Clinical Research Data Entry: Making the Form Do the Work

At a glance

Once you understand the work behind data entry, the next question is how much of that work the person should have to do.

Display the work

Sometimes the person needs to do the work and the form simply needs to make the task clear, organized, and easy to follow.

Support the work

The form can carry definitions, instructions, examples, landmarks, and other guidance so the person doesn't have to remember or find everything themselves.

Do the work

When a task can be calculated, populated, sequenced, or handled with logic, the form may be able to remove that work from the person's list entirely.

There's an assumption built into most data-entry forms: the person does the work, and the form records the result.

Once we start thinking about a form as part of the task rather than simply a collection of fields, that division of labor becomes worth questioning.

Some work requires human observation and judgment. Some exists because the protocol requires it. And some doesn't need to be done by the person at all.

Two opportunities are particularly useful: what someone must know and what someone must do.

What Someone Has to Know

Suppose you enter a particular variable once every six months. You vaguely remember how the study operationalized it, but not exactly.

You now have three options:

• Remember correctly.
• Leave the form and look it up.
• Remember incorrectly, confidently, and enter the value.

That third possibility is particularly troublesome because it may leave no obvious trace and the value may be a valid, possible response.

This is work created by the protocol. The source may be sitting right in front of you, but the study has defined how that source should be interpreted, categorized, or translated into data.

The form can carry some of that protocol knowledge. An informational image, a definition, or a worked example can appear where the decision is made rather than somewhere the person has to leave the form to find it.

That's particularly useful when entry is intermittent, training burden is high, or the judgment is difficult to remember. Weeks or months between participants give definitions time to fade. Staff turnover or multiple sites can increase the number of people who need to learn them. And an incorrect interpretation may still produce a perfectly plausible value.

The form becomes more than somewhere to record the answer. It helps the person produce it.

When Help Becomes Clutter

But help has a cost.

An informational image that's always on screen takes space, pushes fields apart, and adds something to look past every time the form is used. For someone entering data daily who learned the definition months ago, permanent help may no longer be help. It can become clutter.

That's where on-demand help becomes useful.

A small icon, link, or button makes the information available when someone needs it and keeps it out of the way when they don't. This can be particularly useful when the information is quickly learned, entry is frequent, the supporting image is large, or the form is already dense.

The principle is simple: put knowledge where it's needed. Don't make people process it when it isn't.

The same information that helps an inexperienced or occasional user may become clutter for an expert who enters it every day.

What Someone Has to Do

Providing information still leaves the work with the person. The form is helping them do it.

But sometimes there's a more interesting possibility: the form can do part of the work itself.

Consider an animal study in which reward position was counterbalanced across twenty-five trials. Each trial had a configuration that had to be determined, set up, performed, and recorded.

Without help from the form, someone might need to:

• find the counterbalancing schedule
• determine the trial's configuration
• set it up
• hold the configuration in mind
• run the trial
• record the outcome

That's a lot of work surrounding the thing the researcher wants to observe.

The form was built so that selecting the sheet number automatically set the corresponding counterbalanced position information. The variables for each trial were grouped in the order the work was performed.

Now the sequence becomes much closer to this:

Here's the configuration. Run the trial. Record the outcome.

That isn't simply a different layout. The form has taken a job off the person's list, and with it an entire set of opportunities for error.

Display It, Support It, or Do It

There's a useful progression here.

•Display it.

Give the person the fields and make the work clear. The person still performs the task.

•Support it.

Preserve useful structure, match learned routines, provide landmarks, and put definitions or instructions where they're needed. The person still performs the task, but the form reduces the surrounding work.

•Do it.

Calculate, populate, sequence, apply logic, or otherwise perform a task the person should not need to do manually. The form removes work from the person's list.

Put it to work

• Identify the work.

  What does the person actually have to know, remember, find, interpret, calculate, or do before the value reaches the field?

• Ask why the work exists.

  Is it required by the source? Required by the protocol? Or has the form itself created it?

• Decide whether to display, support, or do it.

  Should the person perform the task with a clear interface? Does the form need to provide guidance? Or can the form remove the task entirely?

• Put help where it's needed.

  Provide definitions, instructions, examples, or other guidance at the point of decision, and keep optional help out of the way when it isn't needed.

Every task the form can safely remove is one less task the person has to remember, perform, and get right.

---

Series Summary

We started this series with a room full of people who had plenty to say about what data should be collected. The harder question was what the forms should look like and how the data should get into the forms.

Now we have a better way to answer it.

The source may contain relationships, organization, and cues worth preserving: where information appears, what appears together, and the order it comes in. The task may contain work required by the source, work required by the protocol, and work created by the form itself. And the person arrives with expertise, routines, expectations, limitations, and attention that the form can either leverage or fight.

Traditional validation still matters. We still want to know whether a value is allowable, complete, and consistent. But accurate data entry starts before the value reaches the field.

In the end, it doesn't mean every form should become more visual. Sometimes the best form may be remarkably plain. The point is to make the design fit the source, the task, and the person doing the work.

Sometimes the best change is moving a field. Sometimes it's putting information exactly where the person needs it. Sometimes it's designing around what the person already knows. And sometimes it's deciding they shouldn't have to do the task at all.

Design for the work, not just the fields.