A journal editor indicated that treating a 5-point Likert scale (an independent variable) as a continuous variable was inappropriate in the initial review. I used a single item for a variable and coded it: 1 (strongly disagree), 2 (disagree), 3 (neither agree nor disagree), 4 (agree), and 5 (strongly agree).
However, I believe that treating a 5-point Likert scale measure (an independent variable) as a continuous variable is a common practice.
What are your thoughts on this?
Don Soo Chon, when discussing the Likert Scale, it is important to distinguish between the Likert item rating options and the overall Likert scale.
The Likert item rating options typically follow your options (e.g., Strongly Disagree to Strongly Agree). In terms of assigning numerical values to these item rating options, the numerical values assigned represent the "place" of the categorical, ordered, ratings as in, first place, second place, third place, etc.
The Likert-scale refers to the range of possible total scores obtained when the individual item rating option (or place) scores are added together. Using the 1-5 scoring system to reflect the place of the item rating options, the Likert scale will range from N to 5N, where N represents the number of items in the survey.
One might ask, why is this even important? Well, the selection of statistical techniques varies depending upon whether the data are classified as ordinal or continuous. In short, the Likert item rating options: ordinal; the summation of the item rating option place scores: continuous (under particular assumptions).
Feel free to reach out if you have any questions, David Sul, Ed.D.
In the framework of SEM, indicators with at least 5-point scale are generally acceptable to come out credible parameters estimated.
I was not the reviewer, but could you provide some information on the distribution across the five categories? How many respondents answer affirmative in each category? The Likert-scale is always dubious whether it can/should be treated as continuous. If there is a good distribution across each option (1-5) then it might be ok.
I have heard mix things regarding whether a Likert-scale can be utilized as a continuous variable. I am cautious about it because of the loss of information when you treat ordinal level measures as continuous. When you have a scale 1 to 5 representing levels of agreement, you are treating it as if the difference between 1 and 2 is the same as 4 and 5. That assumptions personally makes me uncomfortable because I do not know for sure if there really is a true difference.
Is there any reason you cannot use it as a categorical variable (i.e., power/degrees of freedom)?
- Badges
- Science topic
- Similar topics
- Geoscience
- Atmospheric Sciences
- Climate Science
- Climate Change