A Likert scale is a rating scale commonly used to assess opinions, attitudes, or behaviors. It presents a statement or question followed by five or seven response options, allowing respondents to select the one that best reflects their feelings or level of agreement. This range of possible answers helps capture the nuances of respondents’ sentiments more effectively.
Likert scales are widely used in survey research. However, they can be susceptible to response biases, such as respondents agreeing or disagreeing with all statements due to fatigue, social desirability, or a tendency toward extreme responses.
What are Likert scale questions?
Likert scale questions are closed-ended questions that allow respondents to indicate their level of agreement or disagreement with a given statement on a numerical scale. The most common Likert scale ranges from 1 to 5, with 1 typically representing “strongly disagree” and 5 representing “strongly agree.”
The format of a typical five-level Likert question, for example, could be:
- Strongly Disagree
- Disagree
- Neither Agree nor Disagree
- Agree
- Strongly Agree
By selecting a number on this scale, the respondent indicates how much they agree or disagree with the statement.
When to use Likert scale survey questions
Researchers use Likert scale questions to obtain more detailed insights than simple “yes or no” questions can provide. For example, if you are surveying customers about their satisfaction with a pair of running shoes, asking, “Are you satisfied with the overall comfort of the running shoes?” gives limited information:
- Yes
- No
This binary question doesn’t reveal the degree of satisfaction or dissatisfaction. Instead, a Likert scale question offers more specific and meaningful data:
How satisfied are you with the quality of the ingredients provided?
- Very dissatisfied
- Dissatisfied
- Neither satisfied nor dissatisfied
- Satisfied
- Very satisfied
Likert scales are particularly effective for assessing subjective characteristics that lack objective measurements, such as attitudes, feelings, or opinions, which can influence behavior.
How to write Likert scale survey questions
When designing Likert scale questions, it’s crucial that each item measures a distinct attitude or trait. To obtain reliable data, you must word your questions with precision. A good rule of thumb is to ensure each question focuses on a single aspect of your topic.
For example, if you aim to assess attitudes toward sustainability practices, you can construct a Likert scale with various questions that capture different facets of this domain.
Here are some tips to keep in mind:
Include both questions and statements
A helpful guideline is to incorporate a combination of questions and statements to maintain the engagement of your survey participants. As you craft these elements, ensure they are clear and straightforward to avoid influencing your respondents’ answers in any direction.
Use both positive and negative framing
Utilize both positive and negative framing. If your questions are solely framed in a socially desirable way, participants may be inclined to agree with everything to present themselves favorably.
Positive framing
The impact of single-use water bottles on the environment is a major concern.
- Strongly disagree
- Disagree
- Neither agree nor disagree
- Agree
- Strongly agree
Negative framing
Banning single-use water bottles does not effectively reduce environmental damage.
- Strongly disagree
- Disagree
- Neither agree nor disagree
- Agree
- Strongly agree
Participants who agree with the initial statement should subsequently disagree with the second statement. By including both in an extensive survey, you can assess the reliability and consistency of participants’ responses.
Avoid double negatives
Using double negatives can cause confusion and misunderstandings because respondents might not be clear about what they agree or disagree with.
Bad Example | Good Example |
I do not purchase products that are not organic. | I strive to purchase organic products whenever they are available. |
– Strongly disagree | – Strongly disagree |
– Disagree | – Disagree |
– Neither agree nor disagree | – Neither agree nor disagree |
– Agree | – Agree |
– Strongly agree | – Strongly agree |
This may lead to unintentional responses due to the complex phrasing. | Positive phrasing is clearer and easier to understand. |
Focuses on what the respondent does not do. | Focuses on the respondent’s proactive efforts to purchase organic products. |
May lead to unintentional responses due to the complex phrasing. | Encourages more accurate responses by using straightforward language. |
Ask about only one thing at a time
Avoid double-barreled questions, which ask about two topics within the same question. When respondents are faced with double-barreled questions, they may:
1. Selectively answer about one topic and ignore the other, leading to incomplete or inaccurate responses.
2. Become confused by the multiple topics being addressed simultaneously, which may cause them to choose a neutral but inaccurate answer in an attempt to respond to both topics at once.
