In attitude measurement, one of aims is to have variance in the response of the users. It is usually suggested to have 5 point scale to measure attitude, while others recommend 7 points. In short, there is not hard and fast rule in deciding the points in Likert Scale.

My opinion is that it depends upon your research objectives and RQs. Also, how would you like to measure the attitudes, singleton item or composite scale. For composite scale use 5 or 7 point scale. For single item, using 3 is also fine (Yes, No, IDK). However, if you wish to use Mean and SD, then go for 5 or 7 point scale. 3 point scale may be suitable for applying chi-square test.

i endorse the answer of Mr Ali Farooq. as it depends on circumstances objectives and nature of behaviour and expected volatality in the human behaviour in given circumstances

No absolute rule exists; however, three points is likely to prove very limiting to your analysis. I would suggest using at least five.

Likert scale originally have 5 (or 7) points. For performing statistical analysis and being able to have a better understanding of the research, it is recommend to use 5 (or 7) points.

But what if students find it confusing ? I always face this problem. Either they choose Neutral or leave the question unanswered but when it is 3 points, I noticed they can express their attitudes towards a particular item easily .Besides, I've read that the new trend in Likert scale is 3 points ... Can I know how reliable is that ?

How many options in the Likert items can depend upon the audience and the sophistication of the audience for the topic.

For example for younger children we may use a three-point scale, and may use smiley faces rather than numbers.  The form at the following link is used with children aged 5 to 8.  It uses both 3-point and 5-point scales, with faces, as well as other types of questions.

http://njaes.rutgers.edu/evaluation/youth/docs/RCEYOUTHEVALUATION_GRK-3_FORM.pdf

Most adults on general questions can very quickly assess things on a 5-point scale.  7-point as well, although I think it takes more time and thought.

Thank you Salvatore.

They are University students . I found them very lazy to think and decide. I can see when moving around how confused they are whether to choose Strongly agree or just agree or move directly to neutral ! I decreased the number of items but still facing the same problem. So, I decide to leave it as 3 points.

If I use Likert scale of 3 points , is it going to affect my results? mean and standard deviation?

I am dealing with students in my research and perhaps lack of interest comes from type of topic being studied and/or too many items asked.  Most of the time if a respondent is unable to find answer to a question after searching their brain, s/he goes for neutral option. To check if this trend is due to attention deficit or lack of knowledge,  perhaps,  adding couple of attention check questions will help you identify the lazy respondents that can bw removed from analysis. 

Nonetheless in some studies where researcher believe that respondent will go for neutral option inadvertently,  4 point scale is also used, by removing neutral option. However, there's another debat that respondents tend to select higher point in absence of neutral point.  

Have you thought of changing design of your questionnaire?  I'm sure how are you distributing the questionnaire among your students, sometimes,  a catchy looking questionnaire itself attract the respondents.  

Well, I would probably say you shouldn't be using mean and standard deviation for Likert items at all...

But I'm sure there is some effect on the summary statistics if you use a 3-point or 5-point scale.  It might make and interesting experiment for your classes.  I don't really know how the two would compare.

In any case, determining comparable summary statistics for the 3-point and 5-point might be tricky.

I would love to know any justification for using a three-point Likert scale, really.

Robert Hess , I think the responses in this thread give some indication of where using three-point Likert-type questions can be useful. As I noted, they are used for younger children who have no trouble with: How much did you learn? None / Some / Lots, but may have trouble with more complicated questions. Another use is when the goal is a rapid and simple assessment. I could imagine this in a marketing survey: What's your opinion of this product: negative / neutral / positive. You might also visit the paper that the O.P. attached.

As a final note --- I'm not in favor of this, but --- it appears people will often reduce 5-point items to a dichotomous or trichotomous response in order to use simple percents (% agree or strongly agree), binomial test, sign test, chi-square test, and so on. This suggests that often people are thinking about Likert-type question responses as negative / neutral / positive in some cases anyway.

Hi Soumia! This is probably a bit on the side but I have interviewed patients from some cultures (e.g., immigrants from some Asian countries) and they seem to avoid taking a stand to attitudinal questions, i.e., they would rather prefer the midpoint on such scales. I've heard that this is more common in culturs that are not so individualistically oriented as you typically see in Western cultures. Does someone have similar experiences?

Maybe students are not "lazy", they are maybe not used to take a stand?

Best

John kåre

I am using 3 point likerts Scale (Most important, important and lest important ) for identifying training needs of dairy framers, the experts feel it easy to respond and as the respondents are experts in the field and the method of analysing the response data is different i.e. Relative importance index for identifying which one among all the needs are most important.

Yes, you can but 5 points will be better for giving more options. It can help you in analysis too.

The scale having more options (5,7,9) will help to perform better in path analysis (SEM).

I agree with Muhammed Ashraful Alam, using 5 and 7 point likert scale is better and common in the studies.

One thing remember, attitude is always a latent variable and usually not directly measured by yes or no. One or two items is not enough to assess attitude.

I think that better 5 point or 7 points Likert scale is suitable for assessing attitudes.

Best Regards

Please see this link:

https://www.formpl.us/blog/point-likert-scale

see this link too:

https://www.simplypsychology.org/likert-scale.html

Actually, its perfectly fine to use your 3-point Likert-like scale for your application. Nasser has a valid point (very). However, you must consider time management, age group, and average response rates.

For example, if you are pressed for time and you send an email survey that is highly accumulative, with 5 to 7 options for each question, it is likely that the survey respondent(s) will not answer all of the questions with a non-neutral answer (as it is faster to pick a middle suggestion).

Therefore, determine timeframe allowed, and complexity (e.g. number of questions) of your survey as you design it. Also, be mindful of whom you shall survey. Using a 3-point Likert has upsides, such as less options, less time wasted, which may encourage your respondent to complete all questions in a timely manner, as it has less.