Consent forms are usually written by people who are not the ones who will read them. They are cleared by ethics committees, copy-pasted between studies, and handed to a participant in the last minute before a session begins.1 Almost every constraint on the form pulls against the reader.
And yet the form is the first piece of writing the participant ever sees from us. It does most of the work of building trust, or losing it. Earlier work on informed consent in lab studies suggests this trust is built less by the words on the page than by the room around the words — the leading, the measure, the absence of an obvious hurry.
A page that is in no hurry
The simplest move is to widen the margins. A consent form whose body sits in a roughly 60-character measure, with generous leading and a single accent colour, reads as something written for a person — not something stamped out for a procedure.
In the rendered HTML version, the same setting is expressed as measure: 60ch in the page's frontmatter. Below the desktop breakpoint, the column simply takes whatever width the viewport allows, with 24 px side padding. The point is not to be precious about 60 characters — it is to never go above 75.
---
template: writeup
measure: 60ch
toc: true
accent: terracotta
---
Leading and rhythm
Newsreader at 18px / 1.65 on a warm off-white background gives the page enough air that a careful reader can stop on a word without losing their place when they look up. That is not just a comfort move — it materially changes how much of the form a participant retains, judging from the short recall task we ran in the v3 round.
The reader has to feel that the page has been arranged for them, and not against them. Everything that follows from that is detail.
What the recall task showed
Six items, scored against a transcript of the form. The effect on the optional clauses is the most interesting row — those are the items participants have always told us they ignored.
| Version | Measure | Recall, required | Recall, optional |
|---|---|---|---|
| v1 — original A4 form | 96ch | 71% | 22% |
| v2 — narrowed margins | 72ch | 78% | 34% |
| v3 — present template | 60ch | 81% | 52% |
Signalling optionality
The harder problem, and the one I have been turning over for the past few months, is how to signal that something is optional without burying the option. The form has to read as a whole — but a participant has to be able to skip a paragraph and feel they have not skipped anything that mattered.
Two things that did not work, and one that did:
- Italics for optional clauses. Read as emphasis, not as optionality. Made the optional clauses feel more important, not less.
- A second column for optional content. Broke the reading rhythm. Participants either read both columns as one and lost the distinction, or read only the left column and missed the optional material entirely.
- A short label, in the meta colour, above the clause. Worked. Reads as a quiet acknowledgement that this paragraph is not load-bearing.
The third pattern is what the present template uses; you can see one in the v3 form in the appendix.
What I am doing next
The next round runs in May with a wider sample and a proper time-on-page measure. If the optional-recall effect holds at n > 60, I think the template is finished and we should put it in the lab's shared materials.
Footnotes
-
On the assumption that the participant has read no email ahead of the session, which we have learned, painfully, to assume. ↩