Get Better Survey Data By Avoiding Overlapping Ranges

When crafting a multiple-choice question the choices need to be:

  • Mutually exclusive, and
  • Exhaustive

In other words the choices should not overlap and they should cover all possible answers to the question. This seems straightforward enough but that’s not always the case. Let’s look at a couple of examples.

Example 1. How long have you worked at your current job?

  1. Less than one year
  2. 1-2 years
  3. 3-4 years
  4. 5-6 years
  5. More than 6 years
  6. Don’t know/Not sure

The choices in this example are certainly mutually exclusive-there is no overlap-but they are not exhaustive. Not all possible answers are covered. Those that have been at their job between 2 and 3 years or between 4 and 5 years don’t have a choice to select. A better set of choices would be:

  1. Less than one year
  2. 1-2 years
  3. 2-4 years
  4. 4-6 years
  5. More than 6 years
  6. Don’t know/Not sure

This set of choices is exhaustive and, for the most part, mutually exclusive. We can certainly live with the small chances associated with someone having worked at their job for exactly 2 years or exactly 4 years.

Example 2. How many different office locations are there at your company?

  1. One
  2. 2-4
  3. 4-8
  4. 8-12
  5. More than 12
  6. Don’t know

The choices in this example are exhaustive but not mutually exclusive. 4 and 8 are contained in each of two different choices. A better set of choices would be:

  1. One
  2. 2-4
  3. 5-8
  4. 9-12
  5. More than 12
  6. Don’t know

This set of choices is exhaustive and mutually exclusive.

These two examples lead us to the following general guidelines:

  • For interval data, the choices should overlap at the end points
  • For count data, the choices should not overlap

Here are two more examples that illustrate these guidelines.

Example 3. Approximately what percentage of your revenues comes from international sales?

  1. Less than 20%
  2. 20-40%
  3. 40-60%
  4. 60-80%
  5. More than 80%
  6. Don’t know

Example 4. How many employees work in your HR Department?

  1. We don’t have an HR Department
  2. 1-5
  3. 6-10
  4. 11-20
  5. 21-50
  6. More than 50
  7. Don’t know/Not sure

Quantitative Question Guide

Pick the right scale (and other survey questions) to collect exactly the data you need.

Get the Ebook

Now let’s look at an exception to the second guideline for count data. When the respondent is providing an approximation or an estimate, then there is no need to provide non-overlapping choices. See examples 5 and 6.

Example 5. Approximately how many employees work at your company?

  1. 1-100
  2. 100-500
  3. 500-1,000
  4. 1,000-10,000
  5. 10,000+
  6. Don’t know/Not sure

Example 6. What is your approximate family income?

  1. < $30,000
  2. $30,000 – $50,000
  3. $50,000 – $100,000
  4. $100,000 – $150,000
  5. $150,000 – $200,000
  6. $200,00 – $250,000
  7. > $250,000
  8. Don’t know/Not sure

For these last two examples the overlap by a single dollar or a single person is of no consequence and including ranges down to a single dollar or person would feign spurious accuracy.

Join the Conversation
  • alonkka

    Thank you. This is very clear.

  • alonkka

    Thank you. This is very clear.

  • MeasurementGeek

    I’d like to see the research behind this. Most survey research suggest intervals should be mutually inclusive regardless of the type of scale. Overlapping categories can cause some issues with what you are able to say with your data and effects the reliability. For instance, I wouldn’t know how to answer the household income question because I fall on one of the amounts that is in two categories. And just to be clear, what you refer to as interval data is not interval because each interval is not equal.

  • MeasurementGeek

    I’d like to see the research behind this. Most survey research suggest intervals should be mutually inclusive regardless of the type of scale. Overlapping categories can cause some issues with what you are able to say with your data and effects the reliability. For instance, I wouldn’t know how to answer the household income question because I fall on one of the amounts that is in two categories. And just to be clear, what you refer to as interval data is not interval because each interval is not equal.