and if admitted, all transcripts will be verified by the Graduate College. Applicants who previously attended the University of Delaware still need to upload an unofficial transcript, but do not need to provide official transcripts for verification. Please do not send any transcripts to the School of Education.
• GRE scores are required. Students typically are expected to have minimum scores of 150 on the verbal and quantitative sections and a 4.0 on the analytic writing section. Most admitted students have far higher than the minimum scores. The GRE is optional for Fall 2025 applicants. Please see the note at the top of this section.
• Three letters of recommendation are required. Applicants should select recommenders who can comment on their potential to succeed in doctoral work.
• Applicants should introduce themselves and discuss educational and career goals related to the PhD in ESRM program and how this program is a good match for their interests. Applicants should identify their focus area and potential research interest.
• While there are no requirements set by the School of Education, personal statements are generally 2-5 pages in length.
• A resume is required.
• No writing samples or supplemental documents are required.
• International applicants must submit scores from either the TOEFL, IELTS, or iTEP Academic Plus. Scores more than two years old cannot be validated or considered official. Required minimum scores for the TOEFL are 100 (internet-based test-iBT), 600 (paper-based test), or 250 (computer-based test). For the IELTS, the minimum score is 7.0. For the iTEP Academic Plus, the minimum score is 4.5.

Application Deadline

The deadline for all applications is December 1  for study beginning the following Fall term.

In general, it is not possible to take required core courses before becoming admitted. The required core courses are generally restricted to students already admitted into the program.

Cost and Financial Support

Our full-time PhD in ESRM students receive guaranteed financial support for four years through a variety of sources, including assistantships and tuition scholarships. Students with assistantships receive 100% tuition scholarship and a 9-month stipend, plus health insurance. Merit-based supplemental funding is available. For more information about this financial support, visit CEHD’s graduate tuition page .

Graduate student assistants work 20 hours a week, engaged closely with their faculty mentors in research and teaching activities. Prospective students can learn more about PhD assistantship experiences through our PhD student spotlights and our PhD student directory .

We also have conference travel funding available through the the SOE and the UD Graduate College.

Graduate Placements and Jobs

Graduates of this STEM-designated degree program will be well prepared for careers in applied education research in several arenas in both the for-profit and non-profit sectors, such as:

• Tenure-track or research faculty at Research-I universities
• Research/evaluation staff at national research organizations (e.g., Abt, AIR, Mathematica, MDRC, RAND, Westat)
• Research/evaluation staff at local research organizations (e.g., Research for Action, Branch Associates, Research for Better Schools)
• Research/psychometric staff at national measurement organizations (e.g., College Board, CTB, ETS, Harcourt/Riverside, Pearson)
• Research/evaluation Staff in federal agencies (e.g., Institute of Education Sciences) and regional agencies (e.g., REL Mid-Atlantic)
• Research/evaluation Staff at local school districts and state education agencies

How to Apply

Applications for all graduate programs at the University of Delaware are done online through the UD Graduate College. To apply to the PhD in ESRM program, complete the steps of the UD online graduate application process . For information about graduate tuition, visit UD’s graduate tuition page for CEHD programs.

Student Spotlight

Kati Tilley

“Through my assistantship, I have gained critical experience in communicating research findings. One of the most valuable experiences I have had was learning to write and present results to a non-academic audience. I led the development of an individualized report of survey results and presented them in person to the schools who participated in our study.”

Request More Information

Visit Campus

Awards and Scholarships

• Major and Minors
• Graduate Programs
• Prospective Students
• Academic Calendar
• Social Media
• 113 Willard Hall Education Building , Newark, DE 19716
• Phone: (302) 831-8695
• Fax: (302) 831-4110

Education & Training

SRC offers a variety of educational opportunities, ranging from a doctoral and masters program in survey methodology to internship opportunities for graduate and undergraduate students. Please select a link below to find out further information on the program of interest to you.

SRC Summer Institute in Survey Research Techniques (SRCSI) (link opens in a new window)

The SRC Summer Institute seeks to provide rigorous and high quality graduate training in all phases of survey research. The program teaches state-of-the-art practice and theory in the design, implementation, and analysis of surveys.

More (link opens in a new window)

Program in Survey and Data Science (PSDS) (link opens in a new window)

The University of Michigan Program in Survey Methodology, established in 2001, seeks to train future generations of survey methodologists. The program offers doctorate and master of science degrees and a certificate through the University of Michigan.

Joint Program in Survey Methodology (JPSM) (link opens in a new window)

The Joint Program in Survey Methodology is a cooperative program between the University of Michigan, the University of Maryland and Westat. The program seeks to train future generations of survey methodologists. The program offers doctorate and master of science degrees.

Fellowships (link opens in a new window)

The Survey Research Center supports its scientists in research and training with fellowships.

• SRC Summer Institute in Survey Research Techniques (SRCSI)
• Program in Survey and Data Science (PSDS)
• Joint Program in Survey Methodology (JPSM)
• Fellowships

Department of Statistics and Data Science

Ph.d. program.

Fields of study include the main areas of statistical theory (with emphasis on foundations, Bayes theory, decision theory, nonparametric statistics), probability theory (stochastic processes, asymptotics, weak convergence), information theory, bioinformatics and genetics, classification, data mining and machine learning, neural nets, network science, optimization, statistical computing, and graphical models and methods.

With this background, graduates of the program have found excellent positions in universities, industry, and government. See the list of alumni for examples.

• For Individuals
• For Businesses
• For Universities
• For Governments
• Online Degrees
• Find your New Career
• Join for Free

Getting a PhD in Data Science: What You Need to Know

A PhD in data science prepares you for some of the most cutting-edge research in the field and can advance your career. But, whether you should pursue one depends on your own personal goals and resources. Learn more inside.

A Doctor of Philosophy (PhD) is the highest degree that a professional can obtain in the field of data science. Focused primarily on equipping degree holders with the skills and knowledge required to conduct original research, a PhD prepares degree holders for advanced professional positions in both industry and academia. 

But, the path to obtaining a PhD is filled with many years of potentially costly study that can be discouraging to those looking for rapid career progression. Before jumping into a doctoral program, then, it’s important to define what your goals are and how a PhD may (or may not) fit into them. 

In this article, you’ll learn more about PhDs in data science, the different factors you should consider before joining one, and types of programs to consider. At the end, you’ll also find some suggested online courses to help you get started today. 

