Form Creation

Part of Census and Demographics

How to design physical census forms that enable accurate, fast, and consistent data collection in field conditions.

Why This Matters

A poorly designed form slows down interviews, introduces errors, and creates data entry nightmares. A well-designed form fits the natural flow of a conversation, makes correct recording obvious, and catches errors before they leave the household. Since a census may involve thousands of households and dozens of enumerators, the cumulative effect of form quality — good or bad — is enormous.

Form design is a craft that combines content decisions (what questions to ask) with layout decisions (how to present them). Content decisions are covered in the Census Design article. This article focuses on the physical and visual design of the form itself: paper, layout, field sizes, instructions, and the small design choices that determine whether enumerators use the form correctly.

Form Structure

Household form vs. individual roster: Two common approaches:

One form per household: The household form has header sections for location information, then a roster (table) with one row per individual and columns for each individual-level question (sex, age, occupation, etc.). Compact — all household information on one or two pages. Works well when household size is consistent and small.

One form per individual: Each person in the household gets a separate form. More flexible for variable household sizes, easier to add or remove individuals. Requires aggregation when computing household-level statistics. Works better for complex questionnaires with many individual-level questions.

Section organization: Group questions by logical category — location and household information, then household-level questions, then individual roster. This matches the natural interview flow (general to specific) and reduces the risk of the enumerator losing their place.

Sequential page numbering: Number every page of every form. If a form comes apart, pages can be reassembled. “Page 2 of 3” is better than just “Page 2” because it flags when pages are missing.

Field Design

Pre-coded boxes: For questions with a fixed set of answers, provide boxes labeled with the codes. The enumerator circles or checks the appropriate code, rather than writing a word. Pre-coded fields are faster to complete, easier to read, and easier to enter into summary counts.

Example instead of writing “Female,” the enumerator circles “2” in a box labeled “Sex: 1=Male 2=Female 99=Not stated.”

Write-in fields: For names, open-ended responses, and items that cannot be pre-coded. Make these fields physically large enough for legible writing — cramped fields produce cramped, illegible writing. Allow 2–3 lines for names; more for free-text notes.

Check boxes: For yes/no questions or multiple-selection questions, check boxes are faster and clearer than write-in responses. Make boxes large enough to check legibly without ambiguity about whether the box was intended to be marked.

Numeric fields: For age, number of household members, etc., provide clearly marked digit boxes. “Age: [][]” (two boxes) for a maximum two-digit age is better than a single large box, because it implicitly communicates the expected format and catches implausible values (three digits for age would be entered in the wrong box).

Visual Clarity

White space: Cramped forms with minimal margins and tiny print are hard to use in field conditions (poor light, tired eyes, physical awkward positions). Leave generous margins and space between sections. Better to use two pages than to cram everything onto one.

Alignment: Questions, their answer options, and their coding boxes should be consistently aligned across the form. Visual alignment allows the eye to scan quickly and reduces the chance of recording an answer in the wrong row.

Instructions placement: Put instructions exactly where the enumerator will need them — next to or under the relevant question, not in a separate instructions section they must flip to. “If household has more than 10 members, use continuation form” belongs next to the row count field, not in a separate manual.

Shading alternating rows: In the individual roster, lightly shade alternating rows. This simple technique dramatically reduces the error rate of recording individual data in the wrong row.

Font size: Minimum 10-point equivalent for print. Larger for field forms used in challenging conditions. Never use font sizes that require close reading in dim light.

Physical Materials

Paper weight: Heavy enough to survive field conditions — being carried in a bag, rained on lightly, written on against a hard surface. Thin newsprint tears and soaks; medium-weight bond paper (80 g/m² or heavier) is adequate.

Ink vs. pencil fields: Forms pre-printed in ink; answers written in pencil (which can be erased and corrected without creating a confusing mess) or in permanent ink (which cannot be altered, desirable for legal forms like death certificates). For most census forms, pencil is practical; for legal vital registration, permanent ink is required.

Paper size: Standard sizes (A4/letter) are easiest to copy and archive. Custom sizes create filing and copying complications. A4 is the right default unless specific content requires otherwise.

Binding: Individual loose forms can be organized in batches. For small surveys, a clipboard with forms is sufficient. For larger surveys, pre-bound booklets with multiple forms per booklet allow quicker handling and reduce the risk of loose forms being lost.

Testing the Form

Cognitive testing: Before printing, have several people who did not design the form attempt to complete it without instructions. Wherever they hesitate, ask the wrong question, or record in the wrong field, there is a design problem.

Field testing (pilot): As part of the questionnaire pilot, test the physical form as well. Watch enumerators completing it in real household conditions. Note: where do they slow down? Where do they make mistakes that have to be corrected? Where does the form’s physical format create confusion with the questionnaire instructions?

Printing check: Before the full print run, review a sample copy from the printer for legibility, correct page order, and complete printing of all fields. Print errors caught before distribution are far cheaper than print errors caught in the field.

Version Control

Every form should be identified with:

• A version number or date (top corner, every page)
• The name of the survey and the sponsoring organization
• A form identifier (form A, form B; household form, continuation form)

Version control prevents the disaster of outdated form versions being used alongside revised ones. Different versions of a form with different question wording or coding produce incomparable data. Explicit version identification on every page makes it possible to track which version was used for which data, and flag any version discrepancies during analysis.