When should we use visuals?

November 2018

Facts icon: a sphere made up of words like: info, science, answers, reality etc. The word FACTS is displayed in bold orange capitals across the middle of the sphere.

Right hand holding screw. Left hand screwing it in with screwdriver with black handle.

Image by 947051 (CC0).

To assemble that chair, start by inserting screw a6 into b1 while holding down c. Got that?

No? Why not? It’s a short sentence (15 words) and there are no big words. Would visuals help? When are they a good idea, when not? What do users prefer?

Let’s look at the evidence on the use of visuals, compare text and visuals, peek at a comic-style employment contract, and finish with some tips (Figure 1).

Words showing the first- and second-level headings in the article. The four first-level headings are: evidence, text or visuals, visual contract, tips. The four second-level headings belong to the ‘evidence’ first-level heading. They are: when visuals work, when visuals don’t work as well as text, simpler is better, how they help.

Figure 1: Structure of this article

By the way, ‘visuals’ in this article refers to static images such as drawings, flowcharts, and photos. It’s used interchangeably with ‘graphics’ and excludes film and videos.

What’s the evidence?

There’s evidence that visuals often work — but also that they sometimes work less effectively than text. There’s evidence that simpler visuals work better. And we have an insight into how they work.

When visuals work

A redesigned procurement guide, a contract with visuals, and procedures with photos, show us visuals can work well.

Redesigned procurement guide

Finnish authorities wanted specific terms and conditions incorporated in procurement contracts (Passera, 2017). Unfortunately, staff didn’t understand the terms and conditions, which led to ineffective procurement and misunderstandings, which in turn ‘resulted in defective service provision, claims, monetary loss, and, above all, citizens unhappy with public services’ (Passera, 2017, p. 9).

So the original 20-page text document was redesigned as a 28-page visual guide with one diagram per page. All text was integrated into the diagrams, and icons provided further visual cues. Three types of diagrams were used:

  • swimlane diagrams to show roles, rights, and responsibilities
  • flowcharts to show possible courses of action and their consequences
  • timelines to show possible scenarios for ending a contract.

To test the visual guide’s effectiveness, 72 staff actively involved in procurement were given ten comprehension questions. Five were to be answered using the original text document, five using the new visual guide.

The result? The visual version performed better. Not only were answers were quicker and more accurate (Figure 2), new and experienced users found it ‘more pleasant, functional, and less stressful to use’ (Passera, 2017, p. 22).

Two line graphs side by side. The one on the left shows the time taken to answer questions. The visual version took 819 seconds, less than the 991 seconds for the text version. The graph on the right shows the accuracy of answers. Answers based on the visual version scored 4.17 points, more than the 3.63 points for the text version.

Figure 2: Answers based on the new visual procurement guide were 17 per cent faster and 15 per cent more accurate than those based on the original text document (adapted from Passera, 2017).

Visually enhanced contract

Visuals can improve contracts too. Researchers asked six comprehension questions about a business-to-business purchasing agreement in English (Passera, Kankaanranta, & Louhiala-Salmien, 2017). Slightly over half the 122 international respondents used the traditional text-only version, while the others used a version that was visually enhanced with diagrams. Native English speakers made up 58 per cent (71) of the respondents. I can’t embed the visuals here, but do have a look at them on pp. 63–66 of the article .

Overall, the visual version was more effective (Figure 3), with those using it being:

  • faster, regardless of their native language
  • more accurate, especially if they were non-native English speakers.

Two line graphs side by side, similar to those in figure 2. The one on the left shows the time taken to answer questions. The visual version took 897 seconds, less than the 1,080 seconds for the text version. The graph on the right shows the accuracy of answers. Answers based on the visual version scored 4.35 points, more than the 3.01 points for the text version.

Figure 3: Those using visually enhanced contracts were 3 minutes faster over the six questions, or half a minute per question. Their answers were 1.3 points more accurate, out of a possible total of 6 points (adapted from Passera, Kankaanranta, & Louhiala-Salmien, 2017).

Because non-native speakers using the visually enhanced version were much more accurate, the researchers suggest that such contracts could be ‘highly beneficial in international business’ (Passera, Kankaanranta, & Louhiala-Salmien, 2017, p. 137).

The authors note several limitations to this research though, including:

  • Not all visuals helped. The authors recommend more user research to see what works.
  • Participants were not ordinary folk like you and me. They had responded to a request for help in the International Association for Contract and Commercial Management’s newsletter, meaning they were interested in contracts and motivated to volunteer. This might affect how generalizable the results are.

I too have a reservation: the visual version wasn’t just different because it had diagrams. It also had wider margins, white space, and key information in bold. Sure, the diagrams would have made a significant difference, but how much did the improved layout help?

Procedures with photos

However good visuals are, they aren’t the answer to everything. Communication experts van Hooijdonk and Krahmer (2008) had 30 adults perform ten easy and ten difficult exercises to prevent repetitive strain injury (RSI). The participants could practice an exercise before actually executing it and moving on to the next one. Instructions were given either as short text (average 27 words), pictures (single photos), or short film clips (average 4.4 sec).

