Getting all emotional with BERT

Bert from Sesame Street

Bipolar emotional response testing (BERT) – sounds like some sort of psychological examination technique right? Well yes it is loosely based on the repertory grid (which is a psychology interviewing technique) but it is also a damn useful means of capturing subjective design feedback from users and is especially useful for comparing different designs. Want to find out which of your 3 designs is more effective? BERT can help. Want to capture which of your competitor sites perform the best? BERT can help. Want to impress your friends and family with an amusingly named user research technique? Most certainly BERT can help!

How does BERT work?

Like all good research techniques BERT is actually very simple. All you do is identify some facets of your design that you’re interested in capturing people’s opinion of (usually about 7-10) and ask them to rate where they feel the design falls between 2 polar extremes of that facet (hence the bipolar part of BERT). For example, if you want to find out how modern users feel a design is then your polar extremes might be ‘Modern’ and ‘Old fashioned’. If you want to find out how intimidating users find a design then your polar extremes might be ‘Friendly’ and ‘Intimidating’. When considering what sort of facets to capture you might want to think about:

  • How would you like users to describe their experience of using the design?
  • What are the goals of the design?
  • What are the key UX requirements for the design? In which areas does it really need to perform?

This all sounds much more complicated than it actually is so hopefully the following example should show what I mean.

An example of using BERT to capture feedback for designs

An example of using BERT to capture feedback for designs

In the example above I’ve used a 9 point scale, but you can equally use a 5 point, 7 point or even no scale at all – by asking users to mark on a line where the design falls.

As with a lot of research techniques one of the keys to BERT is to ensure that you’re capturing only what you need to find out, and that you don’t try to capture too much. As I’ve already mentioned, 7-10 facets is a good rule of thumb and also in an attempt to reduce any bias make sure that there is no obvious order to the facets, and that you mix up the order of the polar extremes so that you don’t have all positive on one side and all negative on the other.

Where can I use BERT?

I’ve already touched on a few contexts where BERT might come in handy, namely testing design alternatives and competitive benchmarking. You might also use BERT to:

  • Capture feedback following user testing
  • Evaluate different visual designs
  • Capture user perceptions of an existing design (e.g. the current website)

One thing to keep in mind when using BERT is that it’s most effective when utilised as a comparison tool. That is, because one man’s ‘Modern’ can be very different from another man’s, you can’t read too much into results on their own because you have no point of comparison. As you are capturing subjective feedback, what is of most importance is how designs differ, rather then necessarily where they fall in the scale (because that scale differs from person to person).

Analysing BERT results

OK, so you’ve identified that BERT could be useful, you’ve determined your facets and you’ve captured a bunch of results from users. So how can you best analyse this information? Well there are number of ways that you can visually display the information to help you to see what findings emerge and how the designs differ.

BERT results as a graph

You could display the results as a graph (I’ve used a bar chart below) to allow easy comparison between designs.

BERT results bar chart

An example of displaying BERT results for a design facet using a bar chart

BERT results as a graphical scale

You could also display the results as a graphical scale, with the size of each circle representing the number (or percentage) of users choosing that rating.

BERT results as a graphical scale

An example of displaying BERT results using a graphical scale

BERT results as a vertical line chart

Another nice way of displaying this information is using a vertical line chart. This is something that the BBC utilised when using BERT for a redesign of the homepage a number of years ago.

BERT results as a vertical line chart

An example (from the BBC) of displaying BERT results as a vertical bar chart

Finding out more about BERT

BERT is not perhaps as widely known (the Sesame street character aside) as it really should be so there aren’t a great deal of articles out there about it. For a really good case study of BERT being used for a redesign checkout out the BBC’s The Glass Wall (PDF, 8Mb), which details a project to redesign the BBC homepage. Also checkout Stuart Church’s Getting all emotional which discusses BERT and it’s uses for web design projects. Oh and don’t let people try to tell that BERT is evil – it’s actually really rather good!