PhD in Data Science: Overview 

A Doctor of Philosophy (PhD) is the terminal degree in the field of data science, meaning it is the highest possible degree that can be obtained in the subject. Holding a PhD in data science, consequently, signals your mastery and knowledge of the field to both potential employers and fellow professionals. 

At a glance, here’s what you should know about a Data Science PhD: 

PhD vs. Master’s Degree in Data Science

There are two graduate degrees in the field of data science: a master’s in Data science and a PhD in Data Science. While both of these degrees can have a beneficial impact on your job prospects, they also have key differences that might impact which one is better for you. 

A Master’s in Data Science is a graduate degree between a bachelor’s and PhD, which usually takes between one and two years to complete. A master’s degree expands on what was learned in undergraduate school through more advanced courses in topics such as machine learning, data analytics, and statistics. Often, a master’s student in data science also pursues original research and completes a capstone project, which highlights what they learned in their program.

A PhD in Data Science is a research degree that typically takes four to five years to complete but can take longer depending on a range of personal factors. In addition to taking more advanced courses, PhD candidates devote a significant amount of time to teaching and conducting dissertation research with the intent of advancing the field. At the conclusion of their doctoral program, a PhD holder in Data Science will complete a dissertation representing a significant contribution to the field. 

Typically, bachelor’s degree holders entering a PhD program are able to earn their master’s degree as a part of their doctoral program. Those entering a master’s program, however, will usually have to apply for a PhD program even if it’s in the same department. 

Skills and curriculum 

Every PhD program is unique with its own requirements and focus. Nonetheless, they do have similar features, such as course, credit, and teaching requirements. To help you get a better understanding of how a doctoral graduate program in data science might be, here’s an example curriculum from NYU [ 1 ]: 

Complete 72 credit hours while maintaining a cumulative grade point average of 3.0 (out of 4.0) each semester.

Core courses in topics like probability, statistics, machine learning, big data, inference, and research. 

39 credit hours for elective courses in such topics as deep learning, natural language processing, and computational cognitive modeling. 

Complete teaching requirements.

Pass a comprehensive exam. 

Pass the Depth Qualifying Exam (DQE) by May 15 of their fourth semester. 

Complete all steps for approval of their PhD dissertation. 

Is a PhD in Data Science worth it? 

A PhD can open doors to new career opportunities and boost your employment prospects. But, it can also take a lot of time and money to complete. Everyone’s personal and professional goals are different, so consider these things when deciding if you should pursue a PhD in Data Science:  

Cost and time

The amount of time and money it takes to complete a PhD are perhaps the most concrete considerations one makes when deciding whether or not they should pursue a doctoral degree. According to research conducted by Education Data Initiative, the average cost of a doctorate degree is $114,300 and takes roughly four to eight years to complete [ 2 ]. 

The exact amount of time and money you might spend obtaining your doctoral degree will depend on your own circumstances and program. Before applying for a doctoral degree, make sure to review each program’s graduation requirements and costs, so you have a clear understanding of what you’re getting into. 

Data Science PhD salary 

While there are no official statistics on the salary gains data scientist earn by getting a PhD, the median salary for all data scientists is much higher than the national average in the United States. According to the U.S. Bureau of Labor Statistics (BLS), for example, the median salary for data scientists was $100,910 as of May 2021 [ 3 ]. 

Typically, the entry-level degree to get a data science position is a bachelor’s degree, meaning that even just an undergraduate degree could help you land a job that earns a higher than average salary. Nonetheless, a PhD will likely prepare you for more advanced positions that could offer higher pay than less specialized roles. 

Data Science PhD programs 

There are several types of doctoral programs that you might consider if you would like to obtain a PhD in data science. These include: 

PhD in data science online

An online PhD program may appeal to individuals who are interested in a more flexible program that allows them to complete their coursework at their own pace. Often, online programs can also be cheaper than their in-person counterparts, though they often offer less opportunities for networking and mentorship. If you’re an independent, self-starter looking for a program that can fit into their already busy life, then you might consider an online PhD program. 

PhD in data science in-person

An in-person PhD program is a more traditional, educational method in which you attend classes on campus with your peers and instructors. In addition to providing doctoral-level instruction, you will also have more opportunities to network and gain more personalized instruction than you will likely encounter through online programs. In-person programs tend to be more expensive and inflexible than in-person ones.

If you prefer real-world instruction, networking opportunities, and a more rigid structure, then you might consider an in-person doctoral program. 

Alternatives 

As an alternative to a PhD program, you might also consider obtaining a master’s degree. While covering some of the same material as a doctoral program, a master’s usually takes much less time and money to complete.

If you’re motivated primarily by the desire to boost your chances of landing a job and gaining financial stability, then a master’s degree program might better help you achieve your goals.

Learn more about data science 

Whatever your educational goals, data science requires extensive knowledge and training to enter the profession. To prepare for your next career move, then, you might consider taking a flexible online course through Coursera. 

The University of Colorado Boulder’s Data Science Foundations: Data Structures and Algorithms Specialization teaches course takers how to design algorithms, create applications, and organize, store, and process data efficiently. Their online Master of Science in Data Science , meanwhile, teaches broadly applicable foundational skills alongside specialized competencies tailored to specific career paths in just two years of instruction. 

Article sources

NYU Center for Data Science. “ PhD in Data Science, Curriculum , https://cds.nyu.edu/phd-curriculum-info/.” Accessed September 27, 2022. 

Education Data Initiative. “ Average Cost of a Doctorate Degree ,  https://educationdata.org/average-cost-of-a-doctorate-degree.” Accessed September 27, 2022. 

US BLS. “ Occupational Outlook Handbook: Data Scientists , https://www.bls.gov/ooh/math/data-scientists.htm#tab-1.” Accessed September 27, 2022. 

Keep reading

Coursera staff.

Editorial Team

Coursera’s editorial team is comprised of highly experienced professional editors, writers, and fact...

This content has been made available for informational purposes only. Learners are advised to conduct additional research to ensure that courses and other credentials pursued meet their personal, professional, and financial goals.

Search form

Overview of survey and data science, the profession.

• Research Opportunities
• Survey Research Center

Since the 19 th  century, survey research has been used as a tool to describe and understand society. Surveys generally ask individuals questions about themselves and what they think, using their answers as the data on which these descriptions are based. More recently other sources of data have become available that may help researchers understand society, in particular data that are generated in massive quantities, such as sensor data, online transactions, search strings and social media posts. Survey methodologists use and develop techniques to collect and interpret these different kinds of data.

