3-point or 5-point Likert Scale?

Home Forums Methodspace discussion 3-point or 5-point Likert Scale?

Viewing 15 posts - 1 through 15 (of 15 total)
  • Author
  • #4059
    Damian Knipe

    Dear members,

    I am not completely convinced which Likert scale works best for me these days in a questionnaire. I have used the 5-point Likert scale more often than the 3-point Likert scale, but tend to recode the 5-point scale into a 3-point scale during my analysis and reporting. What works best for you and why?

    Ahmad Alhaqbani

    Hi Damian,
    I’ve heard that 5-point Likert scale is better than the 3-point in terms of capturing the more accurate feeling or attitude of your participants. I am using the 5-point Likert scale for that reason.

    All the best


    I prefer a six point scale, purely to force fence-sitters to make a choice 3 or 4.

    Al Patenaude


    I found that neither the 3- or 5-point Likert scales work best for me since many of my participant populations take neutral stances; I mainly work with offenders and staff in corrections. By adapting the scale to 4-points, strongly disagree – disagree – agree – strongly disagree, I remove the tendency of these populations to take the “safe” route of replying 3-unsure/uncertain. This is similar to Thomas Groenewald’s six point scale. I’d be interested in seeing the items on the latter scale and hearing what you and others think of this approach and its applicability for your own research. Take Care.

    – Al.


    Peoples views are different one from another. It is therefore important to take care of these differences in choosing the measurement scale. The type of scale must take cognizance of the type of information required. Sometimes, a 3-point Likert scale will suffice. At other times a 5 or 7 point will be required. To my mind the choice of scale is information-required-specific.

    Damian Knipe

    Yes, I understand offering a wider choice of options to respond to a question enables rigour when it comes to analysing the data and producing results. But in reality, how often does a person answer a question by saying I “strongly agree” with what you are saying, most of the people I know usually respond to a statement I make by saying something along the lines of “Yes, I agree or No, I disagree with what you are saying”… It’s good however, to at least offer people a wider choice and then the researcher can then decide how to analyse. Thanks for getting back to me on my query.

    Damian Knipe

    Yes, I have noticed more and more 6-point scales being used in social science research as it cuts out that often annoying, yet vitally important, middle-ground stance. But the word “force” is questionable. Thanks for getting back to me.

    Damian Knipe

    Yes, again interesting stance to take by leaving out the “don’t know”, “unsure”, not applicable”, “unable to respond”, “no opinion” type option! But, could it not be argued that there are people out there who genuinely need to reply by using one of these type of responses becasue that is really their perspective on the statement/question. Thanks for responding to my query.

    Damian Knipe

    Yes, I would tend to agree with you here. The response scale is also specific to the sample you are targetting and the type of statements/questions you are posing. Thanks for your reply.


    It is bettter to collect precise data using 5 point scale. Howevewr at the time of data analysis it is prefreable to collapse into 3 point scale because at the end whar we want to know is wheather they like, dislike on are not sure. At the time of reporting and interprtation you may think of bifurcating the like and dislike catagories further if it so requires.

    Jeremy Miles

    I agree with this – it’s very dependent on the sample.  There is no one answer that will always work best.  I have administered a (6 point) scale where almost every single person either strongly agreed or strongly disagreed with every item.  The six point scale was a waste of time.


    The discussants are not addressing the point that operationalization of a concept means assigning “meanings to numbers” and “numbers to meanings.” There is a distinct difference between conceptual utility and statistical utility. Forcing variability through use of a greater number of scalar points is a statistical utility. It is only a partial treatment of the issues however. One must also be concerned with the intra- and inter-class variability (homoscedasticity and heteroscedasticity) in relation to the kind of statistic one can use. Minimum cell sizes are a critical consideration for any n-way tablular analysis. Often one has no choice but to collapse a 5-n scale to 3 or 2 points in order to have sufficient observations in a cell.


    All the statistical issues, however, mean nothing if there are not a clear theoretical and methodological strategies behind the choices made. “Working best” is probably not the best way to make a choice in number of scalar points. 

    Charles BERG

    I would use the 5 point Likert scale and recode it when I know the distribution. Charles

    Jeff Miller

    On the issue of forcing a polar attitude via even-numbered response options, I want to add that there is more to it than not wanting neutral responses. Certainly, there is disagreement about whether one ever does hold a truly neutral position and might want to eliminate (or not eliminate) a neutral option via an odd-numbered set of response options. And, there is disagreement whether we should “force” non-neutral attitudes when a neutral position may be the reality for the respondent.

    The additional problem is that many respondents gravitate toward the middle response option sometimes as a means of opting out of responding. And, a 3 on a 5-point scale is not the same thing as “I don’t know”, “I don’t care”, and “No comment”. When this happens, you get a confound where you can’t determine whether responses to the middle option on odd-numbered responses scales represent true neutrality or some other attitude.


    What is the role elecrical converer in electronic machine?

Viewing 15 posts - 1 through 15 (of 15 total)
  • You must be logged in to reply to this topic.