In the last blog post, we discussed best practices for survey rating scales and mentioned the use of unipolar and bipolar scales. Since questions regarding which scale is most appropriate often come up in customer discussions, I thought it would be helpful to detail the use of bipolar and unipolar scales.
Survey rating scales are either one or the other—unipolar or bipolar. Unipolar, by definition, means the scale only tracks the presence or absence of a single quality or attribute. Bipolar scales, on the other hand, deal with two attributes.
Examples of Unipolar and Bipolar Scales
Commonly used unipolar scales may include the following choices: not at all satisfied, slightly satisfied, moderately satisfied, very satisfied, and completely satisfied. Statisticians often map these answers to a scale from 0 to 1 (e.g., 0.00, 0.25, 0.50, .75, 1.00 for a five-point scale).
Another typical example of a unipolar scale would be survey questions regarding age demographics. Since the respondents are only considering one attribute, age, the question would be on a unipolar scale.
A bipolar scale not only prompts a respondent to consider two opposite attributes, but it has them determine the relative proportion of each. Where a unipolar scale has one “pole,” a bipolar scale has two opposite “poles” with “neutral” being the midpoint.
A common bipolar scale might include the following choices: completely dissatisfied, mostly dissatisfied, somewhat dissatisfied, neither satisfied nor dissatisfied, somewhat satisfied, mostly satisfied, and completely satisfied. Statisticians often map these answers to a scale with 0 in the middle (-3, -2, -1, 0, 1, 2, 3).
Choosing the Right Scale
Why bother with these definitions? Well, when planning a survey, it’s always best to begin with the end in mind. You don’t want to find yourself mixing and matching scales, negatively impacting your data, and ruining the results you want to achieve. It is critical to know what a bipolar scale is vs. a unipolar scale before selecting your demographic questions because clarity is crucial.
From experience, a unipolar scale works best whenever possible since it’s less confusing to the respondent. Instead of balancing two opposing attributes, they only have to consider one. It’s also more streamlined and offers fewer choices (typically five choices vs. seven). This simplicity comes in handy when your respondent audience contains a larger, random sampling as opposed to a smaller, convenience sampling.
Respondents are much less likely to give incorrect answers to a unipolar question because they don’t have to choose from two sets of polar opposites—which can be more complex than it seems. Many bipolar scales measure one dimension, so synonymous ranking terms tend to overlap.
If this is the case, the question would be a perfect candidate for a unipolar scale. Rarely is a bipolar scale a better choice.
One size doesn’t fit all
Now for another touchy issue among researchers—what size scale to use! Many believe less is more when it comes to response scales, opting for only 3-item scales because they take up less visual space and are less daunting to respondents.
Some researchers prefer 5-point and 7-point options, believing they produce the most reliable results and maximize the reliability and validity of survey results. Yet, others feel a 5-point scale lacks granularity and can lead to distorted results.
Based on previous behavior, some researchers believe respondents will never rate anything a “five” on a scale where “five” is the highest. This results in a prevalence of “four” ratings (thus diluting very satisfied respondents with those who are only somewhat satisfied).
Using a 10-point scale may help to avoid this topping effect. While others opt for an 11-point scale—believing scales with fewer points are more susceptible to grade inflation.
Verint advises the use of 5-point and 7-point scales. These scales have the most research support and, in our experience, deliver the most reliable results.
Research conducted by Jon Krosnick, Professor of Communication at Stanford, and Alex Tahk, a Ph.D. candidate in political science at Stanford, validates this position. Their conclusions are drawn from over 706 tests of reliability spanning 30 different between-subject studies, tested and controlled for sample size and other factors that would affect reliability.
While we at Verint believe 5- and 7-point scales are optimal, we constantly evaluate other scales. The size that’s appropriate for your survey is going to depend on your particular research objective.
Scaling Your Survey
Choosing between a unipolar or bipolar scale is essential for formatting your survey questions. The goal of any great survey is to gain new levels of insight regarding your employees or customers.
To do that, your survey questions need to gain as much data as possible with the most accurate methodology possible. Using a bipolar scale when you could’ve just as easily used unipolar can confuse respondents and significantly skew your results.
What scales do you use, and what research do you cite to determine the appropriate rating scale for your research?
When it comes to leveraging surveys that you can quickly deploy to maximize value from customer and employee feedback, some of the biggest organizations trust in Verint Survey Management. Learn more today or contact our team to schedule a free demo to see this offering in action.