I am a great admirer of Tom Tullis’s work. He’s made important contributions to our understanding of screen design over the years (plus he is even older than me, which gives me great cause for hope!). In a recent poster at CHI 2009, in collaboration with Fidelity Investment colleagues Marisa Siegel and Emily Sun, Tom has done it again. He explained to me their important findings on the use of face images to attract attention on web or intranet pages. The results certainly run contrary to the perceived wisdom in this area (that humans are naturally attracted to facial images).
The Fidelity studies show that not only were users disinclined to look at facial images of the type shown in Figure 1, but that a significant number of users were unable to find the text immediately adjacent when given a task requiring that information.
Figure 1: Link panels with and without facial image (Study II)
In fact, overall performance was negatively affected by the presence of the face: accuracy dropped from 93% to 78%, mean time on task increased from 37 to 54 seconds and perceived task ease dropped from 4.2 to 3.4 (on a 5-point scale).
In a separate study, Tullis and colleagues not only found that facial images reduced performance; they also had the completely unexpected effect of reducing user confidence. Figure 2 shows the quite common approach that was tested (with and without the images). Again accuracy dropped, but this time by a smaller margin, from 94% to 87%. However, trust also dropped, from 6.3 to 6.0 on a 7-point scale. This result may seem small but there is only a 5% probability of it being due only to chance, which is taken by statisticians to mean that it was significant.
Figure 2: Authors’ images reduced trust in content (Study III)
So, contrary to what I remember being told in design seminars years ago, and even contrary to advice that I would have given up until I saw these studies (typified by my article on using images as links – http://www.syntagm.co.uk/design/articles/fitts50.htm) it looks like we need to be careful about how we use facial images in particular.
(You can see the full work-in-progress paper in the CHI 2009 proceedings, which will eventually be available from the ACM Digital Library – www.acm.org/dl)