Throughout the world today, specialists in survey methods are found in academia, government, and commerce. Survey researchers in the academic sector investigate discipline-based questions in fields such as sociology, psychology, political science, public health, communication studies, criminology, economics, and gerontology. U.S. government agencies on health, education, justice, transportation, and labor statistics, as well as the Bureau of the Census, collect and disseminate government survey information. The private sector includes research firms devoted to the measurement of media audiences, user experience, political and other opinion research, market and product research, and customer satisfaction.

The Discipline

Survey methodology is the study of sources of error in surveys--the bias and variability that affect the quality of survey data. As a field of knowledge, a profession, and a science, survey methodology seeks to link the principles of design, collection, processing, and analysis of surveys to an understanding of error.

Achieving high quality survey results within the scientific aspects of surveys requires applying principles from academic disciplines such as statistics, the social sciences, and data science. For example, statistics provides a quantitative foundation to examine sources of error and to summarize their effects. Social and cognitive psychology provides the framework for understanding how human behavior affects accuracy in survey responses. Sociology and anthropology offer theories of social stratification and cultural diversity. Computer science provides principles of database design and human-computer interaction, as well as computational techniques such as machine learning. Because these disciplines all contribute to the foundation of survey and data science, it is an inherently multidisciplinary, dynamic field of study.

The Challenge

Every survey involves a number of decisions about its design and implementation, and each decision has the potential to affect the quality and validity of the results. How will the sample be chosen? What mode will be used to pose questions and collect answers from respondents? All surveys involve compromises, and the challenge for the researcher is to determine how best to use the available resources to produce, on balance, the best results. The Michigan Program in Survey and Data Science prepares students to meet this challenge.

Data Science

Students enrolled in the Master of Liberal Arts program in Data Science will develop the skills necessary to analyze, discover, and innovate in a data-rich world. Students gain hands-on experience conducting interdisciplinary data science research.

National Center for Science and Engineering Statistics

• All previous cycle years

The SED is an annual census of research doctorate recipients from U.S. academic institutions that collects information on educational history, demographic characteristics, graduate funding source and educational debts, and postgraduation plans.

Survey Info

• tag for use when URL is provided --> Methodology
• tag for use when URL is provided --> Data
• tag for use when URL is provided --> Analysis

The Survey of Earned Doctorates is an annual census conducted since 1957 of all individuals receiving a research doctorate from an accredited U.S. institution in a given academic year. The SED is sponsored by the National Center for Science and Engineering Statistics (NCSES) within the National Science Foundation (NSF) and by three other federal agencies: the National Institutes of Health, Department of Education, and National Endowment for the Humanities. The SED collects information on the doctoral recipient’s educational history, demographic characteristics, and postgraduation plans. Results are used to assess characteristics of the doctoral population and trends in doctoral education and degrees.

Areas of Interest

• STEM Education
• Science and Engineering Workforce

Survey Administration

The 2022 survey was conducted by RTI International under contract to NCSES.

Survey Details

Featured Survey Analysis

Doctorate Recipients from U.S. Universities: 2022

SED Overview

Data highlights, the number of research doctorates conferred by u.s. institutions, which began a sharp 15-month decline in spring 2020 due to the covid-19 pandemic, rebounded in 2022 with the highest number of research doctorates awarded in any academic year to date.

Over the past 20 years, most of the growth in the number of doctorates earned by both men and women has been in science and engineering (S&E) fields 

Methodology

Survey description, technical notes, technical tables, questionnaires, view archived questionnaires, featured analysis, related content, related collections, survey contact.

For additional information about this survey or the methodology, contact

Get e-mail updates from NCSES

NCSES is an official statistical agency. Subscribe below to receive our latest news and announcements.

Javascript must be enabled for the correct page display.

• Find a Person
• For EMS Faculty & Staff
• Meet the Dean
• Mission, Vision, Values, and Strategic Plan
• Our Leadership
• Graduate Enrollment
• Undergraduate Enrollment
• Degrees Awarded
• Administrative Structure
• Faculty Honors and Awards
• Accreditation
• 125th Anniversary
• Diversity, Equity, and Inclusion
• Entrepreneurship
• Future Energy-Water-Landscape Initiative
• Online Learning
• Sustainability
• Maps and Parking
• Driving Directions
• Building Locator Map
• EMS Museum & Art Gallery
• Bachelor's Programs
• Master's Programs
• Doctoral Programs
• Undergraduate Certificate Programs
• Graduate Certificate Programs
• Online Programs
• John and Willie Leone Family Department of Energy and Mineral Engineering
• Department of Geography
• Department of Geosciences
• Department of Materials Science and Engineering
• Department of Meteorology and Atmospheric Science
• John A. Dutton Institute for Teaching and Learning Excellence
• EMS Energy Institute
• Earth and Environmental Systems Institute
• Scientific Diving Program
• Miner Training Program
• Fletcher L. Byrom Earth and Mineral Sciences Library
• Open Educational Resources
• Weather Education and Outreach
• Welcome Message from the Dean
• Top Reasons to Choose EMS
• What Can I Study?
• Meet Our Students
• Where Our Graduates Go
• Attend Prospective Student Events
• For Parents and Families
• Costs, Financial Aid, and Scholarships
• How to Apply
• Orientation / First-year Experience
• Academic Planning, Progress, and Requirements
• International Student Resources
• Academic Advisers
• Academic Integrity: Undergraduates
• Avoiding Academic Integrity Violations
• Undergraduate Mentoring Program
• Clubs and Organizations
• EMS Living Learning Community Options
• Honors, Exhibitions, and Award Competitions
• Internships & Career Resources
• Study Abroad / Study Away
• Undergraduate Research
• Ryan Family Student Center
• Faculty Senate Petition Support
• Diversity Resources for Undergraduates
• Tutoring and Academic Assistance
• Graduation & Commencement
• Library Subject Guides
• Media Commons - One Button Studio
• News for Undergraduates
• Find a Program of Study
• Graduate Student Profiles
• Contact Graduate Staff
• Request Individual Visit
• How to Apply to Graduate School
• Assistantships, Awards, Scholarships, and Fellowships
• Academic Integrity - Graduate Students
• Dates and Deadlines
• Research Exhibitions
• Commencement Information
• Diversity Resources for Graduate Students
• Graduate Organizations
• Graduate Student and Postdoctoral Scholar Ombuds Program
• Graduate Students News & Announcements
• Endowed Faculty Positions
• Faculty Directory
• Meet our Faculty
• Research Facilities
• Research with Impact
• Institutes and Centers
• Research News
• Invent Penn State
• Partner With Us
• Recruit Students
• Sponsorships
• About the Office of the Associate Dean for Graduate Education and Research
• Research Administration
• Research and Travel Grants
• Resources for Postdoctoral Scholars
• Subscribe to e-News
• Social Media
• Featured Events
• Lecture Series
• College Magazine
• EMS Hall of Fame Educators
• Find an Expert
• Communications Contacts
• Mentoring Program
• GEMS Alumni Society
• Affiliate Program Groups
• Give to Kiusalaas Fund
• Meet Our Alumni
• EMS Alumni Awards
• University Alumni Awards
• Attend an Alumni Event
• Alumni News
• Development & Alumni Relations Staff
• GEMS Board of Directors
• Campaign Priorities
• Why Giving Matters
• Where to Give
• Ways to Give
• Why We Give
• Obelisk Society
• Giving News
• Development Council

