What is the Central Tendency Bias?

The central tendency bias is when someone filling out a rating scale has a tendency to place most of the answers in the middle of the scale and avoid the extremes. For example, if a customer is rating different aspects of your product, service, or business, on a scale of 1 – 10 then most of the answers responses will fall in the center (4-7) with only a few being rated highly (8 – 10) or lower (1-3).

This may stem from the respondents inability or unwillingness to give an extreme answer. As a result, your data isn’t as clean as it should be. The central tendency bias can occur with internal surveys with your staff, management surveys, and surveys sent out to your customer base.

It occurs with questionnaires that measure the degree of sentiment such as a Likert scale or semantic differential scale.

Examples and effects of the central tendency bias

One of the most common examples of the central tendency bias occurs with internal surveys – especially when managers have to rate employees. It happens with customers to a lesser degree and there are a few reasons for that.

Note: it doesn’t happen with NPS style surveys because there’s only one question being asked.

With internal questionnaires, managers have to rate the people they work with on a daily basis. Those assessments can be the basis of promotions, terminations, and compensation. When there’s an emotional factor at play, the manager may hold back on rating an employee too low.

It occurs to a lesser degree with customers because the stakes aren’t as high. With that being said, they may avoid choosing either extreme because they don’t have the right frame of reference around the question and options.

For example, if you ask about their customer service experience, they may not truly understand what a 1 means or what a 10 means so they err on the side of caution.

In both situations, you don’t get the data you need to make the best decisions.

How to avoid the central tendency bias

Ensure questions (and options) are clear

As mentioned before, the central tendency bias often occurs because people don’t have a reference point for the answer options. What does 10 mean and what does 1 mean?

You can add simple tooltips that give that information. For example, 10 means you’re very excited while one means you’re not excited at all. It adds clarity and allows respondents to give you more accurate answers.

Avoid forcing people to provide justification for answers

In many organizations, managers need to justify why they gave someone a high or low rating. It requires much more thought, especially when they’re rating a large number of people. To avoid the extra work, many will default to the middle option.

Avoid this by not requiring people to provide an explanation for their answers. This is a double edged sword. You get cleaner data but you may not get the thought process that went into the answer which, in many situations, is just as important.

Don’t allow respondents to give the same rating twice

Instead of allowing people to give the same rating over and over again, it’s only possible to give a rating once. For example, if they gave one option a 10 then they can’t give anything else a 10.

If you’re developing a product with multiple features, you can ask them to rate the priority of each feature on a scale of 1 – 10.

Only a single feature can have a priority of one and only a single feature can have a priority of ten. This can be effective because it makes people think more thoughtfully about their answers. It can also give you biased data because people feel strongly about multiple options but are unable to choose it.

For example, they consider someone a high performer in 5 areas but have to rate at least one of those areas poorly or as mediocre.

Here's Why Businesses of All Sizes Rely on KyLeads 

See How KyLeads Can Power Your Business