If I understand your question, you already have data on a Likert scale but would like to transform it to binary outcomes.  I would pick a threshold that allows a reasonable split; you would not want a threshold where 99% of the people are "yes", for example.

Dear Nazneen Akhter 

                       The variable will change its construct and originality. It is better  to keep it like that. 

Mehraj 

Dear All,  who responded my question, many thanks , the response came quite diverse but,  it widen my horizon of understanding on this issue how I can make it up in different angle. Thanks Henry T Robertson, in fact , I advised the student to go in that format. She is already done with her thesis paper &  Thesis defense also completed

Thanks,  Dr. Nazneen