Spatial Data Science

As the use of location-based data becomes more pervasive and complicated, we need experts who can leverage spatial data to develop insights and improve decision-making. The Penn State Master of Science in Spatial Data Science (SDS) can prepare you to work at the forefront of the geospatial industry, where Geographic Information Systems (GIS) and spatial analysis intersect with the principles of data science, visualization, and programming.

What you can learn

As a student in this completely online program, you can learn to:

• Leverage the science of spatial analysis and modeling to develop new workflows to solve problems that impact people and our planet
• Architect and implement solutions that blend computational and visual approaches
• Research, communicate, and critique data quality and the results of spatial analyses

Graduates of this program can pursuing a senior-level career in business, public health, emergency management,natural resources, transportation, energy, or urban development. The following roles are often held by people with this type of degree:

• Geographic Data Visualization Specialist
• Geospatial Analyst
• GIS Software Developer
• GIS/Geospatial Data Engineer
• Spatial Data Scientist

For More Information 

• Penn State World Campus website
• University Graduate Bulletin
• Program Metrics
• Julene Santiago Geospatial Portfolio Program Assistant [email protected] 814-865-2557 (local) 877-778-4471 (toll-free)

The Boston Globe: BU survey says people want social networks, not government, to remove fake news

Excerpt from The Boston Globe | By: Hiawatha Bray | June 7, 2024 | Photo courtesy of Michael Dwyer/Associated Press

As the US Supreme Court prepares to decide how far federal agencies can go in fighting online misinformation, a new survey from Boston University researchers finds that most US residents want to be protected from online lies — but preferably not by the government.

A survey conducted by the Communication Research Center at the university’s College of Communication found that 63 percent of respondents want social media companies like Facebook and X to take down false information, while 57 percent say the companies should provide only limited access to such postings. For example, the companies sometimes choose to limit the number of times certain messages are shared with other users. But the same survey found that only 28 percent favored government regulation of online content, with 46 percent opposed to the idea.

“They don’t want their government actually taking measures,” said Chris Chao Su, an assistant professor of emerging media studies at BU who designed the survey. “They do want some measures to be taken, not by the government but by the platform.”

In March, the US Supreme Court held arguments over an injunction from a federal court in Louisiana which would sharply limit the federal government’s ability to ask social media companies to take down questionable postings. The lower court ruled that the government was attempting to coerce the companies into engaging in censorship that would violate the US constitution if the government did it directly. But the Biden administration has challenged the injunction, saying that the government has a legitimate interest in working with social media companies to limit the spread of false information, and that the government did not cross the line into coercion. A ruling in the case is expected this month.

Last week, in a case that might hint at how the federal court will rule in the Louisiana case, the justices agreed unanimously that the National Rifle Association could sue Maria Vullo, the former head of New York’s Department of Financial Services, after she urged banks not to do business with the NRA. Writing for the court, Justice Sonia Sotomayor said that “the First Amendment prohibits government officials from wielding their power selectively to punish or suppress speech, directly or ... through private intermediaries.”

Whatever the outcome of the Louisiana case, Su said, social media companies themselves need to attack misinformation in a more sophisticated way. Already, Facebook and X take down messages that could be harmful, such as incitements to violence, stalking attempts, or voter suppression efforts. Critics of the companies say they could do much more.

But rather than trying to weed out all misinformation, Su thinks social networks should take a more selective approach.

“It’s not like all misinformation is equal,” Su said. For instance, some research studies have found that false news stories are mostly ignored. “In most cases,” said Su, “the fake news or misinformation has very minimal effect on the general population.”

But if a piece of false information is picked up by politicians, celebrities, or online influencers, it can have a lot more impact. Social networks can proactively counter this, said Su. For instance, a social network might be more aggressive in limiting access to anti-vaccine messages posted by a TV star, while ignoring similar messages posted by a truck driver.

Government agencies can’t track fake news with this level of precision — but the social networks can. “That’s why I think it’s more important for the platforms to take action than the government,” Su said.

Read the Boston Globe Article

Share this:

• Share on Facebook (Opens in new window)
• Click to share on LinkedIn (Opens in new window)
• Click to share on Twitter (Opens in new window)

View all posts

• Policy Guidance
• Submitting Data
• Accessing Data
• About the Genomic Data Sharing (GDS) Policy
• Key Documents
• Preparing Genomic Data
• Extramural Grantees
• Non-NCI Funded Investigators
• Intramural Investigators
• Accessing Genomic Data
• Genomic Data Sharing Policy Contact Information
• Data Standards Services
• Cloud Resources
• Data Commons Framework
• Repositories
• ARPA-H BDF Toolbox
• Accelerating Therapeutics for Opportunities in Medicine (ATOM)
• NCI-DOE Collaboration AI/ML Resources
• NCI-DOE Publications
• U.S.-EU Artificial Intelligence Administrative Arrangement
• NCI Data Catalog
• CDISC Terminology
• FDA Terminology
• NCPDP Terminology
• Pediatric Terminology
• Metadata for Cancer Research
• Informatics Technology for Cancer Research (ITCR) Tools
• Generating and Collecting Data
• Cleaning Data
• Exploring and Analyzing Data
• Predictive Modeling
• Visualizing Data
• Sharing Data
• Cancer Data Science Course
• Training Guide Library
• Cancer Data Science Pulse Blog
• Data Science Seminar Series
• Jobs and Fellowships
• Contact CBIIT
• Organization
• CBIIT Director
• Staff Directory
• Application Support
• Genomic Data Sharing
• NCI-Department of Energy Collaborations
• Cancer Vocabulary
• Learn About Cancer Data Science
• Improve My Data Science Skills

Submit Your Project Pitch to the 2024 Human Tumor Atlas Network (HTAN) Data Jamboree

Do you jamboree? No banjo needed! Just bring your ideas for reusing data, especially spatial omics and single-cell sequence data.

