Workspace
Blog

Research with Beeblio, Part 2: Design the Study and Collect Data

1 Nov 2026·

Turn a question into a method, inspect Beeblio’s survey form and codebook, and preserve a raw-data record before analysis.

Part 2 of How to Finish Your Research with Beeblio. Part 1 left us with a focused question and a checked evidence map. Now we turn them into a method and a data record.

The point of a study plan is to make important decisions before a result can tempt you to change them. In the Everyday Habits & Well-Being demo, the question concerns sleep, activity, leisure screen time, social connection, and well-being. The next step is to define how each concept becomes a variable, who or what contributes an observation, and which analysis can answer the question.

The evidence map from Part 1, still in view as the method takes shape.

Write the method while the decisions are still visible

Open 2-Data/survey-instrument.md and codebook.json. The instrument gives the wording of the example questions. The codebook states the column names, units, ranges, and missing-value conventions. Compare them with the research question in 4-Reports/research-report.md. The distinction between “activity minutes per week” and “well-being score from 0 to 10” matters both for analysis and for the eventual Results section.

Next open 3-Analysis/analysis-notes.md. It lists descriptive statistics, correlations, and a multivariable model. For your own project, put a method and analysis plan in a saved document before looking at outcomes. It should state:

  1. Who or what is eligible, and how observations will be obtained.
  2. The exact definition, unit, and source of every variable.
  3. How missing, invalid, and duplicate entries will be handled.
  4. The planned comparison or model, including relevant confounders and diagnostics.
  5. What the design can and cannot support.

Ask Beeblio to critique an ambiguity, then revise the document yourself. A useful prompt is: “Read my question, instrument, and analysis plan. Find any variable that is undefined, measured inconsistently, or used in the model without a clear rationale.” Keep the revised plan in the project so later analysis can be compared with it.

The plan, committed to writing before the first result exists.

Inspect the survey form as a data source

Open Data and select 2-Data/daily-habits-survey.form.html. The editor has Build, Preview, and HTML source modes. In Build, inspect the demo’s numeric, multiple-choice, and linear-scale questions and the section that groups them. In a new project, create a survey form from Data, add the introduction and consent language appropriate to your study, choose question types, set required fields and option labels, and use Preview to see the respondent experience. Choice and multiple-choice grid questions can be edited in the visual builder too. Test the wording with a pilot where appropriate; document what changed after feedback.

The demo form’s block identifiers mirror the codebook variable names, which makes the response columns easier to trace. Its paired daily-habits-survey.responses.csv illustrates the structure of a form’s submissions. Once a real instrument is approved, use the form file’s Share File → Public link control to open its respondent page and test a submission. Publicly shared form files use a focused respondent view; responses are saved in the paired CSV. In a live project, that file grows as responses arrive. Here the rows were generated with the demo, so opening that file should be treated as inspecting sample data, not as observing a real fieldwork period.

Be careful about what “ready to collect” means. A form preview can confirm that controls work and that questions are understandable; it cannot establish that a measure is valid for your population or that you have permission to recruit participants.

The instrument under construction: each question given a type, a range, and a rule.

Preserve what arrived

The demo also contains 2-Data/daily-habits-survey-raw.csv and daily-habits-survey-clean.csv. Open the raw file first. It contains some blank or NA entries and inconsistent casing in work arrangement. Those details are why a raw snapshot is valuable: it preserves the evidence of what needs a decision. Do not silently edit the raw file to make later tables look neat.

For your own study, make a dated fieldwork note with the collection period, response count, eligibility decisions, and any deviations from the protocol. Freeze a raw snapshot before cleaning. Restrict access to respondent-level data as needed, and decide separately what may be shared. If you return to the project after a gap, Beeblio’s project files and chat history help you recover the earlier decisions, but the saved method and fieldwork note should remain the authoritative record.

The raw record: what arrived, preserved exactly as it arrived.

The handoff to analysis

You should now be able to answer three questions without guessing: What was planned? What was asked or observed? What did the raw data contain? The files that answer them are the instrument, codebook, method or analysis plan, and untouched raw snapshot. Part 3 uses those files to make cleaning choices, produce results, and check figures.

Beeblio Team

researchbeebliodata

The Premise

Why we're building a full-stack research workspace — and finishing the job the IDE started.