RESEARCH DISCIPLINE
Biostatistical Consulting for Research
Statistical Design & Analysis for Health Studies
Biostatistical support for health and clinical science, from the estimand and sample size to the analysis plan, prediction models and reporting, with the question, design and model matched to one another before any result is seen, so that the study can be defended in review.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
Biostatistical Consulting for Research
Biostatistical consulting for research is the statistical design, analysis and reporting of health and clinical studies treated as a methodological discipline: defining the estimand, fixing the design and sample size, and prespecifying the analysis in a plan before results are seen. Studies fail in review less often for lack of software output than for a mismatch between the question, the design and the model, or because the sample size, the missing-data handling and the sensitivity analyses were never planned. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, provides biostatistics consulting through in-house PhD-level statisticians, who work with your team on the plan, the analysis and the reporting, and settle the estimand and the effective sample size before any model is fitted.
WHEN TO ENGAGE
Statistical Support at Different Stages of Biostatistical Research
Research teams usually reach out at a recognizable point, from the first draft of a protocol to a reviewer's objection to a model that does not fit the design.
Designing a Study or Grant Application You are preparing a protocol or funding application and need the estimand, a defensible sample size and a statistical section that a review panel will accept.
Drafting the Analysis Plan The design is settled, and the plan has to specify the primary model, the handling of missing data, the sensitivity analyses and the effect estimates before any results are seen.
Developing a Prediction Model You want a model that predicts a diagnosis or an outcome, and need the sample size, predictor selection, validation and calibration planned so that it does not overfit.
Diagnosing a Design-Model Mismatch An analysis has been run, but the model does not match the design, for example clustered data treated as independent or an estimate that does not answer the stated question.
RESEARCH CONTEXT
Research Designs and Data We Work With in Biostatistics
Biostatistics data analysis starts before the dataset exists as often as after it, so we work with protocols and pilot data as readily as with finished study files.
01
Protocols and Grant Applications
Study protocols at the planning stage, with pilot data or published estimates to size the study and an outcome that is still being defined.
02
Pilot and Feasibility Data
Small preliminary studies whose estimates of variability and event rates have to be used cautiously when sizing a larger study.
03
Prediction and Prognostic Cohorts
Cohorts with many candidate predictors and few outcome events, from which a diagnostic or prognostic model is to be built and validated.
04
Biomarker and Method-Comparison Data
Laboratory or device measurements compared with a reference method, where agreement and the choice of a cut-off need dedicated analysis.
05
Clustered and Multicenter Data
Patients within centers, clinics or families, where the number of clusters, not only the number of patients, limits what a model can estimate.
06
Incomplete Study Files
Datasets in which outcomes or covariates are missing for reasons that call for an explicit assumption and a planned sensitivity analysis.
METHODOLOGICAL CONSIDERATIONS
Statistical Issues We Consider in Biostatistical Research
These checks are made when a plan is written and again when results arrive, because decisions fixed in advance are the ones a reviewer can trust.
Estimand, Design and Model in Line Before any calculation, the stated question is matched to a summary measure, the design to the unit of analysis and the planned model to both, and each gap is closed in the plan.
Sample Size Against Real Information A sample size set for the total N misleads when the information sits in events, clusters or the rarer outcome group, so the calculation and the planned model use the same quantity.
Missing Data in the Plan The plan states why data may be missing, the primary method and the sensitivity analyses, written before results are seen and not after a disappointing estimate.
Confirmatory and Exploratory Work Primary, secondary and exploratory analyses are labeled in the plan, dated versions are kept, and every departure from the plan is documented with its reason.
Overfitting in Prediction Models A model with many candidate predictors and few events overfits, so the sample size for prediction, internal validation and calibration are planned, and apparent performance is not reported alone.
Endpoint Form and Information Dichotomizing a continuous measure discards information and shifts the estimate with the cut-off, so each endpoint's form is chosen deliberately and composite endpoints are reported by component.
STATISTICAL METHODS
Methods Selected for the Research Question and Study Design
Each method family is chosen for the outcome and the data structure of your study, and the line under each title explains why it is used in biostatistical work.
Regression & Predictive Modeling
Continuous and binary outcomes are modeled with covariates chosen before the data are seen and a size the events can support, with performance validated beyond the development sample.
Clinical, Survival & Epidemiological
Prognostic models for time-to-event outcomes and thresholds for diagnostic markers are estimated and validated, with discrimination and calibration reported together.
Longitudinal & Multilevel Analysis
Patients nested in centers and repeated measurements are modeled directly, with the number of centers checked before a center effect is estimated.
Bayesian & Computational Methods
Imputation carries out the missing-data plan, the bootstrap validates prediction models internally, and simulation shows how a design behaves before the study begins.
Comparative & Hypothesis Testing
Primary comparisons are specified with their covariates, scale and multiplicity rule before the data are seen, so each test answers a question named in the plan.
Multivariate, Scale & Reliability
Agreement between a new measurement and a reference method, or between raters, is reported with agreement measures, with correlation shown beside them, never in their place.
DELIVERABLES
Statistical Outputs Prepared for Your Research
The outputs follow the order of biostatistical work, from the check that design and model agree to the dated plan and the programs behind every figure.
Data & Study Review The protocol or draft, outcome definitions, pilot data and any existing analysis are reviewed to confirm that the estimand, design and planned model agree before work proceeds.
Statistical Analysis Sample size or precision calculations, the primary and sensitivity analyses of the plan, or the development and validation of a prediction model, with the diagnostics each requires.
Effect Estimates & Uncertainty Effect estimates on the scale of the estimand with confidence intervals, and for prediction models discrimination and calibration with optimism-corrected performance.
