In a likert scale of attitudes (agreement scale), does it make a difference if I have coded the following 5-points likert scale (strongly agree, agree, unsure, disagree, strongly disagree) with (2,1,0,-1,-2) where 0 is (unsure) in stead of the common coding (5,4,3,2,1) where 3 is (unsure).
The argument that using common coding would obliterate the bi-directionality of the scale
The question is what do you want to do with the data?
You say Likert-scale, but I assume you mean Likert-type scale (i.e. you only have one variable that is measured on the 5 point scale), as a true Likert-scale involves multiple items that are summed together measuring a single construct.
The coding will depend on which of the two options you are using.
If it is the latter, in brms you can run a model with categorical or ordinal scales; and the coding matters more from a theoretical point of view (e.g., can you have negative values in a real world, e.g., hate-love scale where 0 mean ambivalent).
It will only make a difference to your descriptive statistics (lower mean as compared with a 1-5 scale), it will not change your inferential statistics.
The question arises as to why you are wanting to use a negative rating. If you are seeking to measure or items that are negatively worded. Then what you can do is enter the data obtained for the negatively worded items then reverse score them in the dataset. Just be careful to only analyze the reversed scored items. Using a 0 rating means the items are not analyzed by most analysis systems.
Thank you all of your answers. In response to Mircea Zloteanu , I am going to measure a group of constructs and each construct will be measured through Likert scale (items with 5-point agreement rating scale ie. strongly agree, agree, unsure, disagree and strongly disagree). the construct summary measure will be mean percentage score. Does it sound more logical to assign values for each items response by using scale of 2,1,0,-1,-2 RATHER than 5,4,3,2,1?
Mohamed Bala the question here is "what labels make most sense"; is "unsure" the midpoint of agree-disagree? maybe "ambivalent" or "neither agree nor disagree". The step is important when you construct the scale, as you are claim it reflects something about judgement. You might use a slider with 100 points if you think agreement is a continuum. Or you can put "unsure" outside the scale as something to click. This may all sound pedantic, but research has shown it matters. For instance, how would you analyze "unsure"? (see paper by Paul Bukner on analyzing likert data for ways to model this)