Ensuring accurate data entry is central to all research projects. Various data entry validation techniques have emerged in response to this need. This article examines visual validation: the assignment of meaning to the layout of a form in a manner that facilitates accurate data entry. Examples are provided from various research projects.
Data Entry Validation
Consider the following scenario, a rater is presented a series of corneal staining images and asked to assess the extent of pathology in each quadrant and center area of the cornea. It is easy to appreciate the strength of the Visual Validation technique by examining the two alternate form designs below. Form A maps the data entry controls (i.e., pick lists) directly onto a black and white line sketch that matches the spatial layout of the presenting image and corresponding quadrants / center area to assess. In contrast, Form B merely alternates OD vs. OS variables in separate rows, one variable per row. Clearly, the approach used in Form A is much more likely to reduce data entry errors.
Accurate data entry is critical to ALL research projects. Indeed, a number of articles and discussions have covered various data validation techniques and related features, the most frequently covered being: Skip Logic, Range Checks, Pre-Specified Pick Lists, Automated Calculations, and Queries. But don’t blindly bet on what ‘fingers’ can do, the eyes are important too! Although less frequently discussed, Visual Validation is one of the most powerful techniques and a critical consideration when designing data entry forms.
To clarify the intent of this article, basic principles of form design that could be categorized as ‘visual’ and that are important to accurate data entry are not the focus. Some examples of these basic principles include the use of: adequate font size, white space, text alignment, headings, etc. Instead, this article is about integrating the variable data definition with the spatial layout of the form in a manner that facilitates accurate data entry.
So what’s Visual Validation all about and why is it so effective? The essence of the visual validation approach is to assign meaning to the spatial layout of the form. That is, the layout of the form itself becomes part of the data entry verification process. Ideally, the form layout aligns with the data definition of the variables. So what’s the best way to determine the form design? Well, start by considering the characteristics of the data entry person and their preferred methods.
Who’s Entering Data and How?
Heads Down vs. Up Approach
Knowing whether or not data entry is done primarily while looking at the screen is a key form design factor. For instance, dental research staff are accustomed to data entry in a specified order matching of the teeth, and generally don’t look at the screen until the end of a sequence (e.g., all Lingual data points entered). Thus, overlaying variables onto an image of teeth may not be helpful, and may even slow things down. In such cases, the best approach is to arrange the variables in the appropriate specified order with visual outlines that allows for quick vertical visual inspection through the sequence (see below). The efficient visual scan addresses the most common error of being off sequence by one or two variables.
In contrast, other scenarios require close inspection of the screen as a part of the data entry process (see corneal staining ratings mentioned above). In those cases, placement of variables in a manner that maps onto a corresponding image is a must for accurate data entry.
If data entry staff are familiar with location information (e.g., MDs entering data), a simple one row per variable approach without an image map is often sufficient. The next step up is to integrate a reference picture into the form along with the simple one-row-per variable layout (see below).
The highest level and most integrated step is to map the data collection controls directly onto a corresponding image. This approach is ideal when the person entering data may NOT be familiar with the formal location nomenclature (e.g., left gastrocnemius), but is privy to precise location information. Subject data entry is a good example. The form below is from a pain study whereby subjects check the areas associated with pain.
Similarly, technical staff may be familiar with a broad range of location specifics, but not know what’s needed for any given study. That is, the level of granularity defined by the study variables and the associated map depend on the aims of the study. For instance, staff may be familiar with a wide range of electroencephalogram (EEG) lead placement locations, but the operational definition for a specific study might only be ‘right’, ‘midline’ or ‘left’ seizure onset. In such cases, mapping controls directly onto an image is helpful. Below is an example from an epilepsy study.
Types of Visual Validation
Once the characteristics of the data entry person and the preferred methods are fully appreciated, the next step in determining how best to leverage Visual Validation in the form design. To do this, it is important to be aware of the various approaches/techniques and solutions to common issues.
Direct placement of data entry controls onto an image is a frequently used Visual Validation technique and several examples have been presented above. Note that sometimes the data definition doesn’t conveniently map onto an image. For example, variables with an “Other” category often present an issue to be resolved. A common solution is to use a text box that specifies the needed category(ies) (See the grayed text box “Other” below).
When the data definition contains a lateral component it is nearly always best to arrange the form using a left/right configuration. The orientation is typically based on clinical presentation (i.e., as if subject facing forward; see below).
A left/right configuration can be used even when the data definition involves medial variables via placement of these variables between lateral variables. The use of color also helps segment the areas and allows the user to develop a color-side association (see below).
As the saying goes, ‘A picture is worth a thousand words.’ Image placement within a form can be particularly helpful for informational / training purposes. For instance, cases where the benefits of embedded informational images are highest include those when data entry is expected to be intermittent (e.g., weeks or months between study subjects), or when there is a heavy training burden (e.g., high staff turnover rate, large number of staff to train [multi-site trial]) (see below).
In yet another variation, it is sometimes helpful to have a mechanism to present an informational image “On-Demand”. Being “On-Demand” the image is kept out of the way when not needed, thus facilitating data entry (i.e., more of the form is visible). To make the image appear, typically the user clicks a link, button or icon (see below). Situations that work well for an “On-Demand” informational image include:
Fast Learning: Images for which the information is quickly grasped and remembered
High Volume/ Consistent Data Entry: When staff are accustomed to a high volume or frequent/consistent data entry (e.g., daily) the utility of an image taking up space on a form diminishes over time.
Large Image: Images that are large can disrupt the flow of data entry.
Dense Form: A form with a large number of densely packed, grouped variables where space is a premium.
Procedural Oriented Layout
An image is not the only way to leverage spatial information on a form. Procedural order can also be conveyed via the form layout. For example, an animal research study counterbalanced reward (i.e., fruit) position placement in a box across 25 trials. As seen below, the variables captured for each trial are grouped horizontally (only first 4 trials shown).
Side note, logic was programmed into the procedural form above such that selecting the Sheet Number automatically set the corresponding counterbalanced position information.
Taken together, these visual validation techniques are a critical part of form design to ensure research data are accurately entered. All research projects should give careful attention to the use of Visual Validation techniques in their electronic data capture system to increase the likelihood of success.