Yes, we use both, parametric and non-parametric statistics for data analysis.

How to think about Likert scales - the perennial debate between Statistics and Psychology!  One perspective is that Likert scales are responses to words like “somewhat agree” that get translated into discrete numbers (e.g., 1,2,3,4,5,6,7).  But how can we know that the “distance” between 2 and 3 is the same as the “distance” between 6 and 7?  I do think they make a good point, especially for just a single Likert scale item with a small number of choices.

From the perspective of Psychology, I feel Likert scales are, first and foremost, an estimate of an underlying attitude.  Attitude is a theoretical construct about our inclination toward a target (e.g., chocolate ice cream, democracy) with a valence along a continuum (i.e., negative to positive).  A Likert scale item is a coarse estimation of the true value of the attitude. But we rarely use a single item.  Average many Likert scale items that have high reliability (i.e., alpha).  Then your estimate of the attitude is getting less coarse, and a more precise approximation of the true underlying attitude.  As such, it seems reasonable to treat the score as a continuous variable.  Hope this helps your thinking, Damtew! ~ Kevin

I like Kevin's answer. Here's what I have to add:

• I think the interpretation of Likert scales should be done by the labels
• Comparison should be done without quantifying differences.
• Highly reliable (alpha) likert items tend to load on the extremes thus suppressing variance (if this concerns your research in any way)

Dear scholars,

I thank you very much !