STATISTICAL SOFTWARE
SAS Statistical Analysis Service
Rerunnable Programs, Auditable Logs
Our SAS statistical analysis service carries out regression, mixed-model, survival and missing-data analysis in SAS, working from your datasets and programs, with the SAS programs, logs and output provided as agreed so that every result can be rerun and every warning traced.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
What Is SAS Used For in Research?
SAS is a commercial analytics software suite from SAS Institute, used widely in clinical research, public health and institutional data work. Researchers choose it for its statistical procedures, such as PROC LOGISTIC, PROC MIXED and PROC PHREG, and because the data step and the log document how each dataset was built and which notes or warnings every run produced. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, provides a SAS statistical analysis service in which an in-house PhD-level statistician carries out the agreed analysis in SAS, working from your files and keeping the programs and logs so that the analysis can be rerun.
WHEN TO ENGAGE
Four Points at Which Researchers Bring Us a SAS Project
A SAS statistician may be asked to write the programs, to explain a warning, to review code someone else wrote or to keep an established SAS process running.
Analysis Plan Ready to Program A statistical analysis plan or protocol already names the models, and each one has to be programmed, run on the final datasets and checked against the plan before the tables are produced.
Warnings in the Log A run finishes, but the log shows separation, convergence or missing-value notes, and the cause has to be found in the data or the model before any coefficient is trusted.
Programs to Check or Extend Programs inherited from a colleague or an earlier study need correcting or extending, which is SAS programming for research that begins with a rerun on the current data and a full reading of the log.
Existing SAS Process to Keep Your team's data preparation and reporting already run on SAS programs, so the analysis stays in SAS and its programs are organized to run again from start to finish.
ANALYSES IN SAS
Analyses Carried Out in SAS for Your Study Design
The procedure follows the outcome, the design and the correlation among repeated measures, and each card links to the method page behind the choice of model.
Logistic Regression
SAS regression analysis for binary outcomes in PROC LOGISTIC, with the modeled event and the reference categories stated, and results reported as odds ratios with confidence intervals.
Firth's Logistic Regression
When separation is reported or events are rare, penalized likelihood estimation is considered, and its estimates are compared with those of the standard model before one is chosen.
Cox Regression
Time-to-event data are analyzed in PROC PHREG, with the time scale, censoring and handling of tied event times defined, and proportional hazards examined before hazard ratios are reported.
Linear Mixed Models (Multilevel Modeling)
Repeated measures and clustered data are fitted in PROC MIXED, with the covariance structure and the degrees-of-freedom method chosen to match the design and stated in the report.
Generalized Estimating Equations (GEE)
Population-averaged models for correlated binary or count outcomes use the REPEATED statement in PROC GENMOD, with the working correlation stated and robust standard errors reported.
Multiple Imputation
Missing values are imputed with PROC MI, each completed dataset is analyzed, and PROC MIANALYZE pools the estimates under a missing-data assumption that is stated openly.
COMMON SAS ISSUES
Where SAS Analyses Go Wrong Before the Estimate
Many SAS problems do not stop the run: the program finishes, the log carries a note or warning and the output looks complete, so each check below is made before an estimate is reported.
Separation Warnings in PROC LOGISTIC Complete or quasi-complete separation gives unstable or infinite estimates. The warning is traced to sparse cells or a near-perfect predictor, and the model is changed rather than the message ignored.
Event Level in the Response For a binary outcome PROC LOGISTIC models the lowest ordered response level unless told otherwise, so with 0 and 1 codes the odds ratios can describe the wrong event. The modeled level is set and confirmed.
CLASS Coding and Reference Levels The parameterization and reference level of each CLASS variable are set explicitly, because they decide what every coefficient and odds ratio compares, and default coding differs between procedures.
Missing Values and Observations Used A model drops every observation with a missing value on any variable it uses, so fewer observations may be used than were read. The counts in the log are compared with the study sample.
Dates, Formats and Informats SAS stores a date as a count of days from January 1, 1960, so a date column showing numbers lacks a format, and text read with the wrong informat becomes missing. Both are checked against the source.
Notes in the Log Notes about uninitialized variables, missing values generated by an operation or character values converted to numeric do not stop a run. The log is read in full, and each note is resolved or documented.
DELIVERABLES
What You Receive From a SAS Analysis
The files follow a rerunnable SAS project, agreed in the scope and provided where applicable, so that a coauthor or reviewer can follow each result back to the data.
SAS Programs Numbered programs for data preparation, analysis and tables that run in sequence from the source data, with comments stating each modeling decision.
Logs and Output Complete logs from the final run and the procedure output, kept with the programs, so that every note and warning, including the resolved ones, stays auditable.
