Join Jeff Toister for an in-depth discussion in this video Creating scaled questions, part of Using Customer Surveys to Improve Service.
- People seem to have a lot of opinions when it comes to rating scale questions. Should you have an even or odd number of points? Is it better to have five scale points or 10? How should you label the scale to get the best responses? We'll tackle these questions in this video and give you tips for using rating scales effectively. Let's start by looking at whether your scale should be even or odd. An odd scale is the best choice because it has a midpoint, while an even scale does not. Some people feel you should use an even scale to force your customers to decide whether they had a positive or negative experience.
That argument is exactly why an odd scale works best. Any question that causes a customer to adjust their real answer is biased. That means the survey results may tell you something slightly different than what your customer actually feels. People often wonder whether it's better to use a shorter scale like five points, or a longer scale like 10 points. The short answer is: either scale works. As an aside, notice that the second scale has a zero at one end to make the scale odd rather than even.
There isn't one scale length that always works best. On the other hand, the more scale points you have, the more variation you'll get amongst scores. On the other hand, a shorter scale makes it easier for customers to respond. Generally speaking, any scale between five and 10 is fine as long as there are an odd number of total points. Less than five, and customers tend to feel like there aren't enough choices. More than 10, and customers can feel like there are too many choices. People also wonder about labeling the scale.
You may have noticed that the rating scales I've shown you throughout this course have only labeled the end points. That's a best practice, and with good reason. My colleague, Ken Phillips, showed me why at a conference session he facilitated. Ken asked each of the 300 people in the room to assign a percentage to various words he showed on a screen. The first word was "always." Now, if you know the definition of "always," you'd say "always" equals 100%. But that's not what happened, the people in the room gave a wide range of responses from 100% to as low as 75%.
Ken gave us another word, "never." Many people put zero, but some people went as high as 25%. Things got really interesting when he gave us words like "usually," "sometimes," and "rarely." The answers were all over the place. This happened because people tend to apply their own filters to these types of words. You can sidestep this problem a bit and get more consistent responses by labeling just the ends of the scale. Okay, here are a couple of bonus tips for writing scale questions.
First, whenever possible, use a consistent scale with consistent labels. Changing up the scale length or scale labels can confuse customers and lead to some inconsistent answers. Second, your scale should move from left to right with your low point on the left and your high point on the right. That's because English-speaking audiences are used to reading from left to right, so it's a more natural way for them to answer. Of course, if you're writing surveys in a language that's read from right to left, you may wanna reverse this.
Okay, now it's your turn, try to use these tips to review one of your own surveys or to construct rating scale questions for a survey you're developing. The goal is to write rating scale questions that are clear, easy to answer, and reveal useful data.