Submit your short project pitch to the 2024 HTAN Data Jamboree by Friday, July 12!

This in-person event, taking place on November 6–8, 2024, and hosted on the NIH campus in Bethesda, MD, will bring together scientists and coders to create innovative solutions to cancer research problems.

Projects involving data interoperability , such as those that bridge public data with HTAN data , are of particular interest. Organizers also welcome projects that produce tutorial pipelines and educational tools.

Projects could include: 

• building analysis pipelines,
• utilizing visualization techniques,
• developing artificial intelligence/machine learning algorithms,
• employing statistical methods, or
• using existing computational, mathematical, or informatics tools to address cancer-focused questions.

As a participant, you will have access to NCI Cloud Resources , providing an environment to work with HTAN data or combine it with other public data sets , including those created by fellow Jamboree participants.

Don’t miss this opportunity to collaborate, innovate, and make a difference in cancer research! Learn more and submit your project ideas on the HTAN Data Jamboree website .

SUBSCRIBE TO UPDATES

Help us help you: take this 5-minute survey.

We want to know what you want from this website and our weekly email. Your feedback will help us ensure we’re getting you the information you want, how you want it.

Map Options

Education Rankings by Country 2024

There is a correlation between a country's educational system quality and its economic status, with developed nations offering higher quality education.

The U.S., despite ranking high in educational system surveys, falls behind in math and science scores compared to many other countries.

Educational system adequacy varies globally, with some countries struggling due to internal conflicts, economic challenges, or underfunded programs.

While education levels vary from country to country, there is a clear correlation between the quality of a country's educational system and its general economic status and overall well-being. In general, developing nations tend to offer their citizens a higher quality of education than the least developed nations do, and fully developed nations offer the best quality of education of all. Education is clearly a vital contributor to any country's overall health.

According to the Global Partnership for Education , education is considered to be a human right and plays a crucial role in human, social, and economic development . Education promotes gender equality, fosters peace, and increases a person's chances of having more and better life and career opportunities.

"Education is the most powerful weapon which you can use to change the world." — Nelson Mandela

The annual Best Countries Report , conducted by US News and World Report, BAV Group, and the Wharton School of the University of Pennsylvania , reserves an entire section for education. The report surveys thousands of people across 78 countries, then ranks those countries based upon the survey's responses. The education portion of the survey compiles scores from three equally-weighted attributes: a well-developed public education system, would consider attending university there, and provides top-quality education. As of 2023, the top ten countries based on education rankings are:

Countries with the Best Educational Systems - 2021 Best Countries Report*

Ironically, despite the United States having the best-surveyed education system on the globe, U.S students consistently score lower in math and science than students from many other countries. According to a Business Insider report in 2018, the U.S. ranked 38th in math scores and 24th in science. Discussions about why the United States' education rankings have fallen by international standards over the past three decades frequently point out that government spending on education has failed to keep up with inflation.

It's also worthwhile to note that while the Best Countries study is certainly respectable, other studies use different methodologies or emphasize different criteria, which often leads to different results. For example, the Global Citizens for Human Rights' annual study measures ten levels of education from early childhood enrollment rates to adult literacy. Its final 2020 rankings look a bit different:

Education Rates of Children Around the World

Most findings and ranking regarding education worldwide involve adult literacy rates and levels of education completed. However, some studies look at current students and their abilities in different subjects.

One of the most-reviewed studies regarding education around the world involved 470,000 fifteen-year-old students. Each student was administered tests in math, science, and reading similar to the SAT or ACT exams (standardized tests used for college admissions in the U.S.) These exam scores were later compiled to determine each country's average score for each of the three subjects. Based on this study, China received the highest scores , followed by Korea, Finland , Hong Kong , Singapore , Canada , New Zealand , Japan , Australia and the Netherlands .

On the down side, there are many nations whose educational systems are considered inadequate. This could be due to internal conflict, economic problems, or underfunded programs. The United Nations Educational, Scientific, and Cultural Organization's Education for All Global Monitoring Report ranks the following countries as having the world's worst educational systems:

Countries with the Lowest Adult Literacy Rates

• Education rankings are sourced from both the annual UN News Best Countries report and the nonprofit organization World Top 20

Download Table Data

Enter your email below, and you'll receive this table's data in your inbox momentarily.

Which country ranks first in education?

Which country ranks last in education, frequently asked questions.

• Best Countries for Education - 2023 - US News
• Literacy rate, adult total (% of people ages 15 and above) - World Bank
• World Best Education Systems - Global Citizens for Human Rights
• UNESCO - Global Education Monitoring Reports
• World’s 10 Worst Countries for Education - Global Citizen
• International Education Database - World Top 20

Search form

• Prospective Students

Programs of Study

• Graduate Studies at UM
• Student Resources
• MPSDS Informational Webinar

The University of Michigan Program in Survey and Data Science offers programs of study at the doctoral, master's and certificate levels. The PhD and MS programs prepare students for careers in private and academic survey research firms, government agencies, and corporations. The certificate program is designed to provide students with specialized knowledge in survey methodology to enhance skills in current positions and to expand career opportunities. Non-Candidate for Degree status is an attractive option for professionals seeking to increase their skills or continue graduate studies without completing a full degree program and for those who wish to test their capabilities before pursuing a graduate degree. The links below provide detailed information about each program.

• Master of Science in Survey and Data Science
• Doctor of Philosophy in Survey and Data Science
• Certificate in Survey and Data Science
• Departmental Non-Candidate for Degree
• New option  Graduate Data Science Certificate Program

The University of Michigan Program in Survey and Data Science is a STEM program.  STEM is an approach to learning and development that integrates the areas of science, technology, engineering and mathematics. Through STEM, students develop key skills including: problem solving. creativity .

For information about   application procedures   and   deadlines , please see the   admissions   portion of this site. To request additional information, please contact the Program in Survey and Data Science staff.

• Google Scholar
• Program Overview
• Our Faculty
• Awards & Achievements
• JPSM PhD Alumni
• Online Programs
• Onsite Programs
• Non-Degree Seeking Student
• Professional Development
• Online Courses
• Onsite Courses
• All Courses
• The Global COVID-19 Trends and Impact Survey
• Improving the Measurement of Economic Activity
• Improving the Design of Mobile Surveys
• Improving the Analysis of Complex Data in Social & Bio-Medical Settings
• Improving Statistical Methodology with Multiple Complex Big Databases
• Distinguished Lecture Archive
• General Information for Graduates
• General Program Information
• Graduate Student Rights and Responsibilities

JPSM Short Course 2024-2025 Schedule

The Joint Program in Survey Methodology offers a variety of short courses. Be sure to bookmark this new website address. Short Courses are open to the public and admission through the University of Maryland admission office is not required.

