Welcome My name is Adam Nahirnyj and I am a UX/UI designer. Please take a look at my UX/UI, Web Design work. If you have any questions or comments, feel free to contact me.

Survey Design Part 1 – Rating Scales

Surveys might seem like a quick way to gather consumer data, but anyone who has ever spend time designing a survey knows that there is a lot of work involved in asking the right questions and making sure your responses are unbiased.

In the next few posts, I will give some pointers on survey development starting with rating scales.
Most of us have probably been exposed to the two common rating scales: Likert (LICK-ert) and Semantic Differential Scales. While these scales are great for research projects, a poorly constructed rating scale will do more harm than good.

Semantic Differential ScaleSemantic Differential Scale
Likert ScaleLikert Scale

Things to keep in mind:

  • When using Semantic Differential Scales, use careful consideration when selecting the endpoints to anchor the scale. Example: “Superior/Terrible” will product different results then a question anchored with weaker adjectives “Very good/Very bad”. Research has shown that respondents are reluctant to choose extreme descriptors for their response.
  • Consider the meaning of the endpoints. Example: “Friendly/Unfriendly” might have a different connotation than “Friendly/Not Friendly” and product different results.
  • There is evidence to show that there is a bias toward the left side of the scale. Consider the two scales below:
    Likert Scale
    Likert Scale
    Although these two rating scales are “balanced” – meaning they have an equal number of positive responses as they do negative, research indicates that regardless of whether the left descriptor is positive or negative, responses will bias to the left of the scale. Knowing this, we can say that it is important to always keep your positive descriptors on the same side – left or right.
  • How many points should you have in a scale? There is not a magical answer to this question but as a general guideline, scales should have between 5 and 9 points.

Comments are closed.