Tables & Figures Sample size tables with their assumptions, calibration and validation plots, participant flow, adjusted effect tables and shells of the tables the plan commits to.
Statistical Interpretation Estimates are explained against the question in the plan, confirmatory and exploratory results are kept apart, and limits from design and missing data are stated for the discussion.
Analysis Files A dated analysis plan, the sample size workings and programs in R, SAS, Stata or Python, documented so that each reported figure can be reproduced.
REPORTING
Reporting Biostatistical Research
The guideline follows the type of study the plan describes, and each is applied only where it fits.
Randomized trials report to CONSORT 2025 and their protocols to SPIRIT 2025, observational studies to STROBE, diagnostic accuracy studies to STARD 2015 and prediction model studies to TRIPOD+AI. Sample size calculations are reported with every assumption, so that a reviewer can repeat them, and the analysis plan is cited by its version and date. Primary and exploratory analyses are labeled as such, effect estimates carry confidence intervals, and departures from the plan are reported beside the results. The instructions of the target journal and the ICMJE recommendations are checked on top.
ILLUSTRATIVE EXAMPLE
How the Analysis May Look in Practice
An illustration of how a statistical plan is fixed at the design stage, before any data from the new study exist.
ILLUSTRATIVE EXAMPLE · GRANT PROTOCOL DEVELOPMENT · UNITED STATES
A clinical research group preparing a funding application for a randomized comparison needed a statistical section with a sample size, an estimand and an analysis plan that a review panel would accept.
The only estimate of variability came from a small pilot study, and the outcome would be missing for some participants at follow-up. Before any new data existed, we fixed the estimand and the primary model, calculated the sample size over a range of plausible standard deviations, including the upper limit of the pilot’s confidence interval instead of its point estimate, and increased it for expected attrition under a stated missing-data assumption. The plan named the sensitivity analyses, the multiplicity rule for the secondary outcomes and the analyses that would be exploratory. Pilot results were treated as a guide to feasibility, not as an estimate of the effect, and the plan was cited by version and date so that reviewers could see what had been fixed in advance.
Outcome: The group received the sample size justification with every assumption and a sensitivity table, the estimand statement, a dated analysis plan with table shells and the calculation scripts.
This example does not describe a real client or project. We never share details of actual client work, in line with the confidentiality we promise every researcher. This illustrative example is included to show the type, scope, and standard of work involved in this kind of analysis.
RESEARCHER VOICE
What Researchers Say
Feedback from a biostatistician on support that went beyond running software, on handling missing data without distorting estimates and on reporting framed for how studies are reviewed.
“We needed statistical support from someone who understood Biostatistics, not just someone who could run software. The main issue in our data was how missing data should be handled without distorting the estimates. The analysis and reporting were framed around the way studies in our field are actually designed and reviewed, which made a noticeable difference.”
HOW WE ASSESS YOUR PROJECT
We Review Your Actual Research
Your assessment is based on a review of your actual research materials and analytical requirements, not a generic project category.
You Submit Your Project Details
Share whatever is available at the current stage, such as the research aim, protocol or proposal, dataset, variable list, manuscript or statistical requirements.
We Review Your Research and Define the Scope
Your statistically trained Research Coordinator reviews the materials to understand the study design, research questions, data structure, analytical requirements and expected deliverables. If clarification is needed, additional documents or information can be requested before the assessment is finalized.
Your Analysis Is Completed and Delivered
Once the scope and fixed project fee are agreed, the in-house statistician completes the work and the Research Coordinator remains the communication point through delivery, with the analysis, deliverables and timeline as agreed.
SOFTWARE
Statistical Software for Biostatistical Research
Biostatistical work is carried out in the package your group already uses, with programs or syntax provided where applicable.
Statistical Support Within Your Research OnlineSPSS provides the agreed methodological, analytical, and statistical reporting support. The researcher or research team retains the scientific question, substantive interpretation, conclusions, and overall responsibility for the research.
FREQUENTLY ASKED QUESTIONS
Questions About Biostatistical Consulting
Short answers to what research teams ask before sharing a study, from analysis plans and sample size to missing data, confidentiality and the assessment.
What does biostatistical consulting cover in a research project?
It covers the statistical side of a health study from design to reporting: defining the estimand, calculating the sample size, writing the analysis plan, analyzing the data and preparing the reporting. Involvement at the design stage can prevent problems, but a completed analysis can also be reviewed. Our Statistical Methods pages describe the individual methods.
When should a statistical analysis plan be written?
The plan should be written before the outcome data are seen, and in a trial before unblinding. Fixing the primary analysis, the missing-data handling, the sensitivity analyses and the multiplicity rule in advance keeps the choices from being shaped by the results. The same rule holds in an observational study: the plan is finalized before outcome data are extracted or examined.
How is the sample size for my study calculated?
Sample size is calculated for the primary outcome and analysis named in the plan, from the smallest effect worth detecting, the variability or event rate, the expected missing data and the planned model. Pilot estimates are imprecise, so a range of values is shown. For an existing dataset, the question becomes the precision its events can support.
How should missing data be handled in the analysis plan?
The plan begins with why data may be missing, then names a primary method that fits that assumption, often multiple imputation or a model that uses all observed data, and sensitivity analyses that vary the assumption. A complete-case analysis is primary only if its assumptions are plausible for your data.
How is confidential study data handled?
Research data and project details are handled confidentially, and project information is not published. Researcher names and institutions are not used publicly without permission. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You send your research question, protocol or draft plan and any data you have. A statistically trained Research Coordinator reviews them and responds the same day, and the analysis, deliverables, timeline and one fixed project fee are confirmed before work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Discuss Your
Biostatistical Research With Us
Share your research question, protocol or draft plan and any pilot data, even at the design stage. We check that estimand, design and model agree and confirm the approach before work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014