What's the Evidence? Are headings useful?

August 2016

Earnsy Liu, TechCommNZ member and GDID student, looks for evidence (not just opinions) to help you manage the daily conundrums we face in our profession. If you have a question for Earnsy to tackle, please email our Comms Coordinator.

I can’t bring myself to read the articles by a particular financial commentator. He is knowledgeable and respected, with a successful business to prove it – but he doesn’t use headings. His articles are a daunting wall of text. There’s nothing to help me work out what they are about, nor anywhere to stop and think about what I’ve just read. So these days I simply give his articles a wide berth.

Am I the only one who likes headings? I mean names of sections within a document, not headlines or titles. And what good are they really? Apparently they act as signposts, break up text into manageable chunks, and enable readers to scan to get an overview and find information.

Do we know for sure that they are useful? And what makes good headings? Let’s look at evidence from the last 10 years, restricting ourselves to words rather than design elements such as font size or placement.

Headings are useful

Well, headings do get read – that’s been confirmed in eye tracking studies (Hyöna & Lorch, 2004; Hyöna & Nurminen, 2006). Some readers spend more time on them than others do, though: they don’t just read them linearly, but even have a second look before moving on to the next section.

Did you know headings are helpful even for informal product reviews? Over 800 Survey Monkey experiment participants read 11 reviews of everyday products, such as a camera, a tent, and clothing. Reviews with headings were perceived as higher quality than those without. Examples of headings were, in the case of make-up, ‘texture and application’, ‘finish’, and ‘coverage’ (Mackiewicz & Yeats, 2014).

Headings aren’t only useful in English-speaking cultures either. The eye tracking studies above were carried out in Finland. A Dutch study found headings sped up searches (Kools, Ruiter, van de Wiel, & Kok, 2007) – more on this shortly.

Headings help recall

Readers remember more when there are headings. In an eye tracking study, Hyöna and Lorch (2004) gave 66 undergraduates texts on energy and on endangered species, both on a screen. Each participant read a text with headings – single noun phrases like ‘pandas’, ‘wind powers’– and one without, then summarised the texts. Where there were headings:

  • summaries were more complete, although there was no difference in the order of recall
  • topic sentences were read more quickly (a topic sentence is the first sentence of a paragraph that introduces a topic)
  • readers spent less time looking back at topic sentences.

What’s a useful heading?

Several factors seem to contribute to usefulness, including meaningfulness, frequency, length, and using key words up front.

Meaningfulness

One study suggests meaningfulness is key. Kools et al gave 80 university students a letter on healthy eating (five to ten A4 pages) with one of four versions of headings:

  • original headings (fewer headings but more levels)
  • Information Mapping [1] headings, with information in the original order
  • Information Mapping headings, with information in a different order
  • no headings.

After scanning the text for a minute, participants had to find 10 terms in the texts. Those with the no-headings version took longest, but there was no difference between the other groups. The authors suggest that as long as headings are meaningful, users will be able to find what they need. However, they caution that more research is needed, and that the speed may be different with longer texts.

The findings surprise me because there were actually several other differences: order of information, number of pages, number of heading levels, and layout. Did they offset each other, I wonder? And what do the findings mean: how often do users only search for words – aren’t they usually reading to learn or reading to do?

Frequency

Headings every 200 words seem to be useful, as suggested by three experiments on a total of several hundred participants (Schultz & Spyridakis, 2004; Bartell, Schultz, & Spyridakis, 2006). I’ll describe the experiments together as they are similar.

Participants read texts on arthritis, then answered 15 multiple-choice questions. The texts had different heading frequencies:

  • none (control condition)
  • high (every 100 words)
  • medium (every 200 words)
  • low (every 300 words).

As shown below, two of the three experiments suggested medium frequency headings were best for understanding online texts. Frequency didn’t seem to affect understanding of print texts much, with comprehension scores midway between the best and worst scores of the online groups.

Experiment Text giving best comprehension scores
2004 experiment 1 (online only) High frequency
2004 experiment 2 (online only) Medium frequency
2006 experiment (print and online) Print: Similar across frequencies
Online: Medium frequency

But I have several reservations about these studies; I’ll let you in on two. First, the articles only had one level of headings. What frequency should we use for more levels? Second, and more important, the standard deviations [2] of their findings – how much comprehension scores varied – were quite large. The scores of the different groups would have overlapped, so are the results as clear-cut as we’d like to think? Check out Jarrett’s (2007) Good headings help, bad headings hurt for a more detailed discussion on the print vs online study.

Length

Short headings are more useful. We read in an F-pattern online – across the page initially, then scan a column down the left – so keep headings and lists to approximately 11 characters to cater for reading down the left. ‘Users typically see about 2 words for most list items; they'll see a little more if the lead words are short, and only the first word if they're long’ (Nielsen, 2009).

