RESEARCH GUIDE
The Statistical Consultation Process
What to Prepare and Who Decides What
A consultation starts on your side: a defined question and your materials, brought to a statistician who advises on design, analysis and interpretation, while the decisions stay with you. Preparation makes the first conversation specific.
The statistical consultation process is a division of labor: the statistician brings methodological judgment to the design, analysis or reporting of a project, and the researcher brings the subject knowledge, the data and the final decisions. It usually runs from the question and the materials to an agreed scope, then to the analysis or advice, and finally to a record of what was decided and why. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, provides agreed methodological, analytical and statistical reporting support in the areas listed under Services, while the research question, the interpretation and the conclusions remain the researcher’s.
FORMS OF ENGAGEMENT
Types of Statistical Consultation by Who Carries Out the Work
Consultations differ less in subject than in who runs the analysis, who reads the output first and who signs off each choice.
Design Review Before Data The question, the design and the planned sample are examined before collection starts, with Research Planning & Study Design used where a full written plan is needed.
Advice at Agreed Points The researcher keeps the data and runs every model, and the consultant comments on the plan or the output at agreed points, which suits a researcher who must carry out each step personally.
Analysis Run by the Statistician The statistician fits the agreed models, and the researcher receives the code and output with a note on each modeling choice, so the analysis can be rerun and checked.
Shared Analysis The researcher prepares the data and runs the main models, while the consultant specifies the harder parts, such as clustering or missing values, and checks the results against the design.
Reporting Consultation The analysis is settled and the open question is presentation: which estimates belong in the tables, how the methods are described and what wording the results can carry.
Second Opinion on Finished Work A statistician with no part in the original analysis reads it against the design and the claims made in the text, which is often useful once a reviewer has queried the statistics.
ROLE COMPARISON
Researcher and Consultant Responsibilities Compared
Friction in a consulting arrangement usually comes from an unspoken assumption about who decides, and a table makes the usual division explicit.
| Area | The researcher | The statistical consultant | Agreed in the scope |
|---|---|---|---|
| Research question | Defines it from the field and owns it | Tests whether the data and the design can answer it | The exact question the analysis answers |
| Design and data | Knows how the data were collected and what they mean | Reads the structure of the data and what it implies for the analysis | The files, variables and exclusions involved |
| Choice of method | Approves the approach | Recommends a method, gives the reasons and names the alternatives | The models and the checks to be run |
| Interpretation | Decides what the results mean for the field | Explains how far each estimate and its uncertainty reach within the design | The estimates and intervals to be reported |
| Conclusions and authorship | Remain with the researcher, who is responsible for the study | Supplies the methodological reasoning, and any acknowledgment follows the journal’s rules and the researcher’s decision | The outputs to be delivered |
| Records | Keeps the decisions and their reasons | Documents what was run and why | The deliverables and the timeline |
KEY PLANNING DECISIONS
Preparing for a Statistical Consultation
Rough notes on the following points give a first conversation something concrete to test, and a draft is as useful as a finished document.
The Question in One Sentence State what you want to know, about which population and with which outcome, since a consultant can only test whether the data answer a question that has been put clearly.
The Design as It Stands Describe how participants were sampled or allocated, how often each was measured and whether the data are paired or clustered, because those facts decide which methods are open.
Data and Documentation Gather the dataset or a sample of it, the codebook or variable list and a note on coding and missing values, so that questions about the data can be answered from the file itself.
Rules From Outside Note what the journal, the funder, the ethics approval or the program expects, including any preregistered analysis, because these limit the choices before the consultant is asked.
The Outcome You Need Say what you need to leave with, such as a plan, a result, a check of finished work or methods wording to review, so that the scope names a concrete end point.
Who Approves the Analysis Identify the supervisor, co-authors or committee who approve the analysis, since their requirements shape the approach more than the choice of software does.
COMMON ISSUES
Expectations That Undermine the Collaboration
These mismatches are easy to avoid when they are named at the start and costly to discover once the analysis has been done.
Only the Final Numbers Checked Asking a consultant to verify figures when the design was never examined leaves the most consequential choice unreviewed, so the request states which decisions are still open to comment.
A Verdict Expected From the Data No consultant can promise that an analysis will support a hypothesis, since the data decide the result and the work is judged by its soundness and not by its significance.
Interpretation Handed Over Delegating the interpretation leaves the researcher unable to defend it before a committee or reviewers, so each choice is explained and the meaning for the field is decided by the researcher.
