How to use the Likert scale in online surveys?
There are many reasons to create online surveys, what’s yours? You may want to know the level of satisfaction of your customers. This is a very common reason: we are all interested in knowing, for example, if they had a good time in our establishment, if they liked the product we sold them or if our employees took good care of them.
If you share these kinds of concerns and want to get valuable responses, then you should definitely use a very popular resource in online surveys: the Likert scale.
It is a frequently used method to measure attitudes and opinions with a greater degree of nuance than a simple “yes/no” question. It owes its name to the american psychologist Rensis Likert, who published a report in 1932 describing its use, long before online surveys existed!
Introduction to Likert: How does it work?
Let's start from the beginning. To understand the scale, first, it is important to remember what a survey scale is.
A survey scale represents a set of response options, numerical or verbal, that cover a variety of opinions on a topic. It is always part of a closed question, that is, it presents respondents with pre-filled options of answers.
The Likert scale is a measurement tool that, unlike questions that require us to answer yes or no, allows us to measure attitudes and know the degree of agreement of the respondent with any statement we propose.
A classic example of this scale can be when a company asks its customers “What is your level of agreement or disagreement with the affirmation: “I'm satisfied with the services of the company”? and gives them a response with the rating scale “Totally disagree/ Neutral/Agree/Totally agree”.
Continuing with the previous example, the phrase "I am satisfied with the services of the company" is a Likert item. At the same time, the sum of several items together with the respondents’ ratings in each of them make up a scale, in which the interviewers must add the reviews of those items whose content is similar to each other.
How many levels should you include in a Likert scale?
When conducting a survey, the response represents the intensity of agreement or disagreement ("Extremely agree" vs. "Extremely disagree") with the statement of the question. This makes it a bipolar scoring method.
At the same time, the scale also represents the intensity of binary options as “Extremely agree” versus “Not at all agree” or “Extremely disagree” versus “I don't disagree at all”. These options are always linked to the statement of the question. For this reason, Likert is spoken of as a “unipolar” rating method, that is, it allows the respondent to focus on the absence or presence of a single item.
Either of these two options requires complete symmetry, guaranteeing bilateral symmetric distances from an existing or imaginary neutral option and an equal number of positive and negative options.
While there is no clear consensus among researchers on this, the scale generally includes 5 levels, but 7 can also be used. If you decide to add levels, you will get the advantage of more complex and diverse ratings. In other words, in an item with only 5 levels, respondents tend to avoid the 2 extreme options, obtaining very little variation. This is known as "tendency bias".
There are studies that conclude that, from 8 levels, the results obtained are the same, so adding levels will not result in greater variation. It therefore seems that the optimum are the items with 7 levels. However, if you do not need to apply a lot of complexity to your online survey responses, it is likely that with 5 levels your form can provide you interesting information. We suggest you try both alternatives and then decide which one best suits your needs.
Although it is not exclusive, we recommend using odd-level numbers so that an answer can be the only neutral answer. In other words, if the levels go from “totally disagree” to “totally agree” as extremes, option 3 should be “Neither agree nor disagree” so that the respondent has the possibility of expressing himself neutral in a single answer.
Learn how to measure the customer's satisfaction
Let's look at an example. Suppose you have launched a beer (craft or industrial) to the market and you need to know your consumer’s opinion. The first thing you should do is consider which categories you’re going to test: then you decide that you are interested in knowing the opinion about the taste, the alcohol level, the amount of gas, the combination with another food and the price.
For each one of these categories, your form should present a question with a different affirmation. So, continuing with the same example, you could put together the following form:
- Beer X tastes like I expected
- Beer X has a lot of alcohol
- Beer X has a lot of gas
- Beer X is well enjoyed with a hamburger
- Beer X is very expensive
As you will see, some of these statements are made in a negative way. Why? Below we will comment on other biases and how people tend to answer yes, but the short answer is that if people are not happy with the product or service, it is better to know and learn that information immediately.
For each one of these statements, the user will have five options to comment on whether or not they agree with that statement: two extreme (strongly agree/ strongly disagree, two intermediate (agree / disagree ) and one neutral (neither agree or disagree).