Short Courses for 2024-2025 will be delivered online through a combination of synchronous and asynchronous instruction, with the exception of 'Introduction to Survey Sampling '.

• Synchronous: instructor meets with all students at a specified day and time, which will be reflected on your schedule and also in the Schedule of Classes. (Live Sessions)
• Asynchronous: instructor provides content for students to review on their own (Hybrid)

REGISTRATION FOR ALL COURSES WILL CLOSE  5 BUSINESS DAYS PRIOR TO THE CLASS START DATE.

JPSM flat rate is $700.00 per course.

On-Site Class fee is $900: Introduction to Survey Sampling.

(*) These Courses are part of the AAPOR/JPSM Citation Program.

The New JPSM Short Courses Flat Rate is:

Fees and awards are not transferable due to nonattendance. 

Online registration is required. The registration deadline is 5 business days prior to the class start date. Confirmation of acceptance will be sent after the registration form and payment have been processed. Payment by credit card is required. Payment is due at the time of registration. Payments are course and date specific. Please note – if payment is not received at the time of registration, the registrant’s form will be deleted. Registrants are responsible for keeping track of their registrations and course dates. Contact JPSM if you have any questions concerning the status of the registration.

Tax Identification Number (University of Maryland):  55-6002003,  DUNS (University of Maryland):  808124564

Please notify JPSM as soon as possible if you need to cancel your registration. Fees and awards are not transferable due to nonattendance. Cancellation requests should be sent to jpsm-shortcourse.umd.edu. You will be fully reimbursed if you cancel prior to 5 business days before the start of class. Cancellation on or less than 5 business days before the class is not reimbursable but we may transfer your registration to another short course.

Questions for this course should be directed to the JPSM Short Course, University of Maryland, 1218 Lefrak Hall, College Park, MD 20742.

Phone: 301-314-7911

Fax: 301-314- 7912

Email: jpsm-shortcourse [at] umd [dot] edu

Login / Logout

• Mayo Clinic Careers
• Anesthesiology
• Dermatology
• Emergency Medicine
• Family Medicine
• Internal Medicine
• Lung Transplant
• Psychiatry & Psychology
• Nurse Practitioner & Physician Assistant
• Ambulance Service
• Clinical Labs
• Med Surg RN
• Radiology Imaging
• Clinical Research Coordinator
• Respiratory Care
• Senior Care
• Surgical Services
• Travel Surgical Tech
• Practice Operations
• Administrative Fellowship Program
• Administrative Internship Program
• Career Exploration
• Nurse Residency and Training Program
• Nursing Intern/Extern Programs
• Residencies & Fellowships (Allied Health)
• Residencies & Fellowships (Medical)
• SkillBridge Internship Program
• Training Programs & Internships
• Diversity, Equity & Inclusion
• Employees with Disabilities

You're using Internet Explorer - therefore, some pages or features may not display properly. We recommend switching to a modern browser such as Chrome, Microsoft Edge, or Firefox for a smoother experience.

Search life-changing careers..

Search by Role or Keyword

Enter Location

• United States Applicants
• United Kingdom Applicants
• United Arab Emirate Applicants
• Current Employees

Intern- Data Science AI&I (PHD)

• Jacksonville, FL

Not ready to apply? Join our talent community

Under general supervision and guidance: - The analyst/programmer is responsible for the support of moderately complex software-based systems in the supported areas. - Contributes to the design, development, implementation, and maintenance of custom software, or the installation and maintenance of purchased software systems. - Produces documentation such as systems requirements, designs, and plans as requested by the work unit leadership. - Work will be reviewed for quality, timeliness, and adequacy at predetermined milestones.

Please note that visa sponsorship is not avaliable for this role. 

Working towards a PhD in Computer Science or related Informatics field with a GPA of 3.0 or greater.Experience with Java, SQL, and XML.

About our location

Jacksonville, Florida

We would love to connect with you.

Click the button for a list of our upcoming events.

Join Our Talent Community

Sign up, stay connected and get opportunities that match your skills sent right to your inbox

Email Address

Phone Number

Upload Resume/CV (Must be under 1MB) Remove

Job Category* Select One Advanced Practice Providers Business Education Engineering Executive Facilities Support Global Security Housekeeping Information Technology Internship Laboratory Nursing Office Support Patient Care - Other Pharmacy Phlebotomy Physician Post Doctoral Radiology Imaging Research Scientist Surgical Services Therapy

Location Select Location Albert Lea, Minnesota Arcadia, Wisconsin Austin, Minnesota Barron, Wisconsin Bloomer, Wisconsin Caledonia, Minnesota Cannon Falls, Minnesota Chandler, Arizona Chippewa Falls, Wisconsin Decorah, Iowa Duluth, Minnesota Eau Claire, Wisconsin Fairmont, Minnesota Holmen, Wisconsin Jacksonville, Florida La Crosse, Wisconsin Lake City, Minnesota London, England Mankato, Minnesota Menomonie, Wisconsin Minneapolis-St. Paul-Bloomington, Minnesota New Prague, Minnesota Onalaska, Wisconsin Osseo, Wisconsin Owatonna, Minnesota Phoenix, Arizona Prairie du Chien, Wisconsin Red Wing, Minnesota Rice Lake, Wisconsin Rochester, Minnesota Saint Cloud, Minnesota Saint James, Minnesota Scottsdale, Arizona Sparta, Wisconsin Tempe, Arizona Tomah, Wisconsin Waseca, Minnesota Zumbrota, Minnesota

