Biostatistician Job Description

A biostatistician answers one question for a study team: will this design and this analysis give an answer people can trust? The job sits between the scientists who ask the clinical question and the data that will answer it. In drug development that means sitting at the protocol table long before the first patient is enrolled, and staying until the last regulatory question is closed.

Before the data exist

The decisions that shape a study come early. The biostatistician picks the primary endpoint's analysis, calculates how many patients are needed to detect a clinically meaningful difference, and sets up randomization and stratification so known prognostic factors are balanced across arms. They define the estimand, which pins down what to do when a patient stops treatment, switches to rescue medication or dies before the endpoint is measured. O*NET's task list for Biostatisticians (15-2041.01) includes calculating sample size requirements for clinical studies and writing detailed analysis plans for research protocols or reports.

The statistical analysis plan is the contract. It names every test, the order of hypotheses for multiplicity control, the handling of missing data and the sensitivity analyses that will stress the main result. It is locked before unblinding so nobody can choose the flattering analysis after seeing the numbers.

While the trial runs

During the study the biostatistician works with data management on edit checks and with programmers on analysis dataset specifications. For trials with interim looks, they set the stopping boundaries and may support an independent data monitoring committee through a separate, unblinded statistician so the sponsor team stays blind.

After database lock

Then comes the production work: efficacy and safety tables, subgroup forest plots, survival curves, and the statistical sections of the clinical study report. O*NET lists preparing tables and graphs to present clinical data or results among the core tasks, along with developing models to track biological phenomena such as the spread of infectious diseases, work that is common in public health roles.

The reviewer across the table

FDA has statisticians on the other side of the table. The Center for Drug Evaluation and Research's Office of Biostatistics describes its work as including review of proposed studies intended to provide regulatory evidence, checking the alignment of the study design with the clinical and regulatory questions, the reliability and relevance of endpoints, and the appropriateness and prespecification of the analysis. A sponsor biostatistician's job is, in part, to anticipate each of those questions and answer it in the plan before a reviewer has to ask.

Outside industry the same skills apply to cohort studies, disease surveillance, hospital quality data and grant-funded research, with the analysis plan going to a journal or a funder instead of a regulator.

Education

Biostatistics is mostly a graduate-degree field. ONET rates Biostatisticians at Job Zone Five, extensive preparation needed. Among respondents to ONET's survey for the occupation, 58% said a master's degree is required, 29% said a doctoral degree is required and 13% said a bachelor's degree is required.

As a rough guide, a master's in biostatistics or statistics is the common starting credential for trial analysis and programming roles, while a doctorate fits methods development, academic faculty posts and the most senior design roles. Coursework that pays off on the job includes linear and generalized linear models, survival analysis, longitudinal and mixed models, causal inference, Bayesian inference and the design of experiments.

Technical skills

O*NET's technology list for the occupation includes SAS and R, along with Structured Query Language (SQL) and Bash. In regulated trial work the practical skills are narrower and deeper:

• Writing reproducible SAS or R code and validating a colleague's output by independent double programming
• Reading and specifying CDISC SDTM and ADaM datasets, the formats submissions are built on
• Group sequential boundaries, alpha spending and graphical multiplicity procedures
• Multiple imputation, tipping-point analyses and reference-based imputation for missing data
• Simulation of design operating characteristics, which adaptive and Bayesian designs often call for

Domain knowledge

Clinical teams speak in endpoints, so learn the endpoints of the therapeutic area you work in: response criteria in oncology, rating scales in neuroscience, exacerbation definitions in respiratory disease. Know the ICH guidelines that govern trial statistics and how a protocol, analysis plan and clinical study report fit together. Public health and academic roles add their own vocabulary: incidence and prevalence estimation, survey weighting, and the reporting checklists journals use for observational studies.

Credentials

No license is required to practice. The American Statistical Association awards two voluntary credentials: GStat (Graduate Statistician) and PStat (Accredited Professional Statistician). Check the ASA's own accreditation pages for current eligibility rules before applying.