Based on the information collected, you can analyze each item separately or add them all together to get a score.
Apply the Likert scale in your online surveys
Now, how can we present it in online surveys?
Online survey platforms offer tools to organize a Likert scale question vertically or horizontally.
Some researchers suggest that if it is presented horizontally, what is known as “left-side bias” can occur in the respondent. This means that when the options are placed on the left side of the scale there may be a tendency among respondents to select options on that side and translate into something stronger for the positive options on the left side.
Again, the consensus of the researchers is not complete, as other analysts suggest that there is a similar, but even stronger, selection bias for vertical Likert scales, as respondents tend to omit the lower response options and select the most frequent options.
On the basis of these claims, certain scale designs appear better suited to avoid survey bias and sloppy question answering than others. But ultimately, you should be the one to decide what is best for the purpose of your online survey.
When to use Likert scales?
You can use this methodology to measure different behaviors of the audience that responds to your survey. It can help you discover the level of agreement with a statement, the frequency with which a certain activity is carried out, the level of importance attributed to a certain factor, the assessment of a service, product, or company or the probability of taking a future action.
These scales have a variety of purposes and advantages when used in online surveys. They are ideal for capturing attitudes, opinions, emotions, and comments on any given topic. They are great for being very specific on a topic rather than getting lost in generic questions.
In addition, these types of questions are universally applicable, easy for the respondent to evaluate, and quick to complete, thus avoiding survey fatigue.
How to treat the results?
This tool is considered as an ordinal type scale, which allows you to calculate the median (the value of the variable that occupies the central place in an ordered series of data) and the mode (the value most frequently in one of the distributions of data) of the results.
The value assigned to each position is arbitrary and will be determined by the survey researcher / designer himself. Given this value, we can calculate the mean, the median, or the mode. The median and the mode are the most interesting metrics, since making an interpretation of the numerical mean if we handle categories such as "agree" or "disagree" will not provide us with much information.
Once the form is finished, you can analyze each item separately or, in certain cases, add the responses of a set of Likert items and obtain a total value.
When adding the responses to the elements, you must take into account that all the elements measure the same, that is, have the same weight. In the case discussed above, an analysis of the results would provide us with a coefficient of consumer satisfaction with the new beer.
Other data to take into account in the analysis (here we will get a little technical, so if you are not a statistics lover you can continue reading the following section and, if necessary, ask a colleague for help to weigh the results):
- On the Likert scale, dispersion is calculated through the interval between quartiles (the standard deviation cannot be calculated), or it can be analyzed using non-parametric techniques.
- The responses to the items can be added together, and it should be noted that all items must measure the same (eg attitude towards foreigners). An analysis of variance could be applied.
What are the advantages of using Likert in online surveys?
Let's review some of the advantages the scale brings.
First, from a survey design point of view, Likert is an easy scale to build. On the other hand, we offer the respondent the facility of being able to graduate their opinion in the face of complex statements.
It works very well for online surveys: it is very visual, the respondent can make comparisons between items, as well as modify and adjust their response easily.
However, not all that glitters is gold, and some pollsters point to drawbacks to using the Likert scale. On the one hand, two people can obtain the same value on the Likert scale, having made different choices. Furthermore, they say, it is difficult to treat neutral responses, of the type "neither agree nor disagree."
Another phenomenon that often happens is that respondents tend to agree with the statements presented, which is known as “knowledge bias” (acquiescence bias).
Let's get to it
You should be the one to decide how, where, when and why to apply the Likert scale in your online survey. The important thing is that you start from an objective and, in order to measure the results, analyze them and improve the design of the online survey in a new iteration. Everything in the digital ecosystem is dynamic and surveys are no exception!
We invite you to visit our website to learn how to use the Likert scale to create dynamic and engaging online forms for your target audience. Our online surveys have an average completion rate of 83%, which will allow you to obtain key information to enhance the relationship with your customers. Create your first online survey in just a few steps with Survey Kiwi!