Area of Interest Select One Nursing Research Laboratory Medicine & Pathology Radiology Surgery Physical Medicine & Rehabilitation Finance Licensed Practical Nurse (LPN) Cardiovascular Medicine Facilities Neurology Psychiatry & Psychology Pharmacy Emergency Medicine Surgical Technician Respiratory Therapy Ambulance Services General Services Environmental Services Radiation Oncology Anesthesiology & Perioperative Medicine Social Work Mayo Collaborative Services Orthopedics Hematology Hospital Internal Medicine Mayo Clinic Laboratories Administration Gastroenterology & Hepatology Medical Oncology Cardiovascular Surgery Housekeeping Obstetrics & Gynecology Family Medicine General Internal Medicine Patient Scheduling International Pediatrics Transplant Desk Operations Information Technology Oncology Senior Care Ophthalmology Artificial Intelligence & Informatics Critical Care Dermatology Engineering Global Security Healthcare Technology Management Linen & Central Services Nephrology & Hypertension Community Internal Medicine Hospice & Palliative Care Office Support Health Information Management Services Neurosciences Rheumatology Biochemistry & Molecular Biology Cancer Center Clinical Genomics Education Otolaryngology (ENT) Pulmonary/Sleep Medicine Sports Medicine Endocrinology Health Care Delivery Research Physiology & Biomedical Engineering Regenerative Biotherapeutics Urology Business Development Digital Epidemiology Infectious Diseases Legal Molecular Medicine Neurologic Surgery Quality Surgical Assistant Travel Communications Comparative Medicine Development/Philanthropy Human Resources Immunology Information Security Mayo Clinic Platform Molecular Pharmacology & Experimental Therapeutics Pain Medicine Primary Care Spiritual Care Allergic Diseases Bariatric Medicine Cancer Biology Center for Individualized Medicine Clinical Trials & Biostatistics Dental Specialities Geriatric Medicine & Gerontology Marketing Occupational/Preventative Medicine Spine Center Urgent Care Volunteer Services

• Research, Jacksonville, Florida, United States Remove

Confirm Email

By submitting your information, you consent to receive email communication from Mayo Clinic.

Join our talent community.

Join our global talent community to receive alerts when new life-changing opportunities become available.

If you want to know what it's really like at Mayo Clinic, just ask. You'll find that our pride–in where we work, and in what we do–is a common trait. You will also find a lot of inspiring stories about lives changed for the better.

Nurse Residency Program

The Nurse Residency Program (NRP) is for all nurses with less than 12 months of experience, to be completed within the first year. NRP provides a framework for a successful transtion to a professional nurse by promoting educational and personal advancement.

As your career evolves, our compensation and benefits packages are designed to change with you — meeting needs now, and anticipating what comes next. We know that when Mayo Clinic takes care of you, you can take better care of our patients.

Equal opportunity

All qualified applicants will receive consideration for employment without regard to race, color, religion, sex, gender identity, sexual orientation, national origin, protected veteran status, or disability status. Learn more about "EEO is the Law." Mayo Clinic participates in E-Verify and may provide the Social Security Administration and, if necessary, the Department of Homeland Security with information from each new employee's Form I-9 to confirm work authorization.

Reasonable accommodations

Mayo Clinic provides reasonable accommodations to individuals with disabilities to increase opportunities and eliminate barriers to employment.  If you need a reasonable accommodation in the application process; to access job postings, to apply for a job, for a job interview, for pre-employment testing, or with the onboarding process, please contact HR Connect at 507-266-0440 or  888-266-0440.

Job offers are contingent upon successful completion of a post offer placement assessment including a urine drug screen, immunization review and tuberculin (TB) skin testing, if applicable.

Recruitment Fraud

Learn more about recruitment fraud and job scams

Advertising

Mayo Clinic is a not-for-profit organization and proceeds from Web advertising help support our mission. Mayo Clinic does not endorse any of the third party products and services advertised.

Advertising and sponsorship policy | Advertising and sponsorship opportunities

Reprint permissions

A single copy of these materials may be reprinted for noncommercial personal use only. "Mayo," "Mayo Clinic," "MayoClinic.org," "Mayo Clinic Healthy Living," and the triple-shield Mayo Clinic logo are trademarks of Mayo Foundation for Medical Education and Research.

Any use of this site constitutes your agreement to the Terms and Conditions and Privacy Policy linked below.

Terms and Conditions | Privacy Policy | Notice of Privacy Practices | Notice of Nondiscrimination

© 1998-2024 Mayo Foundation for Medical Education and Research (MFMER). All rights reserved.

Deep Learning and the Brain

Clip, llava, and the brain, insights into multimodal transformers from neuroscience.

Jonathan R. Williford, PhD

Towards Data Science

How do recent multimodal transformer networks, like CLIP (Radford et al. 2021) and LLaVA (Liu et al. 2023), compare to the brain? Are there similarities between the attention in these networks and the brain? In this article, I look at these transformer architectures with an eye on the similarities and differences with the mammalian brain.

What stood out to me was that vision transformers, CLIP, and LLaVA perform a type of processing analogous to pre-attentive visual processing in the brain. This processing is done in the initial feedforward visual responses to a stimulus before recurrence. Although a lot can be accomplished in a feedforward way, studies have shown that feedforward pre-attentive processing in the brain does have difficulty with:

• Distinguishing the identity or characteristics of similar types of objects, especially when objects are close together or cluttered or the objects are unnatural or artificial (VanRullen 2007).
• More complex tasks such as counting or maze or curve tracing tasks.
• Perceiving objects that are more difficult to see, such as where it is difficult to perceive the boundaries of the objects.

In contrast to the feed-forward processing, one of the things that stands out with the brain is the richness in the interaction of areas, which I will discuss in more detail in the next section.

Bidirectional Activity in the Brain

In most current deep learning architectures, activity is propagated in a single direction, for example, an image might be given as input to a network and then propagated from layer to layer until you get to a classification as the output.

The brain is much more interesting than these feedforward models. In the visual system, a stimulus will initially propagate from lower- to higher-level visual areas in a feedforward fashion, then the higher-level areas will exert influence over the lower-level areas as depicted in Figure 1.

Some of this feedback is the conscious top-down attention that allows us to allocate more resources to objects and features of interest and disambiguate stimuli that are either complex or ambiguous. Another part of this feedback is automatic and allows higher-level areas to infuse the lower-level areas with information that would not be known in just the feedforward manner.

Conscious top-down attention is thought to support consciousness of visual stimuli. Without conscious access to lower-level areas that encode borders and edges, we wouldn’t have as spatially precise a perception of borders. Tasks like mentally tracing a curve or solving a maze would be impossible.

One example of automatic unconscious feedback is border-ownership coding which is seen in about half of the orientation-selective neurons in visual area V2 (Zhou et al. 2000, Williford and von der Heydt 2013). These neurons will encode local information in about 40 ms and, as early as 10 ms after this initial response, will incorporate global context to resolve occlusions — holding the information about which objects are creating borders by occluding their backgrounds.