In one task, 80 participants were given links truncated to 11 characters and asked what they thought they would find if they clicked the links. In another task, they had to figure out which link to click on to find particular information. The results showed that the best links were specific, in plain language, and started with key words (Nielsen, 2009):

Characteristics Examples
Best Were specific
Used plain language and naming conventions

Started with key words

Gift Cards (Gift Cards & E-Gift Certificates)

New custome (New customers apply online now)

Worst Were general
Used made-up terms
Did not start with key words
Introducing (Introducing Chase Exclusives Special Benefits for Checking Customers)
Working whi (Working while you study: paying tax)

Profit Acce (Profit Accelerator Overview)

Although the research was done on links, Nielsen tells us similar principles apply to headings. The short headings in the product review and energy-endangered species studies above may be testimony to this.

Fortunately we don’t need to count off [3] exactly 11 characters when writing headings in ‘real life’, because users can read more if they wish.

Key words up front

Another factor is starting with key words. Redish (n.d.) recommends starting with a verb. In her unpublished study from the 1980s, when headings all began with ‘how to’, users ‘could not easily find the right section because their eyes stayed on the "how to" and all the headings seemed the same’. In contrast, beginning with verbs allowed users to ‘see the different actions right away’.

Not helpful Helpful
How to get a permit
How to fill out the permit

How to change your address

Getting a permit
Filling out the permit

Changing your address

Blind users appreciate headings that start with key words too. Between November 2002 and February 2003, Theofanos and Redish (2006) observed 16 blind users who used screen readers to read websites. They found that just as sighted readers scan with their eyes, blind readers ‘“scan with their ears”, listening to just enough to decide whether to listen further’, rather than reading every word.

Redish feels ‘Although both of these studies are more than 10 years old, I would strongly argue that the results related to headings are still valid’ (Redish, G., personal communication, June 30, 2016). Given that Nielsen’s more recent study (above) also found it was better to start with key words, I’d agree.

Suggestions on writing headings

So research suggests it is worthwhile putting effort into writing good headings, specifically:

  • make them meaningful
  • use them regularly
  • keep them short and specific
  • start with key words.

Redish recommends ‘it is best to write the headings first, as an outline, and then fill in the text under the headings’ (personal communication, June 30, 2016). She shares useful tips in her article Headings. Perhaps I should tell that financial commentator about it.

Acknowledgements

A big thank you to Janice (Ginny) Redish for her helpfulness and useful tips.

References

Bartell, A. L., Schultz, L. D., & Spyridakis, J. H. (2006). The effect of heading frequency on comprehension of print versus online information. Technical Communication, 53(4), 416-426. Retrieved February 25, 2016, from Ingenta Connect.

Hyöna, J., & Lorch, R. F. (2004). Effects of topic headings on text processing: Evidence from adult readers’ eye fixation patterns. Learning and Instruction, 14(2), 131-152. Retrieved June 26, 2016, from Research Gate website: https://www.researchgate.net/profile/Jukka_Hyoenae/publication/222136283_Effects_of_topic_headings_on_text_processing_Evidence_from_adult_readers'_eye_fixation_patterns/links/0912f50ceee507ec44000000.pdf

Hyöna, J., & Nurminen, A.-M. (2006). Do adult readers know how they read? Evidence from eye movement patterns and verbal reports. British Journal of Psychology, 97(1), 31-50. Retrieved from Academic OneFile.

Jarrett, C. (2007). Good headings help, bad headings hurt. Retrieved June 30, 2016, from Usability News website: http://usabilitynews.bcs.org/content/conWebDoc/48173

Kools, M., Ruiter, R. A. C., van de Wiel, M. W. J., & Kok, G. (2008). The effects of headings in information mapping on search speed and evaluation of a brief health education text. Journal of Information Science, 34(6), 833-844. Retrieved February 25, 2016, from Research Gate website: https://tinyurl.com/Koolsetal2008

Mackiewicz J., & Yeats, D. (2014). Product review users' perceptions of review quality: The role of credibility, informativeness, and readability. IEEE Transactions On Professional Communication, 57(4), 309-324. Retrieved February 25, 2016, from http://jomack.public.iastate.edu/Mackiewicz%20and%20Yeats%20-%20Product%20review%20users'%20perceptions%20of%20review%20quality.pdf

Nielsen, J. (2009). First 2 words: A signal for the scanning eye. Retrieved June 28, 2016, from Nielsen Norman Group website: https://www.nngroup.com/articles/first-2-words-a-signal-for-scanning/

Redish, G. (n.d.). Headings. Retrieved June 27, 2016, from Plain Language.gov website: http://www.plainlanguage.gov/howto/guidelines/headings.cfm#6

Schultz, L. D., & Spyridakis, J. H. (2004). The effect of heading frequency on comprehension of online information: A study of two populations. Technical Communication, 51(4), 504-516. Retrieved from Academic OneFile.

Theofanos, M. F., & Redish, G. (2006). Guidelines for accessible and usable web sites: Observing users who work with screen readers. Retrieved June 28, 2016, from Redish & Associates website: http://redish.net/images/stories/PDF/InteractionsPaperAuthorsVer.pdf


[1] An approach to structuring documents by applying a set of principles and techniques. For an overview, see http://www.technicalauthoring.com/wiki/index.php/Information_Mapping

[2] Standard deviation indicates how similar things are. If we were all a similar height, the standard deviation of our heights would be small. Conversely, if they varied greatly, the standard deviation would be large.

[3] To count characters in MS Word: highlight the text, go to the Review tab, and click Word count.