Stepping a form blocks overview and cross-checking

What it is

A single-screen form lets people glance across the whole picture: what was asked, whether answers agree. Paging splits that picture into routes. Overview disappears; cross-checking becomes going back a screen, relying on memory or a draft. Whether shipping matches billing, a company name matches a tax ID, or a date entered early is the same event as a date entered later—these cross-field constraints are no longer co-present. This entry is about that structural cost of stepping. It is not about field count driving abandonment, and not about whether a stepper says 2 of 5.

Why it happens

Checking depends on simultaneous visibility. Vision can compare two passages on one screen; once a “next” sits between them, comparison moves into working memory, which is small and lossy. People also store “that item lived lower on the page”; paging destroys the spatial anchor. If Back reloads or clears values, the cost jumps from a glance to retyping, and most people skip the check and submit inconsistencies. Branching makes it worse: later choices depend on earlier answers that are no longer in view, so people either guess or submit blind. Stepping therefore exports consistency load even as it lowers load on the current screen.

Studying it

Give a task that requires cross-field agreement (address vs invoice title, start vs end date, repeated email) and compare one page with paging; split the paged arm into “free back-navigation” vs “going back loses data.”

Independent variables: paging, whether back-navigation keeps values, a read-only review before submit, whether cross-field rules fire only at submit. Dependent variables: inconsistency rate, voluntary back-navigation, edits on a review screen, post-submit correction requests.

Lab participants know they will be scored on checking and over-use Back. Real users rarely page backward for consistency. A review screen adds its own leave point; separate “fixed on review” from “abandoned on review.” Do not use completion time alone—paging can be faster and still leave more internal contradictions.

Where it stops holding

When steps are truly independent (pick a plan, then a ship-to, with no field-level constraint), lost overview is cheap. Screen-reader users already move linearly; their extra cost is focus management and whether values survive, not lost glancing. Legal consent and e-signature steps sometimes cut overview on purpose so attention stays on the current statement. A two-step email-plus-code flow has almost no cross-field relation; it does not need a review screen for this reason.

Applying it

• Keep fields that must agree on one screen; if they cannot, echo a read-only summary of the earlier answer on the later screen, with an edit control that returns to the source.
• Before submit, show a read-only review grouped the way people think, not the way the wizard was engineered; each group can edit in place or jump back.
• Back-navigation must not clear input or rerun option-changing requests unless the person explicitly changed a prior answer.
• Verify by planting two facts that must agree (two addresses, a date range) and counting contradictory submits; if a review screen exists, see whether contradictions get edited there or ignored.

Related

• Within the group: H1.01.1 The number of fields shown at once drives abandonment · H1.01.2 Multi-step forms need a total and a current position
• Adjacent: H1.02 Field order and grouping · H1.16 Post-submit result presentation · H7.05 Order confirmation
• Search terms: form review · simultaneous visibility · cross-field consistency