Shaping Up Data-Entry Forms: Designs That Put In the Work

Once we've decided what data to collect, it's tempting to start designing forms.

Don't. At least not yet.

Before deciding what the form should look like, it helps to understand the work the form is being asked to support. Three things shape that work: the source, the task, and the person.

The source tells us about the information before it reaches the form. The task tells us what must happen to turn that information into data. The person tells us who is doing that work.

Together, they give us a practical way to think about a data-entry problem before we start moving fields around the screen.

The Source

Where does the information come from, and what is it like before it ever reaches the form?

A value doesn't always exist somewhere waiting to be copied into a field. Three characteristics of the source are particularly useful to consider: abstraction, accessibility, and structure.

Abstraction

How far is the source from the value we want to collect?

At one end, the relationship may be nearly one-to-one. A lab report says 4.8. The task may simply be to enter 4.8.

Move further away and some interpretation may be required. At the other end, the source doesn't contain the value at all. It contains information from which someone must derive or judge the value.

Consider a video of a hand tremor. Someone must examine the video, judge the severity, and translate that judgment into the categories defined by the study (e.g., mild).

Abstraction tells us how much interpretation stands between the source and the value we want.

Importantly, abstraction doesn't tell us what the form should look like. A highly abstract source may have no useful visual structure for the form to reproduce. A very concrete source may have structure that is critically important. That's a separate question.

Accessibility

How does the person get to the source?

Sometimes the source is sitting directly in front of them. Other times, getting to it is a task by itself. Maybe they must find the participant in an EMR, choose the correct visit, open the right report, find the appropriate section, and locate one measurement among several similar ones.

Every one of those steps matters because the characteristic error may not be an obviously wrong value. It may be the right-looking value from the wrong source: wrong participant, wrong visit, wrong report, wrong time point. The value itself can look perfectly reasonable.

That makes seemingly small things like breadcrumbs, dates, visit identifiers, source labels, or other orientation cues potentially important parts of the form. They don't just help someone find information. They can help someone confirm that they found the right information.

Structure

Does the source already contain organization that carries meaning?

Maybe it's spatial, with a finding occurring in a particular location on an image. Maybe it's sequential, with information produced or presented in a meaningful order. Maybe it's hierarchical. Maybe it has sides, regions, or temporal relationships that matter.

This gives us a particularly useful question:

Does the source carry structure the form can preserve or destroy?

That's different from abstraction. A source can require considerable judgment but have little meaningful structure for the form to preserve. A clinician deciding whether an adverse event is mild, moderate, or severe may have to weigh several pieces of information, but their arrangement in the source may not matter. Conversely, a lab value may require almost no interpretation at all. If the report says 4.7, the value is 4.7. But whether that 4.7 came from the correct participant, visit, or time point is essential to its meaning.

Abstraction is about how much interpretation stands between the source and the value. Structure is about whether the organization of the source carries meaning worth preserving.

When useful structure is already there, throwing it away may force the person to reconstruct it themselves. We'll see exactly what that looks like in Part 2.

The Task

What work must happen to turn the source into entered data?

"Data entry" makes the process sound like a single action. It rarely is.

Someone may have to retrieve the source, observe something, interpret it, apply a definition, translate terminology, remember where they are, perform a calculation, follow a sequence, locate the right field, and finally enter a value.

But not all that work comes from the same place. A useful question is: Why does this work exist?

Work Required by the Source

Some work exists because of the nature of the information itself. If we're rating hand tremor, somebody must observe and judge the tremor. If a measurement must be obtained, somebody or something has to obtain it. If the answer requires talking with a participant, that information must be elicited.

A better form can't simply make necessary observation or judgment disappear. But we can ask whether the person entering the data needs to do all of the surrounding work. Can the source be brought closer to them? Can information be pulled in automatically? Can part of the work be supported? Can a device or system do something we're currently asking a person to do?

Work Required by the Protocol

Other work exists because the study has decided how the source will become data. What counts as diffuse rather than localized? Which categories are allowed? What operational definition applies? What sequence is to be followed?

The source didn't create those requirements. The protocol did.

That distinction matters because protocol knowledge doesn't necessarily have to live in the person's head. Definitions can appear where they're needed. Reference images can help with difficult judgments. Instructions can be available at the moment of decision. Study logic can sometimes be incorporated directly into the workflow.

Work Created by the Form

And then there's work we created ourselves.

Translate left eye into OS. Remember which column represents OS. Find the Bottom row. Move across the table. Locate the correct dropdown.

None of that improves the observation or satisfies a scientific requirement. The form created the work. If the form created unnecessary work, perhaps the form can remove it.

The goal is to distinguish the thinking the task requires from the thinking our form unnecessarily adds.

The Person

Finally, who is doing the work?

The same source and the same task may call for very different forms depending on the person sitting in front of them. Rather than trying to catalogue every human characteristic that could matter, two questions get us surprisingly far.

What Can We Leverage?

Expertise. Experience. Learned routines. Familiar terminology. Expectations about what comes next. A mental map developed through thousands of repetitions.

People don't arrive at a form empty-handed. A dental hygienist may know the order of a periodontal examination so well that they barely need to think about where the next measurement belongs. A coordinator who enters the same form every day may have internalized definitions that an occasional user has long since forgotten.

Good design can take advantage of what people already know rather than making them relearn it.

What Might Get in the Way?

Maybe they're new to the task. Maybe six months pass between uses. Maybe staff turnover is high. Maybe their attention is divided between the form and a patient. Maybe interruptions are common. Or maybe a deeply learned routine conflicts with the order we've imposed on the screen.

Even expertise can become an obstacle if the form fights what the person has already learned. That's why asking whether a form is "intuitive" isn't quite enough.

Intuitive to whom, doing what?

Put Them Together

Source. Task. Person. None determines the form by itself.

A source may contain spatial structure, but that doesn't automatically mean the form should look like the source. A task may require considerable judgment, but that doesn't mean every part of that judgment needs to live in someone's head. An experienced person may bring knowledge we can leverage, while a novice doing the same task may need something entirely different.

The three also interact. Who the person is can change how much work a task requires. The structure of the source can change what the person has to translate or remember. And the way we design the form can either add to that work or take some of it away.

Next in Part 2, we'll start putting this framework to work with probably its most intuitive component: spatial relationships.