Results showed that participants who had seen pictures (Figure 4):

  • learnt fastest
  • practised about a fifth of the exercises (more than participants in the film condition but less than those in the text condition)
  • executed the exercises quickest
  • executed a similar number of exercises correctly to the film condition for easy exercises
  • executed a similar number of exercises correctly to the text condition for difficult exercises.

The authors concluded:

  • no one medium was superior in every way
  • pictures were most efficient for learning and execution times
  • pictures fared extremely well for easy exercises. It might therefore be ‘particularly efficient’ to illustrate complex exercises with ‘a series of pictures depicting key stages of the procedure’ (van Hooijdonk and Krahmer, 2008, p. 59).

Four line graphs. The top left graph shows learning speed. Participants in the picture condition learnt fastest and participants in the text condition learnt slowest. The top right graph shows amount of practice. Participants in the text condition practised about half the exercises, those in the picture condition practised about twenty per cent, and those in the film condition practised almost nothing. The bottom left graph shows execution time. The picture condition was fastest and the text condition was slowest by far. The bottom right graph shows the number of correct exercises. For easy exercises, participants in the picture and film conditions performed similarly, and were better than those in the text condition. For difficult exercises, however, text and picture participants performed similarly. They did worse than on the easy exercises, and worse than film participants. Conversely, film participants did about the same as on the easy exercises.

Figure 4: While no medium was perfect, the picture condition had ‘nearly optimal results for easy exercises’ (van Hooijdonk and Krahmer, 2008. p. 59). Participants shown pictures learnt fastest, practised significantly less than those given text, and executed exercises quickest (adapted from van Hooijdonk and Krahmer, 2008). There were ten easy and ten difficult exercises.

The researchers believed it took longest to learn using text because:

  • ‘it takes more mental effort to read a text than to look at a picture or watch a film clip’ (van Hooijdonk, 2008, p. 58)
  • text participants practised substantially more.

Oh, guess which medium participants liked most? Not photos. Not film – They didn’t have a clear preference.

When visuals don’t work as well as text

Visuals aren’t always better than words. Dutch health scientists Veldwijk et al (2015) sent a questionnaire about vaccination to parents of new-borns. They asked nine questions twice, once as words and once as graphics, making 18 questions in all.

Most respondents said words and graphics were both understandable and clear (Figure 5) …

A line graph showing whether tasks were understandable and clear. Most participants found both words and graphics understandable and clear, but ratings were higher for words.

Figure 5: Although most of the 959 respondents said words and graphics were understandable and clear, graphics seem less so (adapted from Veldwijk et al, 2015).

… but preferred words (Figure 6).

A column graph showing whether tasks that were clearest and easiest used words or graphics. Between 50 and 60% of participants felt word tasks were clearest and easiest. Approximately 10% favoured graphics tasks. About one third felt there was no difference.

Figure 6: Asked which condition was clearer and easier, 5½ times as many respondents said words were clearer, while 6½ times as many said words were easier (adapted from Veldwijk et al, 2015).

Interestingly, only 13 per cent answered both word and graphics questions consistently, and some were inconsistent with up to eight answers. Those who saw graphics first were significantly more inconsistent. (This makes me wonder if the words and graphics sent different messages. Hold this thought — we’ll revisit it towards the end.)

So words were better. To be honest, I found the graphics hard to understand. Look at them here and see what you think.

Simpler is better

You know the keep-it-simple mantra? It applies to visuals too.

Psychologist Butcher (2006) demonstrated this in two experiments. She started by assessing 74 undergraduates on their existing knowledge of the heart and circulatory system. Next, she gave them information on the topic in text, text with simplified diagrams, or text with more detailed diagrams. Finally, she assessed what they had learnt. (‘Simplified’ and ‘detailed’ are not absolute descriptions of the diagrams but how they compare to each other, as shown on p. 3 of the article ).

She found (Figure 7):

  • participants given simplified diagrams learnt more
  • simplified diagrams worked better than detailed diagrams, especially for those with less prior knowledge
  • detailed diagrams helped develop mental models, but were no different to text-only information on other measures
  • those with high prior knowledge did not benefit more from detailed diagrams
  • the condition made no difference to inference.

Four column graphs. The top left graph shows mental model improvement. Participants shown simplified diagrams improved most and those shown text improved least by far. The top right graph shows general knowledge improvement. Simplified diagram participants improved most. Text and detailed diagram participants improved less, and about the same as each other. The bottom left graph shows memory test scores. Simplified diagram participants had the highest scores. The bottom right graph shows inference question scores. There was little difference between the participants in the different conditions.

Figure 7: Simpler diagrams facilitated learning best. Participants given such diagrams improved their mental models and general knowledge more than participants who were given text or detailed diagrams. They also remembered more and made more inferences. In contrast, those given detailed diagrams performed like those given text, except that they improved their mental models more (adapted from Butcher, 2006).

