As a practitioner of usability or user experience, a common way that you will attempt to investigate a user (or player, or customer)’s perceptions is through designing and implementing a survey. In designing a survey, its important to consider the format that questions come in, especially with common question types such as “How frustrating did you find this level?.” Today we’ll look at one of the most common question formats, the Likert scale, and the implications that using it has on your studies.
What is the Likert scale?
Lets start with an example.
Most people have seen a Likert scale before. Do you agree with this statement?
- Strongly agree
- Neither agree or disagree
- Strongly disagree
Often used to gauge opinions, they are especially important for people involved with measuring usability or player experience, as they can help quantify subjective things like a user’s experiences. They are usually in the form of a statement, followed by a selection of statements, to indicate how far someone agrees with the statement. They can often be used to quantify things like ease-of-use, or fun, which would be impossible to quantify through other methods. Hence they are of particularly important for us, since user experience is essentially abstract.
Different kinds of Likert scales.
The essential question when it comes to implementing a Likert scale, is how many responses to offer.
‘Forced Choice’ scales are those which have an even number of options. Essentially this means missing out the ‘neither agree or disagree’ option, and forcing the participant to make a selection (see what they did with the name? very clever!). This would be done to force participants to show an opinion, but there are dangers inherent with this. Forcing a response may give a larger degree of ‘static’ in the responses, reducing their accuracy, since the responses may not map their opinions. People who don’t agree or disagree may not be happy about being forced to give an opinion, reducing their likelihood to answer later questions accurately. However if your aim is to support a conclusion that people do/do-not like a system, you may be willing to risk these to prove your point when designing the survey.
If you select to use a scale with an odd number of options, there are a few issues that should be kept in mind when deciding between a five or seven point scale. The most obvious difference is that a finer grain of responses can be analysed from a seven point scale, as it can represent a wider range of views. Also, take into account that it’s been shown participants shy away from the ‘edges’, the extreme like and dislike options offered. This means a five point scale will likely only get responses in the ‘slightly’ columns from all except the most ardent fanatics. Again, you have to consider whether a wider range of responses is useful to the topic you are exploring.
Should you use a Likert Scale
Ultimately if you are trying to track opinions, a Likert scale is a good method of accessing this data. There is no all-encompassing correct answer for which scale is appropriate, the context of use and what you want to find out will all affect this. As long as you keep in mind that not only the phrasing of the question, but the range and number of responses you offer will affect the results, and anticipate this affect, you can’t really go wrong. Happy surveying!