Clinical Research Data Entry: Why the Same Form Fails Different Users

At a glance

The best form depends not only on the information being collected, but on what the person entering it already knows and how they do the work.

Use the map they already know

Experienced users may already know the anatomy, sequence, terminology, or workflow. Making them use a second representation can add work rather than remove it.

Design for how they actually work

Some people watch the form continuously. Others enter from memory or a practiced sequence and only glance at the screen at natural stopping points. The form should support those moments.

Design against the likely failure

A good form doesn't just make entry easier. It helps prevent or expose the mistakes most likely for that person and task.

We ended Part 2 with what looked like a useful rule: when the source contains useful relationships, preserve them.

But that rule leaves out something important: the person using the form.

Now consider periodontal data collection. Several measurements per tooth. A full mouth. A specified order. Anatomy and spatial location everywhere.

Sounds perfect for a visual representation of the mouth.

Except the person entering the data may already have one.

The Map May Already Be in Their Head

An experienced hygienist doesn't need a picture of a mouth to know where they are. They already have one.

That map doesn't have to be anatomical. It may be a familiar sequence, an expected order, where information normally appears, how related information is grouped, or a pattern the person has repeated hundreds of times.

The order of the teeth, where the examination starts, the sequence of measurements, and where it ends have been repeated so many times that the sequence itself has become a map.

Often two people are involved. One examines and calls out numbers. The other enters them. The person entering the data may barely look at the screen while the sequence is running: 3... 2... 4... 3...

Put a dental chart on the screen and ask that person to locate each measurement visually, and you've handed an expert a second map to interact with. Now they have to reconcile the one in their head with the one in front of them.

That's work the corneal rater needed and the periodontal team doesn't.

How Much of the Map Do They Already Carry?

That's the question that helps determine the design.

•They already have it.

An experienced hygienist working through a tooth sequence they know cold: the anatomy, the order, and the rhythm of the examination all at once. Don't make an expert abandon a useful structure merely because the form can present a different one. Match what they already use and get out of the way.

•They have part of it.

Someone knows the anatomy or can identify the location but doesn't know the study's terminology, boundaries, or level of granularity. Or they know the workflow but not this study's categories, or where information normally appears but not how this study groups it. A reference image, a label, or a small amount of orientation may be enough.

•They don't have it.

A participant can point to where something hurts but doesn't know the anatomy. Or a coordinator may be transcribing from an image or report without the knowledge needed to orient themselves: the anatomy, the sequence, the terminology, or what belongs with what. Here, the form may need to provide more of the structure itself.

Get this wrong in either direction and you've added work. Impose a map on someone who already has one and they spend attention reconciling the two. Withhold one from someone who needs it and they spend attention translating.

The same source may need a different form depending on who's entering the data.

When Does This Person Look at the Form?

There's another thing we need to know about the person, and it's easy to miss unless you watch the work.

In a well-practiced workflow like periodontal data entry, the person may not monitor the screen continuously. They sample.

What they carry isn't only what comes next. It's when they expect to look.

They orient at the beginning: Am I in the right place?

Then the familiar sequence runs.

At a natural stopping point, they look again: Am I where I expect to be?

That's very different from someone checking the screen after every entry, and it suggests something important about form feedback.

Feedback isn't valuable simply because it's always visible. It needs to be available when the person uses it.

Give Them Landmarks

For periodontal entry, the characteristic error isn't necessarily entering an impossible number. It's getting off by one or two positions in the sequence. Once that happens, perfectly reasonable measurements can begin landing in the wrong fields.

And once you know that's the failure you're designing against, the form changes.

Keep the fields in the same sequence as the examination. Add visual outlines that allow someone to scan the sequence quickly. If there's a meaningful transition in the examination, give the eye somewhere to land there. A divider, spacing, or consistent change in color can create a landmark at the point where someone naturally reorients. The outline does more than mark a location. It marks a predictable point in the sequence.

Now someone who glances at the screen has places to check: Am I where I expect to be? If the sequence has drifted, the design may help them notice before they get much farther.

Notice how different this is from the corneal example. Spatial correspondence reduces the opportunity for a mapping error. Sequence outlines don't prevent drift. They make drift easier to see.

Different failures. Different design responses. Understanding how the task fails helps tell you which tool you need.

Give the Eyes Somewhere to Land

The same thinking applies when the person is looking at the screen continuously. Consider a matrix.

Read the question on the left. Move across several response columns. Choose an answer. Return to the left for the next question. Do that enough times and the rows start to look alike.

A characteristic failure is simple: the right response goes into the wrong row.

Icons in the left column can give each row a distinct anchor, so the eye returns to something recognizable rather than a wall of similar text. A head for understanding. An ear for hearing.

But only when the mapping is obvious. Look at the hand used for remember on that form. A hand doesn't mean memory to most people. An icon someone must decode has simply replaced one kind of work with another.

Again, the question isn't whether icons are good. It's whether these icons help this person do this task.

Sometimes the Form Should Match the Person, Not the Anatomy

Laterality provides another good example.

Suppose the subject is sitting across from the assessor. The assessor already works from that clinical presentation and has a familiar way of orienting right and left. We can arrange the form around that view.

Medial variables can appear toward the middle of the form, with lateral variables toward the outside. Right and left can be positioned the way the assessor encounters them rather than according to an abstract convention. Orientation may not be the only convention they bring, either: they may work through the structures in a habitual order, which is worth finding out rather than assuming. Color can further separate the sides so the association becomes easy to recognize after a few uses, particularly for someone who already has a well-established mental map.

The point isn't that every laterality form should use one orientation. Instead, preserve the orientation, sequence, and useful conventions the person already works from rather than making them mentally translate into a new one.

That's the lesson the corneal example couldn't teach us by itself. The source matters, but so does the person looking at it.

What Else Is the Person Doing?

So far, we've mostly looked at the work of getting an observation into the correct field. But some work is required by the source, some by the protocol, and some is created by the form.

The person may have to remember exactly how the study defines a variable. They may need to look something up. They may have to calculate something, determine a sequence, or follow a procedure before there's even a value to enter.

So here's the next question: How much of that work does the person need to do at all?

Put it to work

• Find out what they already know.

  What organization does this person already carry into the task? Consider anatomy, terminology, familiar locations, sequence, grouping, conventions, and routines.

• Watch when they look at the form.

  Do they continuously monitor the screen, or do they work through a familiar sequence and reorient only at certain points?

• Identify the characteristic failure.

  Are they likely to reverse sides, lose their place, drift by one position, put a response in the wrong row, or struggle to orient themselves?

• Design for that person and that failure.

  Preserve useful structures and conventions the person already relies on, provide landmarks where they reorient, and supply additional structure only where they actually need it.

Don't ask whether the form is intuitive. Ask whether it supports this person doing this task.