Another example of this unconscious feedback was shown by Poort et al. (2012) using images like that in Figure 2. In the Macaque early visual cortex V1, neurons will tend to initially (within 50–75 ms of stimulus presentation) encode only the local features within their receptive fields (e.g., green square). However, after around 75 ms, they will receive feedback from the higher-level areas and tend to have a higher response when that texture belongs to a figure, such as this texture-defined figure above. This happens even when attention is drawn away from the figure, however, if the monkey is paying attention to the figure the neurons will on average respond even more.

One way to look at this bidirectional interaction is that each neuron greedily uses all available predictive signals constantly. Even higher-level areas can be predictive, especially when visual borders do not correspond to significant first-order contrast edges.

Transformers

With all the talk about attention with the introduction of transformers (Vaswani et al. 2017) and with the ability to generate sentences one word at a time, you might be led to believe that transformers are recurrent. However, there are no internal states kept between the steps of the transformer, only the previous output is provided as input. So, the recurrence is limited and does not have the bidirectionality that is ubiquitous in the brain. Transformers do have multi-headed attention, which is like being able to attend to a fixed number of things simultaneously (8 in the original paper). Hence, image transformers can be seen as analogous to pre-attentive feedforward processing with some modifications.

Radford and colleagues from OpenAI introduced CLIP in their 2021 paper “Learning Transferable Visual Models from Natural Language Supervision”. The idea behind CLIP is simple and is shown in Figure 3. It takes a bunch of image and caption pairs from the Internet and feeds the image to an image encoder and the text to a text encoder. It then uses a loss that brings the encoding of the image and the encoding of the text closer together when they are in the same pair, otherwise the loss increases the distance of the encodings. This is what CLIP gives you: the ability to compare the similarity between text and images. This does allow it to be used for zero-shot classification, as shown in Figure 4. CLIP does not, by itself, generate text descriptions from images.

The image encoder and text encoder are independent, meaning there is no way for task-driven modulation to influence the image encoding. This means that the image encoder must encode everything that could be potentially relevant to the task. Typically, the resolution of the input image is small, which helps prevent the computation and memory requirements from exploding.

Large Language and Vision Assistant (LLaVA) (Liu et al. 2023) is a large language and vision architecture that extends and builds onto CLIP to add the ability to describe and answer questions about images. This type of architecture interests me because it can attempt tasks like those used in Neuroscience and Psychology.

LLaVA takes the vision transformer model ViT-L/14 trained by CLIP for image encoding (Figure 5). The first paper uses a single linear projection matrix W to convert the encodings into tokens. The tokens calculated from the images Hᵥ and the text instructions Hq are provided as input. LLaVA can then generate the language response Xₐ one token at a time, appending the response so far as the input to the next iteration.

I won’t go into the details of how LLaVA is trained, but it is interesting how they use ChatGPT to expand the caption (Xc) in Figure 5 to form instructions (Hq) and responses (used to train Xₐ) about an image and the use of bounding box information.

In version 1.5 of LLaVA (Liu et al. 2024), some of the improvements they made include:

• The linear projection matrix W is replaced with a multilayer perceptron
• The image resolution is increased by using an image encoder that takes images of size 336x336 pixels and splits the images into grids that are encoded separately

Task-driven attention in the brain can dynamically allocate resources to the object, location, or features of interest, which allows the processing of information that would otherwise be overwhelmed by clutter or other objects. In LLaVA, the image encoder is independent of the text instructions, so to be successful it needs to make sure any potentially useful information is stored in the image tokens (Hᵥ).

LLaVA and CLIP lack bidirectional and recurrence with internal states, which constrains their processing. This is especially true for image processing since image processing is done independently of the text instructions. Most convolutional neural networks also share these limitations. This leads me to my conjecture:

Conjecture: Most convolutional, vision transformer, and multimodal transformer networks are restricted to processing that is analogous to pre-attentive feedforward visual processing in the brain.

This is not a criticism as much as an insight that can be informative. Feedforward processing can do a lot and is fast. However, it is not as dynamic as to what resources can be used to be used, which can lead to informational bottlenecks in cluttered scenes and is unable to encode enough information for complex tasks without an explosion of the size of the encodings. Creating models that work in a feedforward fashion is an important stepping stone because of the difficulty of adding recurrence and bidirectional processing.

Some networks are not limited to pre-attentive feedforward networks, but currently, most of the architectures lag behind those of transformers. These include long-short term memory models (LSTMs) and, more recently, the Mamba architecture, which has several benefits over transformers ( Gu and Dao 2024 ). Extended LSTMs (Beck et al. 2024 , Alkin et al. 2024 ) have recently been proposed, which help close the gap between transformers and LSTMs. Diffusion models also have a limited type of recurrence that uses the image as the state between iterations.

B. Alkin, M. Beck, K. Pöppel, S. Hochreiter, and J. Brandstetter, Vision-LSTM: xLSTM as Generic Vision Backbone (2024), http://arxiv.org/abs/2406.04303 .

M. Beck, K. Pöppel, M. Spanring, A. Auer, O. Prudnikova, M. Kopp, G. Klambauer, J. Brandstetter, and S. Hochreiter, xLSTM: Extended Long Short-Term Memory (2024), http://arxiv.org/abs/2405.04517

A. Gu and T. Dao. Mamba: Linear-Time Sequence Modeling with Selective State Spaces (2024) http://arxiv.org/abs/2312.00752

H. Liu, C. Li, Y. Li, and Y. J. Lee “ Improved Baselines with Visual Instruction Tuning (2024) Proc. of IEEE/CVF CVPR .

H. Liu, C. Li, Q. Wu, and Y. J. Lee, Visual Instruction Tuning (2023), https://doi.org/10.48550/arXiv.2304.08485

J. Poort, F. Raudies, A. Wannig, V. A. F. Lamme, H. Neumann, and P. R. Roelfsema. The Role of Attention in Figure-Ground Segregation in Areas V1 and V4 of the Visual Cortex (2012) Neuron

A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, and J. Clark. Learning Transferable Visual Models from Natural Language Supervision (2021) ICML

R. VanRullen, The Power of the Feed-Forward Sweep (2007) Advances in Cognitive Psychology

A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, Attention Is All You Need (2017) NeurIPs

J. R. Williford and R. von der Heydt, Border-Ownership Coding (2013) Scholarpedia

H. Zhou, H. S. Friedman, and R. von der Heydt. “ Coding of Border Ownership in Monkey Visual Cortex (2000) The Journal of Neuroscience

Originally published at http://neural.vision on June 19, 2024.

Written by Jonathan R. Williford, PhD

Text to speech