Scope That Grows Midway Extra outcomes or comparisons added during the analysis change the workload and the agreement, so they are raised openly and the scope is revisited before the work is extended.
One Meeting Expected to Settle Everything A single conversation rarely resolves a complex design, because answers to the first questions often raise new ones about the data, so later questions are expected and recorded as they arise.
Data Shared Beyond the Consent Identifiable records sent outside the project can breach what participants consented to, so the ethics approval is checked first and the data are de-identified before they are shared.
REPORTING
Keeping a Record of What Was Agreed and Done
A shared record protects the researcher's ownership and lets a supervisor, co-author or reviewer see where each analytic decision came from.
A consultation leaves a trail that others can read: the scope, the analysis plan, notes on each decision and the output that followed. That record shows which choices were methodological advice and which were the researcher’s, and it supports the methods section, which names the software, its version and each analysis carried out. Under the ICMJE recommendations, a contributor is first judged against the authorship criteria, and those who do not meet them are acknowledged with their agreement. Many journals also ask for a contributor roles statement, and the CRediT taxonomy lists roles such as methodology and formal analysis. At OnlineSPSS, authorship remains the researcher’s and no intellectual contribution is claimed.
STATISTICAL METHODS
Method Families a Consultation Can Draw On
The family is set by the question and by how the observations are structured, which is often the first thing a consultant asks about.
Comparative & Hypothesis Testing
Group differences and categorical association
Regression & Predictive Modeling
Outcomes explained by several predictors
Multivariate, Scale & Reliability
Scales, correlation and many variables
Longitudinal & Multilevel Analysis
Repeated measures and nested data
Clinical, Survival & Epidemiological
Time-to-event and risk data
Qualitative & Mixed Methods
Interviews, text and integrated designs
ANALYSIS SERVICES
Where Each Service Fits in a Consultation
What follows a first conversation depends on the project, from a written plan to the analysis, the reporting or a reading of results already produced.
Research Planning & Study Design
The question, the design and the analysis plan examined together before data are collected, with the sample size and the analysis settled for each question.
Quantitative Data Analysis
The agreed models fitted to your data and explained to you, with diagnostics read against the design and every estimate given with its interval.
Results Reporting & Interpretation
Results formatted for the journal or program you are submitting to, with tables, figures and a methods description that match the analysis, and the interpretation remaining yours.
Independent Statistical Review
A finished analysis read against its design, with notes on what holds and what should be revisited before the results are reported.
FREQUENTLY ASKED QUESTIONS
Questions About Working with a Statistical Consultant
A consulting arrangement raises practical doubts before it starts: what it involves, how to judge the advice, whether a supervisor should join and how much the consultant must know about your field.
What does the statistical consultation process involve?
The arrangement rests on three agreements: the question the analysis will answer, what the consultant will examine or carry out and how the results will be recorded, while the researcher keeps the decisions and the conclusions. At OnlineSPSS the agreed analysis is carried out in-house by a statistician, with a statistically trained Research Coordinator as the main contact, and the stages are set out under How It Works.
How can I tell whether a consultant's advice is sound?
Sound advice ties each recommendation to your design and data, states the conditions it rests on and names the alternatives, so that you can explain the choice yourself. It reports estimates with their uncertainty and says plainly where the data are too thin to settle a question. If a recommendation cannot be explained in terms of your own study, it is worth asking for the reasoning before adopting it.
Should my supervisor or a co-author be part of the conversation?
It is often useful to include the person who approves the analysis, at least in the first conversation or through a shared record of it, because their expectations shape the method. The researcher remains the one who explains and defends the analysis: the supervisor states what the program or the field expects, the consultant gives the methodological reasons, and differences are discussed openly and decided by the researcher.
Does the consultant need to know my research field?
A consultant needs sound methods knowledge and a willingness to learn how the measures and the design work in your field, while the substantive expertise stays with the researcher. Terms, conventions and the size of a meaningful effect come from the field, and a short description of them in the first conversation prevents misreadings. How OnlineSPSS is organized is set out under About Us.
How do you protect my research data and confidentiality?
Materials brought to a consultation can include unpublished data, drafts and reviewer comments, and your data and project details are handled confidentially. Project information is not published, and 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 share your research question, design and relevant materials, such as a protocol, an analysis plan or a dataset, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then 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
A Clear Question Makes
a Better First Conversation
Your notes on the preparation points above can travel with the request, and a statistically trained Research Coordinator replies the same day to agree how the work will be divided.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014