Help in categorizing data

#1
Hi,
I am working on my dissertation and am stuck with the data analysis. :confused:

I am dealing with 4 variables, all of which are categorical variables. The unit of my data analysis is a dyad while the unit of data collection is an individual.

All the variables are measured on an ordinal scale (likert like) and they have to be categorized. The variables are value similarity (to leader and to organization), leadership views and commitment.
One variables, commitment, has to be categorized into high and low. (Thinking of a mean + or - half std. dev. split? or something else)
The other variable, leadership, has to be categorized into high and low. (Thinking of a median split? though there is a lot of literature against it)
Also for value similarity i need to compare the scores of the boss and subordinate and then categorize it. I have no clue how to do it.

Please help me figure this out. I would appreciate all help.
 
Last edited:

terzi

TS Contributor
#2
Hi natashaakaul,

Actually, I would argue against (almost) any form of categorizing data in a way that reduces information. In your case, a likert scale may possibly lead to better and more interesting results than dichotomous measures. If you give us more details or explain why you want to categorize your data that way, we could get a better glimpse. Still, I'd search for a better solution that does not involve that transformation, since there are many powerful statistical techniques which can be used for that type of data.