How they help

What is it about visuals that improve understanding? Butcher (2006) performed a second heart function experiment. It was like the first, except the 34 undergraduates were asked to self-explain their learning. An example of self-explanation is: ‘Well, it says that the blood that empties into the right atrium is dark, because it picks up the carbon dioxide from the blood cells. So, I’m guessing it turns it dark’ (Butcher, 2006, p. 196).

Learning results were similar to the first experiment. But the self-explanation showed that those given diagrams inferred more and integrated inferences better. Those given simplified diagrams made the most integration inferences (Butcher, 2006; Figure 8).

A line graph showing the number of inferences made. Simplified and detailed diagram participants made a similar number of inferences, more than twice the number that text participants made.

Figure 8: Diagrams seem to summarise information and provide additional learning support, enabling learners to infer information (adapted from Butcher, 2006). I think the number of integration inferences are included in the number of self-explanation inferences.

The author concluded that ‘diagrams provide a type of cognitive support’ (Butcher, 2006, p. 194), which may explain why participants given diagrams — especially simplified diagrams — learnt more.

So … text or visuals?

Reading text requires more skill and effort, and involves linear processing. In contrast, visuals:

  • communicate information like orientation and location more easily (cited in van Hooijdonk, 2008)
  • break up text and reduce complexity (cited in Passera, Kankaanranta, & Louhiala-Salmien, 2017)
  • help readers create mental models of themselves carrying out procedures (cited in Passera, Kankaanranta, & Louhiala-Salmien, 2017)
  • help readers develop mental models when learning (Butcher, 2006)
  • explain steps, action sequence, interactions, and components of systems more effectively than text (cited in Passera, 2017).

But as we’ve seen, visuals can’t do everything. Research shows integrating text and visuals helps readers (cited in Passera, Kankaanranta, & Louhiala-Salmien, 2017):

  • understand better
  • complete tasks faster
  • learn to use devices faster
  • feel more positively about instructions.

A visual contract

By the way, engineering and infrastructure consultant Aurecon introduced comic- style employment contracts in New Zealand in August. They explain why and how in a 2-min video .

Although the company involved lawyers in the new contract, other lawyers warn of risks of consistency, interpretation, and perhaps discrimination. (Back to the possibly inconsistent text vs visual messages earlier – do those findings justify this fear about inconsistency?)

Tips

Tempted to use diagrams? Before you get your coloured pencils out, here are a few tips:

Twenty-six coloured pencils lined up neatly in a row.

Image by Bru-nO (CC0)
  • Ensure visuals are appropriate and useful (Passera, Kankaanranta, & Louhiala-Salmien, 2017). For example, diagrams may need to highlight key relationships or encourage inference (Butcher, 2006).
  • Choose the right medium for your purpose. For example, van Hooijdonk and Krahmer (2008) suggest with RSI exercises:
    • To encourage practise, use text.
    • To enable quick learning, use pictures.
    • To facilitate accuracy (correct task execution), use film clips.
  • Keep things simple. Don’t try too hard to make visuals realistic, because ‘adding realistic detail almost always increases diagram complexity’ (Butcher, 2006, p. 183).
  • Involve lawyers and experts when drawing diagrams for contracts. When they worked on the visual terms and conditions, they uncovered mistakes that were hard to spot in text. Plus, ‘lawyers creating rough sketches as raw materials for designers’ sped things up ‘considerably’ (Passera, 2017, p. 25).
  • Prioritise documents that ‘will be used, unchanged, for a long period of time’ (cited in Passera, 2017) when investing in good design.

Happy visualising.

References

Butcher, K. R. (2006). Learning from text with diagrams: Promoting mental model development and inference generation. Journal of Educational Psychology , 98(1), 182–197. Retrieved from https://pdfs.semanticscholar.org/fd4c/2c51c23e1e61404c1ac670bf6a20f4394b5a.pdf

Passera, S. (2017). Flowcharts, swimlanes, and timelines: Alternatives to prose in communicating legal–bureaucratic instructions to civil servants . Retrieved from https://pdfs.semanticscholar.org/7ddc/b142452465c755b34c2f0aed7987be5ac87f.pdf

Passera, S., Kankaanranta, A., & Louhiala-Salminen, L. (2017). Diagrams in contracts: Fostering understanding in global business communication. IEEE Transactions on Professional Communication, 60(2), 118–146. Retrieved from IEEE database.

van Hooijdonk, C., & Krahmer, E. (2008). Information modalities for procedural instructions: The influence of text, pictures, and film clips on learning and executing RSI exercises. IEEE Transactions on Professional Communication , 51(1), 50–62. Retrieved from IEEE database.

Veldwijk, J., Lambooij, M. S., van Til, J. A., Groothuis-Oudshoorn, C. G., Smit, H. A., & de Wit, G. A. (2015). Words or graphics to present a discrete choice experiment: Does it matter? Patient Education and Counseling , 98(11), 1376–1384. Retrieved from https://www.sciencedirect.com/science/article/pii/S073839911500275X