Free resource: Open the printable consulting proposal checklist to document scope, deliverables, fees, revisions, and open questions.
A statistical consulting proposal should make clear what question the work will address, what you will receive, what assumptions the quote relies on, and what happens when the project changes. “Analyze the data and provide results” leaves too much open for either party to plan reliably.
Use this checklist to compare proposals or prepare a scope discussion. It is a practical planning guide; the details should reflect your project and the agreement you ultimately make.
State the research question, outcome, target population, and intended comparison in plain language. Identify the audience: a dissertation committee, journal, funder, internal team, or another decision-maker. Specify whether the work is exploratory, confirmatory, descriptive, predictive, or intended to support a causal interpretation.
For a fictional training study, “compare six-month outcomes between two formats, accounting for baseline scores and repeated observations” is more useful than “run statistics.” The proposal should also explain which conclusions the design cannot support.
Describe the expected files, approximate dimensions, identifiers, codebook, repeated or clustered structure, and known missingness. Identify who will supply the data, resolve coding questions, and confirm the meaning of variables.
Make data cleaning explicit. Combining exports, resolving duplicate records, creating analysis variables, and investigating discrepancies can require substantial work. If the quote assumes an analysis-ready dataset, define what that means and how unexpected problems will be handled.
Outline the proposed methods and the rationale for them without pretending every choice can be finalized before review. Include relevant diagnostics, missing-data handling, sensitivity analyses, and planned comparisons. Where decisions remain open, state how they will be made and documented.
An agreement to investigate a model’s suitability should not become a guarantee that the model will work or yield a desired finding. Record substantive changes to the analysis plan rather than silently replacing it after viewing results.
Specify the formats and level of explanation you need. Manuscript drafting, presentation design, and editing may be separate tasks. If code depends on licensed software, confirm that your team can use it.
List milestones and the information needed to start each one. Distinguish the consultant’s work time from delays awaiting files, decisions, or feedback. Include time for the researcher to check that variable definitions and interpretations match the study.
Define completion by the agreed deliverables and quality checks, rather than statistical significance or journal acceptance. Set a practical process for raising questions or reporting errors after handoff.
State whether fees are fixed, hourly, capped, or staged, along with payment milestones. Identify assumptions behind the estimate and any exclusions. Specify how additional work will be proposed and approved before it is undertaken.
Examples of scope changes include a new outcome, another data wave, a revised questionnaire scoring rule, or a request for additional subgroup analyses. Distinguish corrections to agreed work from new work. See the consulting-cost guide for the factors that affect effort.
Specify how many review rounds are included, what each round covers, and the time window for feedback. Clarify whether committee or journal reviewer responses are part of the initial engagement or a later phase.
A request to explain an existing decision differs from a request to collect new data or fit a different family of models. Decide how those situations will be scoped rather than assuming all future revisions are included.
Agree on who may access project data, the transfer method, storage arrangements, and retention or deletion expectations before sharing records. Address institutional requirements and restrictions on using external services. An initial inquiry can usually begin with a data description instead of participant-level files.
Discuss ownership and permitted use of deliverables. For research publications, agree on how contributions will be recognized and revisit the discussion if the role changes. Authorship depends on contribution and the applicable journal criteria, not simply payment.
Ask: “Could someone unfamiliar with our conversation tell what is included, what I must provide, and what would require a new agreement?” If not, clarify the proposal before work begins.
Use the free analysis-planning worksheet to prepare your question and intended outputs. Send DASS a project inquiry for a scope discussion. For provider selection, read how to choose a statistical consultant.
Reference: The ASA ethical guidelines provide a foundation for expectations around competence, methods, limitations, confidentiality, and stakeholder communication.
One email whenever we publish something new. No spam, unsubscribe anytime.
{{ subscribeErrorMsg }}