To ensure clear and accurate responses, it’s best to break down double-barreled questions into separate, single-topic questions that allow respondents to provide focused and specific answers.
Be clear
Clarity in your communication is paramount for ensuring the accuracy of your data:
- Formulate your questions with precision, ensuring they are easily understood and leave no ambiguity.
- Select language and styles that align well with your audience’s preferences and understanding.
- Avoid using technical jargon that might alienate or perplex those responding to your queries.
How to write Likert scale responses
The wording of your Likert scale response options is just as important as the questions themselves. Here are some best practices:
Decide on a number of response options
Providing more response options can yield deeper insights but may make it more challenging for participants to select a single answer. Offering fewer choices means capturing less detailed information, but the scale is more user-friendly. Researchers commonly use five or seven response options.
Selecting an odd number of responses is advisable, as it includes a middle point. Conversely, an even number of responses eliminates the neutral option, compelling participants to take a stance.
Here’s a table comparing the 5-item and 7-item scales:
5-Item Scale | 7-Item Scale |
“How frequently do you use public transportation?” | “How frequently do you shop online for clothes?” |
1. Never | 1. Never |
2. Occasionally | 2. Rarely |
3. Sometimes | 3. Occasionally |
4. Often | 4. Sometimes |
5. Always | 5. Often |
6. Very often | |
7. Always | |
Suitable for questions with fewer distinctions | Allows for more granularity in responses |
Easier for respondents to process and answer | May be more challenging for respondents to differentiate |
May result in less precise data | Provides more detailed and precise data |
Choose the type of response option
Selecting the correct type of response options is crucial for accurately capturing participant perspectives across various domains.
Type | Scale Items |
Agreement | – Strongly Agree |
– Agree | |
– Neither Agree nor Disagree | |
– Disagree | |
– Strongly Disagree | |
Quality | – Very Poor |
– Poor | |
– Fair | |
– Good | |
– Excellent | |
Likelihood | – Extremely Unlikely |
– Somewhat Unlikely | |
– Likely | |
– Somewhat Likely | |
– Extremely Likely | |
Experience | – Very Negative |
– Somewhat Negative | |
– Neutral | |
– Somewhat Positive | |
– Very Positive |
Consideration of “Don’t Know”: Some surveys include a “don’t know” option to differentiate between respondents who lack sufficient information to form an opinion and those who are neutral. However, offering this option might lead some respondents to select it by default.
Choose between unipolar and bipolar options
With a unipolar scale, you measure a single attribute (e.g., agreement). In contrast, a bipolar scale allows you to assess two attributes (e.g., agreement or disagreement) along a spectrum. The choice between these formats depends on the information you aim to collect from respondents.
Unipolar Scale | Bipolar Scale |
“How satisfied are you with the quality of healthcare services you receive?” | “How satisfied are you with the range of investment options available through our financial services?” |
1. Not at all satisfied | 1. Extremely dissatisfied |
2. Somewhat satisfied | 2. Dissatisfied |
3. Satisfied | 3. Neither dissatisfied nor satisfied |
4. Very satisfied | 4. Satisfied |
5. Extremely satisfied | 5. Extremely satisfied |
Measures satisfaction from a neutral point (absence of satisfaction) to a high point | Measures satisfaction on a continuum from dissatisfaction to satisfaction |
Suitable when dissatisfaction is not a relevant or meaningful option for respondents | Appropriate when both satisfaction and dissatisfaction are relevant and meaningful to respondents |
May be preferred when the focus is solely on the presence and degree of satisfaction | Provides a more comprehensive view of respondents’ attitudes, including dissatisfaction |
Unipolar scales, which only have positive values, work best when a 5-point response format is utilized. Conversely, if you want to allow for a broader spectrum of responses, including both positive and negative values, then bipolar scale items are recommended.
Bipolar scales have positive and negative values and function optimally with a 7-point response format, with three scale points on each side of a genuinely neutral midpoint.
Use mutually exclusive options
Avoid overlapping response items. Respondents are more likely to make random choices when two items share similar meanings.