Experience path

As a rough guide, new master's graduates often start as a statistician or statistical programmer supporting one study, take on analysis plan authorship within a few years, and move to study or program lead after roughly 5 to 8 years. Internships at sponsors, contract research organizations or academic coordinating centers can help, since they expose students to the documentation habits regulated work demands.

BLS projects employment of statisticians to grow 11 percent from 2025 to 2035, from 31,300 jobs to 34,700. For mathematicians and statisticians combined, it expects about 2,000 openings a year on average over the decade, and says many of those openings are expected to come from the need to replace workers who transfer to different occupations or exit the labor force. BLS does not project biostatisticians separately, so these numbers cover statisticians in every field.

Where the jobs and the pay sit

BLS puts the federal government first among the largest employers of statisticians, with 16% of the occupation's jobs in 2025, followed by educational services and by research and development in the physical, engineering and life sciences, at 15% each.

Pay varies by setting. In the May 2025 OEWS industry estimates, statisticians in Scientific Research and Development Services had a median annual wage of $130,310, with 5,080 employed. The median in the Federal Executive Branch was $132,620, and in Pharmaceutical and Medicine Manufacturing it was $140,680. In Colleges, Universities, and Professional Schools the median was $81,830. Those are statisticians of every kind in each industry, not biostatisticians alone.

The regulatory toolbox is changing

Two FDA drafts matter to anyone designing trials. On Jan. 12, 2026, FDA published draft guidance, Use of Bayesian Methodology in Clinical Trials of Drug and Biological Products, to meet a commitment under the Prescription Drug User Fee Act (PDUFA) VII. FDA's announcement lists uses including determining futility or success earlier in adaptive trials, incorporating prior clinical study data, real-world evidence and external or nonconcurrent controls, and supporting primary inference in a trial. It says Bayesian methods may be especially valuable for rare or pediatric indications. FDA's guidance page still lists it as a draft as of November 2026, and CDER's Small Business and Industry Assistance program held a webinar on it on Oct. 14, 2026.

The second is ICH E20, Adaptive Designs for Clinical Trials, endorsed as a draft for public consultation on June 25, 2025, and issued by FDA as draft guidance in September 2025; FDA still lists it as a draft. It says a structured report of design simulations should be included in regulatory submissions before the trial is conducted. Design simulation and the choice of prior distributions, a topic of FDA's webinar, are skills worth building for anyone heading into trial design.

AI and the statistician

The American Statistical Association's Statement on the Role of Statistics in Data Science and Artificial Intelligence, dated Aug. 4, 2023, says statistical education must keep evolving and that the next generation of statistical professionals needs a broader skill set and must be more able to engage with software and machine learning engineering experts. FDA's Office of Biostatistics lists artificial intelligence and machine learning among the programs it supports with statistical leadership. The practical reading for job seekers: whatever tool writes the code, the design judgment, the prespecified plan and the accountability for a submitted result rest with the statistician.

Dear Hiring Manager,

I am applying for the Biostatistician position on your clinical statistics team. I hold an M.S. in biostatistics and have spent three years as a statistician at a mid-sized contract research organization, supporting Phase II and Phase III studies in respiratory and metabolic disease.

My current role covers the full study cycle. I wrote the statistical analysis plan for a Phase III asthma trial with an annualized exacerbation rate primary endpoint, specified negative binomial models with an offset for exposure time, and set a hierarchical testing order for four key secondary endpoints. When the sponsor's clinical lead wanted to add a responder analysis late in the study, I documented it as a post hoc exploratory analysis rather than folding it into the prespecified family, and explained why in a short memo the team later reused for other programs.

I program in both SAS and R. I independently double-programmed the primary efficacy tables for two database locks, and I built an R simulation tool our group now uses to check power for recurrent-event endpoints under different dropout assumptions.

What draws me to your team is your work on adaptive dose-finding designs. I have taken graduate coursework in Bayesian methods and would welcome the chance to apply it to a program where interim decisions shape the next study. I also value the way your group pairs statisticians with clinical scientists from protocol drafting onward, which matches how I like to work.

Thank you for your consideration. I would be glad to walk through a sample analysis plan or simulation report in an interview.

Sincerely,

Jordan Ellis