Are 1-5 satisfaction ratings continuous or discrete?

What's really being asked

Whether you understand that the numeric labels on a Likert-style scale carry ordinal properties, not necessarily interval or ratio properties, and that this classification dictates which summary statistics and inferential tests are mathematically appropriate. Interviewers want to see that you do not treat every number as continuous by default.

A GOOD ANSWER COVERS four things in order. First, classify the data as discrete because there are only five possible values, and more importantly as ordinal because the numbers imply rank but not equal spacing between satisfaction levels. Second, explain that ordinal data supports counts, frequencies, medians, and modes, while means and standard deviations assume equal intervals that may not exist between 1 and 2 versus 4 and 5. Third, note that inferential statistics should generally be non-parametric, such as Mann-Whitney U for two groups or Kruskal-Wallis for multiple groups, rather than t-tests or ANOVA. Fourth, acknowledge the practical reality that many UX researchers treat 5-point scales as interval when the distribution is reasonably symmetric, but defend that choice explicitly rather than ignoring the scale type.

The mistakes people make

Calling the data continuous because it is numeric. Computing arithmetic means without comment. Jumping straight to parametric tests like the independent samples t-test or one-way ANOVA without checking assumptions or acknowledging the ordinal limitation. Another red flag is conflating discrete versus continuous with nominal versus ordinal, for example saying it is discrete therefore you cannot rank it.

What usually comes next

The interviewer may ask when it is acceptable to treat Likert data as interval, how you would visualize the distribution, or what you would do if the data is heavily skewed. They might also ask how sample size affects the robustness of parametric tests with ordinal data, or whether collapsing a 1-5 scale into satisfied versus dissatisfied changes the measurement level.

A concrete example

Suppose you compare satisfaction before and after a redesign. If you treat the 1-5 scale as ordinal, you report the median shifted from 3 to 4 and use a Wilcoxon signed-rank test to assess significance. If you treat it as interval, you might report a mean increase from 2.8 to 3.4 and use a paired t-test, but you should justify that the psychological distance between each point is roughly equal and that the sample distribution is not severely skewed.