Bad Example | Good Example |
“The depletion of ozone layer is increasing UV radiation exposure.” | “Overuse of pesticides and herbicides is harming pollinator populations.” |
– Strongly agree | – Strongly agree |
– Agree | – Agree |
– Neither agree nor disagree | – Neither agree nor disagree |
– Indifferent | – Disagree |
– Disagree | – Strongly disagree |
– Strongly disagree | |
Includes an irrelevant option (“Indifferent”) | Maintains a consistent and relevant scale |
Unbalanced scale with an extra option on the disagreement side | Balanced scale with an equal number of options on both sides |
May confuse respondents and lead to less accurate data | Clear and concise, likely to yield more accurate responses |
How to analyze Likert scale data
Approaches to Analyzing Likert Scale Data When examining the data gathered through Likert scale questions, the first important consideration is the nature of the data itself. Likert-derived data can be viewed as either ordinal-level or interval-level data. However, most researchers treat Likert data as ordinal, recognizing that the distances between response options may not be equal.
Beyond this foundational data type decision, you must also determine which statistical methods, both descriptive and inferential, would be most appropriate for summarizing and analyzing the Likert scale data you have collected. Descriptive statistics can be leveraged to present the data clearly, numerically, or visually, providing a high-level overview.
When working with Likert scale data, carefully evaluating the properties and assumptions associated with ordinal-level measurement should guide selecting the most suitable analytical approach.
Example: Descriptive statistics
Ordinal data: To understand the distribution of responses, you calculate the median, or the middle score, for each question. Additionally, you create a frequency table for each question to display the number of responses for each item choice.
Interval data: You calculate the sum of scores for each participant across all questions. Then, you determine the range of scores, identifying the highest and lowest scores in your sample. You also compute the variance to understand how much the scores differ from the mean.
You can also utilize inferential statistical methods to draw broader conclusions and test hypotheses.
Example: Inferential statistics
Ordinal data: You hypothesize that customer satisfaction is linked to the likelihood of recommending a product. You conduct a Mann-Whitney U test to compare satisfaction scores between customers who would recommend the product and those who wouldn’t.
Interval data: You explore whether income level influences online shopping frequency. Using an ANOVA test, you examine if there are significant differences in Likert scale scores across different income groups.
Finally, ensure that you explicitly indicate whether you treat the data as interval or ordinal in your analysis.
Analyzing data at the ordinal level
Researchers usually treat Likert-derived data as ordinal. Here, response categories are presented in ranking order, but the distances between the categories cannot be presumed equal.
Example 1: Ordinal-Level Analysis
- Consider a 5-point Likert scale where 1 = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree.
- On this scale, a response of 4 is more “agreeable” than a 3, 2, or 1. However, the difference between a 4 and a 5 cannot be assumed to be the same as that between a 2 and a 3.
- To analyze this ordinal-level data, you would use descriptive statistics like the median or mode to capture the central tendency. You could also create bar charts to visualize the frequency of responses for each item.
- Appropriate inferential statistics for ordinal data include Spearman’s rank correlation, Mann-Whitney U test, Kruskal-Wallis test, and chi-square tests of independence.
Analyzing data at the interval level
In some cases, researchers may treat Likert-derived data as interval-level data. This assumes that the response categories are ranked and represent equal intervals between them.
Use ANOVA or Pearson’s correlation for inferential statistics, assuming the data is at interval level. Calculate each participant’s total score for descriptive statistics, then find the sample’s mean score and standard deviation.
Advantages and disadvantages of Likert scales
Like any research method, Likert scales have pros and cons. Understanding these can help you determine when to use a Likert scale effectively. Developed by Rensis Likert, the Likert-type scale is a widely recognized ordinal scale used for the measurement of attitudes and opinions.
It is commonly used in questionnaires to collect quantitative data, particularly in market research. The scale used allows respondents to express the intensity of their feelings or agreement with a statement, making it an ideal tool for measuring attitudes and perceptions in a structured way.
Advantages | Disadvantages |
Provide a standardized way to measure subjective data | Responses may be subject to bias (e.g., social desirability bias) |
Allow for easy quantification and statistical analysis | The distance between scale points is not always equal |
Offer a balanced scale with a neutral midpoint | Respondents may interpret the scale points differently |
Are relatively simple for respondents to understand and complete | Data analysis can be more complex than simple yes/no questions |