Getting all emotional
Bipolar Emotional Response Tests, or BERTs (not to be confused with other Berts) are a method for measuring people’s subjective emotional opinions about websites – or anything else for that matter – along a range different dimensions. For example, you might ask someone to rate a site on a scale from Professional to Amateur. Or Friendly to Impersonal. Or any paired responses that you might be interested in. If you do this for a number of different pairs, you can effectively create an “emotional fingerprint” of how your users or clients perceive a site. I first became aware of BERTs a few years ago when I saw them used in the The Glass Wall, the BBC’s fantastic document about how they redesigned bbc.co.uk. Unfortunately, this document isn’t officially available on the web (although it is out there if you know where to look for it). On the few occasions I’ve used them, I’ve found BERTs to be extremely useful, yet for some unfathomable reason they don’t seem to be used very often in web development projects.
BERTs for users & clients/stakeholders.
BERTs are a surprisingly good tool to use as part of a stakeholder analysis. You can get stakeholders to rate competitor sites, or even come up with a hypothetical “desired” emotional fingerprint for the site that you intend to design. In this way, BERTs can contribute to the overal design strategy for the site. This is very much a “big picture” approach, and doesn’t consider specific bits of functionality, but it complements other approaches (e.g. questionnaires/interviews) extremely well. You’re also collecting quantitative data, so if you get a broad range of BERTs it’s easy to see what the central tendencies are for each bipolar combination and where there is consensus or disagreement among the stakeholders.
As well as contributing to a design vision, BERTs also provide invaluable baseline data. It’s possible to plot the clients’ desired emotional fingerprint for the site against their (or the users) actual responses to the finished (or prototype) site. I did this for the first time recently and was staggered to get a correlation of 0.93. In other words, the project delivered pretty much exactly what they asked for in terms of the emotional perception of the site. Lots of client brownie points!
You can also use BERTs track user perceptions of your prototypes as they iterate. In this way, you can potentially plot a trajectory for your emotional design, or compare your results to those obtained from testing competitor sites.
There are also additional benefits. These include:
- Speed and cost. In some cases, BERTs can be very quick and very cheap. In fact, I can’t think of a quicker or cheaper method for user research!
- They’re (relatively) simple to write and understand. It’s very easy to write a piss-poor user survey or questionnaire; even the BBCs latest online survey contains more schoolboy errors than you can shake a stick at (apologies to intelligent schoolboys everywhere). By virtue of their simplicity, BERTs appear to be a little more robust. There aren’t that many meaningful pairs of responses that you can come up with, so the more you do BERTs the easier it ought to get. However, you still need to be sure that you (i) choose sensible scales that cover all the emotional attributes that are important, which is perhaps easier said than done, and (ii) make sure that the bipolar extremes are easily understood by your users (as in all things involving real users, a little preliminary testing goes a long way).
Aim to standardize them as best you can.
Everyone who is tested needs to have some familiarity with the site. For naive users (in the sense of those never having seen the site before) it’s important to do some sort of basic test or walkthrough to expose them to the site’s content and functionality and design in a standardized way. In the case of stakeholders, you’d hope that they’d all be pretty familiar with the site, but in my experience people tend to only be concerned with one or two aspects of the site. It never hurts to encourage them to refresh their memories of the whole site.
In short, BERTs have the potential to be a really useful and relatively inexpensive tool in web development projects. I’ll certainly be using them more in the future, and it’ll be interesting to see whether they become more widely adopted in the user experience community in the future.