Hi Ahmed,

the most important concern when you apply numbers and lables to response categories is that of communication - that is, the numbers should be in aggreement with the psychological state that you want to know. I find the label "uncertain", too, not reallly the psychological midpoint between the other to agreement- vs. disagreement tendences. Rather would I use "neither nor" to reflect that the respondent has no tendency.

With regard to using uncertain to a separate category: Yes, you can do this but there should be no number at all, as "uncertain" is exactly that: no response as he/she does not know. This implies that this prerson is missing for that variable which complicates analysis. Again the question is theoretical: Do you think, every member of the population has some opinion about the topic than "forcing" a response could be the most valid way to go. When, in contrast, the answer requires some background information or experience, than - indeed - it would be most useful to "select out" the person. But this changes the overall perspective on your population - as your target population is those respondents who have that knowledge/experience. Think about that in the first place: What is your population.

HTH

Holger

Thanks a million. The populations of the study were the EFL teacher educators and school teachers. What do you exactly mean by:

" numbers ... to response categories is that of communication - that is, the numbers should be in engagement with the psychological state that you want to know" . What "psychological state" DO you mean?

If 0 is assigned for No Opinion what effects does it have in comparison with assigning 3 to No Opinion?

Ahmad R. Beigi the value '3' does not mean 'no opinion'. If you wanted to give the opportunity to answer 'no opinion' you should have create a addidional option. It seems you used the wrong wording for category. It should have been 'neither disagree or agree'. This is something different from 'uncertain'. 'Uncertain' is not an admissable answer within the range 'strongly agree' to 'strongly disagree'. Optimal scaling could give you some insight how people perceived this category. Optimal scaling can show if this catagory is in the middle of the range or at either end of the scale. It will do so in order to optimize the correlation between scores. It is a kind of non lineair factoranalysis.

Yes that is true. For the Likert scales, the participants provided their responds on a five-point: “(1) SD = Strongly Disagree, (2) D = Disagree, (3) N = No Opinion, (4) A = Agree, (5) SA = Strongly Agree which was given a number from (1) to (5), respectively and with negatively worded items, the scores were reversed” (Dornyei, 2003, p.37) before analysis with SPSS software.

Now the question is can we possibly assign "0" instead of "3" for No opinion in a research?

Hi Ahmad,

the "psychological state" is the phenomenon you want to measure. My sentence that the numbers have to support the meaning of the question means the following: People try to make sense what you really want to know with a question. If the numbers or labels don't concur with that, a different interpretation may arise.

Let me give you an answer. A few years ago, the German research institute GESIS made an experiment. They gave two groups of respondents the same question: "how successfull were you in your life until now" -- however one group got a scale from 0 to 10 and the other from -5 to +5.

As a result the, mean of the second group was substantially higher. The reason is that the first scale was interpreted in terms of "low success" to "high success" whereas the second as "failure" to success" (because of the negative values". Hence, decide what you want to know and choose numbers that communicate that aim.

HTH

Holger

Dear Ahmad R. Beigi

In your Likert scale, the '3' would more correctly refer to 'neither agree nor disagree', which is different from 'no opinion'.

(1) SD = Strongly Disagree, (2) D = Disagree, (3) N = Neither agree nor disagree, (4) A = Agree, (5) SA = Strongly Agree

In this case the '3' can not be replaced with '0'.

If you need to have a category 'no opinion' you would need to develop a new set of categories.

For the accuracy of the research, I think participants should be able to choose the "no opinion" option that can receive the value "0" in the scale construction. In this way, you can analyze the frequency of choices for each answer variant. Also, in order to have no significant changes in the results at the time of data processing, the value "0" can be declared as missing in the SPSS program. Good luck!!

Thank you all, I finally assigned 0 for "No opinion" in a five-point Likert scale.

Instead of labeling it "No opinion" we can use "Neutral" and assign "3" numeral.