Free resource: Open the printable consulting proposal checklist to document scope, deliverables, fees, revisions, and open questions.
Choose a statistical consultant by examining how they would approach your research question and what they would deliver. A list of software packages or a low hourly rate does not establish that someone can handle your design, explain the assumptions, or help you respond when the results are questioned.
Start by describing your intended claim, data structure, project stage, and deadline. You do not need to settle on a statistical test before asking for help. A useful first conversation should clarify the question and identify what remains uncertain.
Ask about work with the features that make your project challenging: repeated observations, clustered sampling, missing outcomes, questionnaire scores, small groups, or a particular reporting audience. Familiarity with your subject area can help, but it does not replace methodological competence.
For example, a fictional researcher studying students within classrooms would ask how the consultant would account for clustering and distinguish classroom-level from student-level comparisons. “I use regression” is less informative than an explanation of how the design changes the analysis and uncertainty.
Ask what the consultant can handle directly and when they would involve another specialist. A clear account of limits is more useful than a promise to handle every method.
A responsible consultant may recommend a design review or a short paid diagnostic phase before quoting the full analysis. That can be reasonable when the data condition or feasibility is uncertain. Ask what that phase delivers and how you will decide whether to continue.
Request an anonymized or synthetic example of a relevant deliverable. Confidential client files are not necessary. Look for clear interpretation, stated limitations, documented decisions, and a connection between the research question and the results.
Specify whether you need tables, figures, reproducible scripts, a methods description, or a walkthrough. A folder of software output may not meet your needs even if the calculations are correct. Conversely, a polished report without enough documentation may be difficult to reproduce or revise.
Identify who will do the work, who will answer questions, and how you will review interim decisions. Ask whether meetings and explanations are included, how corrections are handled, and whether you will receive the files needed to reproduce the final outputs.
If you are a graduate student, confirm your department’s rules on outside assistance. Agree on a role that lets you understand and defend your work. Statistical coaching and contracted analysis support involve different responsibilities.
Compare deliverables, assumptions, revision coverage, schedule, and exclusions before comparing the total price. A proposal that includes data reconciliation, sensitivity analyses, and reviewer support covers more than one that includes a single model and table.
Do not assume a fixed price is always preferable. A defined deliverable may suit a fixed fee; uncertain diagnostic work may suit hourly billing with a cap or a staged agreement. Read what a consulting proposal should include and what drives consulting costs.
Be cautious about guarantees of statistically significant results, publication acceptance, or a particular conclusion. Ask for an explanation if a provider selects a method before understanding the design, cannot describe limitations, or requests sensitive records before agreeing on a suitable transfer process.
These expectations align with the American Statistical Association’s ethical guidelines, which emphasize competence, transparent methods, confidentiality, and resistance to predetermined results.
Bring a short research question, a data dictionary or description of the data structure, and the actual deadline. See the first-call preparation guide. Use the free analysis-planning worksheet to organize your intended outcome and comparison.
Tell DASS about your project if you would like to discuss the support it needs. Keep participant records and confidential information out of the initial inquiry.
One email whenever we publish something new. No spam, unsubscribe anytime.
{{ subscribeErrorMsg }}