you first select the indicators for happiness, racism and motivation.Then calculate Z score of each indicators of happiness, racism and motivation . After getting  Z score for the happiness, racism and Motivation. You can categorized the z  score value as  Very High, High, Moderate, Low and Very Low.

you first select the indicators for happiness, racism and motivation.Then calculate Z score of each indicators of happiness, racism and motivation . After getting  Z score for the happiness, racism and Motivation. You can categorized the z  score value as  Very High, High, Moderate, Low and Very Low.

Dear Endro,

 I believe that you wish to create cut-offs for the scores that you have derived from each construct (happiness, racism and motivation) from your questionnaire.

I need to clarify something a couple of things -

1) When you state that these scores were valid and reliable, does that mean you have conducted validity/reliability testing on each scale or did you just emulate the a pre-existing valid/reliable scale without personally testing validity/reliability in this current experiment?

2) Are you validating a newly created questionnaire?

 These questions are very important because -

1) You can get cut-offs by comparing your newly developed scale with another scale which has some form of categorization (ROC curve method)

2) You can also get cut-offs by using scatterplots (Scatterplot method)

 I will need more information before I could help you out. But these are, from my personal point of view, things that you ought to think about first. Hope this helps.

Rabin

Converting continuous variables into categorical variables amounts to a loss of information, in the sense that you can do considerably more in terms of statistics with continuous variables than you can with either ordinal or nominal variables. So the first question is why you want to convert your original variables into categorical variables.

That said, the simplest format is to use something like quartiles, where you would divide your original variables into four categories that each have the same number of respondents in each category.

Using the mean and SD you might define anything more than 1 SD above the mean as "high," anything more than one SD below the mean as "low" and thus anything one SD on either side of the mean as "medium." Of course, if you have anything like a normal distribution, that would put about two-thirds of the cases in the "medium" category.

@Deepak Wankhede: thank you for the advice i will take a look on that.

@Yogarabindranath Swarna Nantha: first of all, thank you for your reply. Below are my respond to your questions:

1) When you state that these scores were valid and reliable, does that mean you have conducted validity/reliability testing on each scale or did you just emulate the a pre-existing valid/reliable scale without personally testing validity/reliability in this current experiment?

2) Are you validating a newly created questionnaire?

No i just emulate the pre-existing validity and reliability. So is your suggestion still applicable after i answered you?

3) You can also get cut-offs by using scatterplots (Scatterplot method)

Well this is interesting, is it applicable for my situation? If yes, can you give me links of paper for me to learn this method?

@David L Morgan: Hi thank you for your advice below are my answer for your questions. 

1) So the first question is why you want to convert your original variables into categorical variables

i dont want to convert my variables to categorical variables. I just simply need to categorized them for descriptive explanation. Because one of my research question is to find out the level of racism, happiness, and academic motivation of the students. 

However, the main idea of my research is a predictor study. So to find out whether racism or happiness will predict academic motivation, i will treat the data as a continuous varible, so you dont need to worry about that.

i just need to categorize the the variables for descriptive purposes and will not statistically analyzed it using categorical form.

Yup thats what i mean, about the SD and Mean. Im sure will take a look more thoroughly into it.

Feel free to comment or reply back. i am all for feedback and suggestion :)

Dear all, 

1) In relation to the answer of David Morgan, after using the correct test /models for quantitative variables, I usually categorize them, obtaining  more interpretable results . Than I usually complement  the  analysis of quantitative variables, with models for categorical ones.

2) If Wilibrordus Endro has different continous variables, I will procede as Deepak Wankhede said.

3) "i just emulate the pre-existing validity and reliability. So is your suggestion still applicable after i answered you?"

if I was you I would analyse also if your scale is reliable. One thing is apllying a known reliable scale needed for your analysis, another is compementing that giving information about the results you obtain with the same scale.

Helena 

Agree with Helena.

De acuerdo a las variables que estés trabajando en tu proyecto defines las categorías para el cuestionario. Se trata de identificar las dimensiones que cruzan un problema o tema en especial. Si quremos validar el desempeño de un docente, una categoría para identificar y validar es el aprendiza de sus alumnos . Otra  su formación.

Good morning/afternoon/evening everyone.

 Dear Helena Pestana

1) In relation to the answer of David Morgan, after using the correct test /models for quantitative variables, I usually categorize them, obtaining  more interpretable results . Than I usually complement  the  analysis of quantitative variables, with models for categorical ones

Thank you for your suggestion. I have never thought of combining multiple regression analysis with categorical interpretation.  I will try to apply this also to my research.

2) If Wilibrordus Endro has different continous variables, I will procede as Deepak Wankhede said

i do have 3 different continuous variables. Hence, i will try to apply z score technique for my categorization.

3) if I was you I would analyse also if your scale is reliable. One thing is apllying a known reliable scale needed for your analysis, another is compementing that giving information about the results you obtain with the same scale

Again, thanks for your advice. Indeed i will also analyze the validity and reliability of my questionnaire in my research and compare it to the existing validity and reliability value. However due to time limit. I will not able to use test-retest techniques. I can only apply for split half technique.

Dear Yogarabindranath Swarna Nantha

Hi, thanks for the confirmation of Helena answer. As the final thought, i will go with Depak suggestion. Thank you

Have a nice day

Dear Mariela Salgado Arango

I am really sorry. I dont understand your answer. I tried to use google translate but made confuse. Do you mind answering again in english? Thank you very much.

Dear Wiliobrordus, 

Please see the translation I  made for you. I hope it is ok!

 "De acuerdo a las variables que estés trabajando en tu proyecto defines las categorías para el cuestionario."= "According to the variables you're working on your project you define the categories to the questionnaire."

2) "Se trata de identificar las dimensiones que cruzan un problema o tema en especial. Si quremos validar el desempeño de un docente, una categoría para identificar y validar es el aprendiza de sus alumnos . Otra  su formación."

= You have to identify the dimensions crossing a problem or particular topic. If we validate the performance of a teacher, a category is to identify and validate  students. Another training..

Helena

Dear Helena Pestana:

Thank you for the translation. Indeed, these explain why a lot of questionnaire dont have any category. Most of them say : "The higher the score, the higher the level"". Again thank you, this website help a lot.

:)

Dear Wilibrordus,

From my perception, I have realized that your question in RG is little bit vogue and it can be clear if you mention some questions.I think in your questionnaire if you do not make higher or lower category then you can make response category like Yes/No and make analysis by Multiple response analysis.

Dear Md. Saif Uddin Rashed...

indeed you are correct, i can use multiple regression without even categorized it.

however i do need to categorized it for descriptive purpose, because one of my research question is to know how high is the level of these variables in my participant.

Thanks... have a nice day...

:)