Analysis Dataset A permanent SAS dataset with variable labels and formats, plus a data dictionary of the derived variables and every missing-value rule applied.
Tables and Figures Publication-ready tables and figures in Word, Excel or another format as agreed, with effect estimates, confidence intervals and the sample size behind each.
Statistical Interpretation A written interpretation of each estimate within the limits of the design, including how warnings and missing data were handled, for use in your own methods and discussion.
Release and Settings The SAS release and the system options and macro variables the programs rely on are recorded, so that the same results can be produced in another SAS environment.
OTHER PLATFORMS
When Another Platform May Be More Appropriate
The platform follows the method and the team, and these are the situations in which an analysis that could run in SAS may fit another platform better.
ILLUSTRATIVE EXAMPLE
How a SAS Analysis May Look in Practice
An illustration of how a warning in the log is handled before a rare-event model is reported.
ILLUSTRATIVE EXAMPLE · VACCINE SAFETY SURVEILLANCE RESEARCH · UNITED STATES
In a registry extract already prepared in SAS by its data manager, a university pediatrics group needed to test whether a rare adverse event was associated with a vaccine dose schedule.
The model, a logistic regression with the dose schedule and a small set of covariates specified in advance, was fitted in PROC LOGISTIC. The first run completed but the log reported quasi-complete separation: in one site category no adverse events had occurred, so its coefficient had no finite estimate. Rather than suppress the message, the cell counts were examined, sparse sites were combined under a documented rule, and a penalized likelihood estimate was run beside the standard model as a sensitivity analysis. The modeled event level and the reference categories were confirmed in the output, and the observations used were reconciled with the registry count.
Outcome: The group received the odds ratio with its 95% confidence interval from the primary model and from the penalized sensitivity analysis, the SAS programs and logs, and a permanent analysis dataset from which the analysis reruns from the registry extract.
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.
RESEARCH DISCIPLINES
Research Fields Where SAS Is Common
SAS clinical data analysis has a long history in pharmaceutical and clinical trial research, and SAS is also common in public health and institutional data work, where the same questions about warnings, missing data and rerunnable programs arise.
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.
RESEARCHER VOICE
What Researchers Say
Feedback from a CRO biostatistician on separation warnings in PROC LOGISTIC and a workflow kept in SAS.
“We already had most of the project in SAS, but our main issue was separation warnings in PROC LOGISTIC and unstable coefficient estimates. The team worked from the files we already had, cleaned up the workflow, and returned the analysis in a form we could reproduce. I especially appreciated not having to rebuild everything in another program.”
Statistical Support Within Your Research OnlineSPSS provides the agreed methodological, analytical and statistical reporting support, including the SAS programs and logs described in the scope. The researcher or research team retains the scientific question, substantive interpretation, conclusions and overall responsibility for the research.
FREQUENTLY ASKED QUESTIONS
Questions About SAS Analysis for Research
Short answers to what researchers ask before sharing a SAS project, from the analyses we carry out and existing programs to file standards, differences from other packages and confidentiality.
What does a SAS project with OnlineSPSS usually cover?
Most SAS projects cover regression, mixed-model, survival, repeated-measures and missing-data analysis, delivered with the programs, logs and output needed to rerun them. Where a statistical analysis plan exists, the programs follow it model by model; where none exists, the models are agreed during the assessment.
Can you work from my existing SAS programs and datasets?
Yes. Your programs are first run as they stand on your datasets, and the log from that run shows what each data step and procedure actually did before anything is changed or added. If the programs are finished and the question is whether the results hold, Independent Statistical Review checks them, including any parts drafted with an AI tool, and sets out what should change.
Can the programs and files follow the conventions my study uses?
They can, where this is agreed in the scope. If your study uses a defined data structure, such as CDISC-style analysis datasets, or a programming convention of your team, share the specification at the assessment, and how it is followed is confirmed before work begins. Regulatory submission requirements and any formal validation remain with the study team.
Why do my SAS results differ from R for the same model?
Usually because the defaults differ, not because one package is wrong. PROC PHREG handles tied event times with the Breslow method unless told otherwise, while the coxph function in R uses Efron’s method, and PROC MIXED applies its own denominator degrees-of-freedom method where lme4 reports none. When two packages disagree, each default is identified and matched before the results are compared, and the setting used is stated in the methods text.
How do you protect my research data and confidentiality?
Your SAS datasets, programs and logs are handled confidentially, and project information is not published. 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 share your research question, design and relevant materials, such as the datasets and any programs, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are confirmed before any 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
SAS Analysis With Us
Share your research question, datasets and any existing programs or logs, and the models you have in mind. We review the files and confirm the SAS setup before any work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014