It is very common to assign numerical values to Likert-scale items if you plan to use your items in quantitative statistical analysis. The numerical values that you propose properly reflect the ordinal (= ordered categorical) nature of the variables. I don't think you necessarily have to cite a reference for that but pretty much any book on measurement and/or scale development should cover this issue.

Hi Enuma. Yes, you can assign values to your Likert scale in the manner you've described. However, this coding scheme is somewhat unconventional. In typical Likert scale coding, adjacent categories are usually coded with adjacent numbers, and the scale is often centered around a neutral midpoint. Your proposed coding places 'Neutral' at 0, which is a common practice, but it reverses the order of 'Dissatisfied' and 'Very Dissatisfied', which might be confusing or lead to errors in analysis.

If you have a specific rationale for this coding scheme (perhaps related to the specific nature of your analysis or the way you wish to interpret the data), it can be used, but it's important to be clear about this choice in your documentation and analysis.

As for reference materials, most standard texts on survey research or data analysis will discuss Likert scale coding but they may not cover unconventional coding schemes like the one you've proposed. Here are a few general references that might be helpful:

Remember, if you use an unconventional coding scheme, it's particularly important to explain your rationale and approach in any reports or papers you produce, to ensure that your readers understand how to interpret your findings correctly.

Often the neutral category would be nearer the center of the numeric values. Not sure why you are wanting to do this however.

Also, you might clarify if thus are dashes or minus signs. I am assuming dashes.

Honestly, you might explain what you are trying to do with this coding scheme. (And probably what statistical analysis you are going to perform.)

Usually, "neutral" would be between "satisfied" and "dissatisfied".

If you want to remove "neutral" responses from being analyzed with other responses, you could do this. But you would have to separate those responses from the other responses, not just give them a different numeric value.

If you consider the neutral response as reflecting "don't know" or "does not apply" then there would be some logic to separating the response out. However, I'm assuming that the original questionnaire placed "neutral" in the middle position, and people will weigh that position in their responses. If you compare responses to a 7-point scale with those to a 5-point scale, you'll find that some of the middle responses will move to a position such as "slightly satisfied" or "slightly dissatisfied." Thus, the position of the response category has meaning to the respondents, and that should be taken into account in the analysis.

The answer also depends on how you plan to perform the analysis. If you only plan to calculate percentages and to focus on the percentages who agree versus those who disagree, then it may make sense to place the neutral responses to the side. However, if plan to calculate means, then the neutral responses should be in the middle. Otherwise, you are saying that the distance from neutral is smallest when comparing neutral to very dissatisfied, and largest when comparing neutral to very satisfied. That could produce misleading results.

Thank you very much everyone. This is a lot clearer

Ifeakachukwu Enuma, there is a presumed order to the item rating options you present. In your case, the concept being rated through the use of these ordered rating options is that of satisfaction and it ranges from neutral (level 0), passes through very dissatisfied (level 1), and arrives at very satisfied (level 4).

Neutrality tends to mean not in favor of one side or another (in a dichotomous situation). As such, I would locate the neutral category at the space in between dissatisfied and satisfied in the satisfaction item scale you have developed.

In terms of assigning numerical values to these ratings, the numerical values assigned represent the "place" of the categorical, ordered, ratings as in, first place, second place, third place, etc. The ordered place of the categorical ratings is not affected by choice of the numerical value assigned to the starting point which represents the first place option. However, most researchers assign the value of "1" to the "first place" option in the item rating scale, for obvious reasons.

Finally, the Likert-scale refers to the possible total scores obtained when the individual item rating (place) scores are added together. Using the 1-5 item rating score options, the Likert scale will range from N to 5N, where N represents the number of items in the survey.

Feel free to reach out if you have any questions, David Sul, Ed.D.