Dear Tezana,

There are two meanings of Likert scales. It is either a single item with a number of discrete values, such as five, or it is several items which are combined together into a single variable. I suspect you are talking about the former.

That being the case you have ordinal data which means that you should not be calculating a mean and a standard deviation as these forms of summary statistic are only permissible with scale data. The kinds of summary statistic you are permitted to calculate are minimum, maximum, median, quartiles, range and interquartile range.

If subsequently you wish to classify a data series as low, medium or high then you need to define how you wish to this this, e.g. by on the basis of the median.

I totally agree with Peter Samuels' answer. If I may add to it, if your intent is to describe your results and if your scale ranges from "totally disagree" to "totally agree", something else you can use is percentages. For example, for each question you can look at the percentage of respondents that agree or totally agree vs the percentage of respondents that disagree or totally disagree. That might me more informative than the median.

Categorization of the answers into low, medium and high is largely subjective. A suggestion is" totally disagree " and "disagree"- low, " neutral " - medium and "agree" and "totally disagree" - high. If as usual the respective scores are 1, 2, 3, 4 and 5 the means between 1 and 2 are adjudged low; the means between 2 and 4 adjudged medium and the means between 4 and 5 adjudged high.

Hello, Tefera

In my opinion,

First you need to define and classify. Forinstance, those who score 1&2 may be classify as low, 3 as medim and 4&5 as high. Then you can calculate how many of them are low, medium and high fore the specific item and for total score. But your classification/ catagory should be sound and litrature based. Be aware of, your classification may be the reverse if you have used 1- strongly agree to 5- stongly disagree.

I do not think you need to categorize your results. What is needed of you is to report them in meas, standard deviation or better still in percentage. Alternatively you may still go ahead and obtain your weighted means.

Tezana Bekele During my research, have never done those three against the average scale, because I don't believe the average results, except special measurement model, in a study of data can talk if we right to manage or use the data for analysis, need some references to basic footing, Why? because to the low, medium, high, and each has a different meaning of the measurement results.

More to study descriptive so that the value of low, medium, high and become more meaningful to me.

You have to calculate the mean of the likert scale questionnaires and make categorized as above the mean and below the mean

I had been dealing with the same issue for many days. I treated my 6 point Likert scale as Continuous data and was looking for credible source to divide it into Low, Medium and High.

At one point, I found a book on survys in social research. De Vaus (2002) state, "This approach to collapsing

categories has the advantage of letting the data define what is low, medium or high rather than us imposing some external, unrealistic definition." (p.165).

The author also suggests diving the sample by the third. So, in my case, I divided 6 point Scale by 1.66 each.

Hope this helps!

Nice Contribution I Putu - Artaya

As to me, categorization needs some references and or standards on the area that you need to study. But in general you should be reasonable and logical while categorizing the response. For example, if you are studying challenge using likert scale, you shoud categorize participants to have challenge if and only if their score is 4 and 5 (agree and strongly agree) for the positively stated sentece of the challenge question. e.g. My teacher's response to my questions is making me suffer a lot. 1. strongly disagree 2. disagree 3. neutral 4. agree 5. strongly disagree. In this particular example we can say that the students as having challenge if he agrees with the statemet; otherwise he has no challenge concerning this particular question since he is in the opposite side. I hope this helps....

I think based would be to to re categorize the data under the scale of 1 – 5 (1.00 – 1.79 = Strongly Disagree, 1.8 – 2.59 = Disagree, 2.6 – 3.39 = Neutral, 3.4 – 4.19 = Agree, 4.2 – 5.0 = Strongly Agree). This in turn helps to manage outliers on a particular statement and appears a accomadative since it observes a range. Try reading Ruiz (2002), u will be more guided on both mean and standard deviation under a five point scale

Dear Martin Kasenge

Would you please provide me the full information of Ruiz (2002). Many thanks in advance.

Dear Mahmoud Alquraan

Please try to read this work by Jonald(2019).

https://gssrr.org/index.php/JournalOfBasicAndApplied/article/download/9874/4329/

Elias Semajeri Ladislas thank you for sharing this information

Criteria for constructing judgments on the

statistical means (M) of responses

Range Degree

M ≥ 3.66 High

2.34 ≤ M < 3.67 Moderate

M < 2.34 Low

This is come from to formula that 5-1/3= 1.33

So from 1-2.34 consider low and so on as I mentioned before...

Would anyone please provide me the complete information of Ruiz (2002)? Many thanks in advance.