Dear Mollalign,

I faced similar kind of problem while doing my M.Phil research on effect of individual's personality trait on treatment realated decision making. Here is how i completed the research.

1. You ll find many researchers has stated that data collected through 5 or 7 points likert scale may be considered as quantitative data.

2. Run Shapiro Wilk test to check normality of your data, collected through the 7 point likert scale. If it satisfies normality you are good to go.

3. There still ll be difficulty in convincing some researchers as necessarily everybody doesn't accept the aforesaid logic. So you may conduct qualitative techniques by grouping the data so collected, like 1-3 as not likely (if at negative side) 4 as neutral and 5-7 as likely.

4. Also, here is a summary of qualitative vs quantitative

1 sample- t/z test(quan) & ks test(qual)

2 independent sample- fisher t test (quan) & MannWhitney U (qual)

2 paired sample- paired t (quan) & Wilcoxon/Mcnemar(qual)

More than 2 sample- ANOVA(quan) & KW ANOVA( qual)

Likert scale responses can be reported qualitatively or quantitatively - this depends on your research question.

Look at what your research question is - this will then dictate your data analysis technique.

Dears, thank you for your response!

For Samantha Curle ,Basically likert responses are mostly subjective in measuring attitudes, character, and personality traits. In such a case, there is no definite measuring and inferring approach with respect to qualitative and quantitative techniques.

Even if the research question dictates the techniques of data analysis, still the way we dictated is difficult for me. That is the problem what i got now. here is sample question.

Another approach is to say that this is an empirical question. If you believe a Likert scale is at least ordinal then you should believe that there is some transformation that changes the spacing between values so that they are interval. Optimal scaling can be used to find these transformations. This is, for example, done in the PRINCALS procedure, often called the principle components analysis for ordinal variables, but the key bit within this procedure is the optimal scaling transformations.

First, I am not clear what you mean by the "qualitative" interpretation of items with Likert scores. Most people interpret qualitative as coming for open-ended rather than closed-ended questions.

Second, there is a major difference between separate Likert-scored items and scales that you create by combining such multiple items. Because there have been so many questions here about creating scales from Likert-scored items, I have collected a set of resources on this issue:

https://www.researchgate.net/post/How_can_one_help_people_create_simple_scales_from_Likert-scored_items

Dears,Really thank you for your response! I appreciate it.

For David L Morgan , how to get a good alpha coefficient for arranging likert items into scales? and how to arrange likert items for testing using alpha value in different packages, is it enough having theoretical knowledge of the likert scale in question?

From, https://data.library.virginia.edu/using-and-interpreting-cronbachs-alpha/

"Cronbach’s alpha is not a measure of dimensionality, nor a test of unidimensionality. In fact, it’s possible to produce a high α coefficient for scales of similar length and variance, even if there are multiple underlying dimensions. To check for dimensionality, you’ll perhaps want to conduct an exploratory factor analysis. "

what will happen if I am not going to use exploratory factor analysis?

If you do not have a prior theoretical basis for separating your items into sets, then you should start with Exploratory Factor Analysis. In doing so, I would suggest you use either Principal Factors or Maximum Likelihood for your extraction method and oblique rotation.

If you skip using EFA and you do have multiple dimensions (ie.e., more than one factor in a set of items), this will lower your alpha.

Also, on a more technical note, while it is possible to have an alpha value above .7 with a set of items that contain two factors, this will only happen when the two factors are highly correlated.

David L Morgan ,can you suggest me on important procedures for doing Principal Factors or Maximum Likelihood specifically using R, please.

According to https://stats.stackexchange.com/questions/215404/is-there-factor-analysis-or-pca-for-ordinal-or-binary-data answers,

" Traditional (linear) PCA and Factor analysis require scale-level (interval or ratio) data. Often likert-type rating data are assumed to be scale-level, because such data are easier to analyze. And the decision is sometimes warranted statistically, especially when the number of ordered categories is greater than 5 or 6. (Albeit purely logically the question of the data type and the number of scale levels are distinct.)

What if you prefer to treat polytomous likert scale as ordinal, though? Or you have dichotomous data? Is it possible to do exploratory factor analysis or PCA for them?

There are currently three main approaches to perform FA (including PCA as its special case) on categorical ordinal or binary variables (read also this account about binary data case, and thisconsideration about what might be done with ordinal scale).

Factor scores in approaches (2) and (3) are more difficult to estimate than factor scores in classic FA or in approach (1). However, several methods do exist (expected or maximum aposteriori methods, maximum likelihood method, etc.).

Factor analysis model assumptions is chiefly the same in the three approaches as in traditional FA. Approach (1) is available in R, SPSS, SAS (to my mind). Approaches (2) and (3) are implemented mostly in specialized latent-variable packages - Mplus, LISREL, EQS.

For Daniel Wright , have you tried optimal scaling transformation using R packages such as homals and aspect? And what will be the theoretical basis to use transformed data?

" Optimal scaling approach (a family of applications). Also called Categorical PCA(CatPCA) or nonlinear FA. In CatPCA, ordinal variables are monotonically transformed ("quantified") into their "underlying" interval versions under the objective to maximize the variance explained by the selected number of principal components extracted from those interval data. Which makes the method openly goal-driven (rather than theory-driven) and important to decide on the number of principal components in advance."

why we need to maximize the variance,what will be its importance?

See Article Gifi Methods for Optimal Scaling in R: The Package homals

Quality is creed in every food items respect not only for the research but also in other areas including medicine ,grocery . It is in this areas I have expressed my similar views which I submit herewith for your kind perusal

Quality should be our creed & to me apart from research nothing to be carry out at the cost of quality as it has a permanent symbol & with the quality whatever the action we may take in any areas including research may turn out to be everlasting & also as a symbol of permanency.

To me Quantity has a reference to measurement & any of our action under a quantity initial may have a shinning picture but ultimate result in any areas will be having a reflection of quality .

This is my personal opinion