Z scores + 1

#1
My prof told me that we could model a continous variable as a dichtotomous or even as a trichotomous one without the disadvantages of a median or mean split. He gave me this text passage:

For that purpose the complete data set was used to model a group scoring moderate on health value and a group scoring high on health value by subtracting 1 from the mean centered scores on health value and adding 1 to the mean centered scores on health value, respectively (Cohen, Cohen, Aiken, & West, 2003; Siero, Huisman, & Kiers, 2009).

I am not too sure how to do this... when i simply add 1 to the z scores of my independant variable, nothing really changes - at least not the main and/ interaction effects. And how would I do this, when i want to model a trichotomous variable?
 
#3
Sadly these are whole books... I even own one of these books, but I have no idea where to look, i already read the chapters about zscores but couldnt find information about the aforementionend procedures. I could however post other text passages of papers using these procedures, but it is never really explained or at least i dont understand it.

-To study the pat- tern of persuasion due to the EMDR manipulation in levels of self-esteem, the complete data set was used to model a group scoring low on self-esteem and a group scoring high on self-esteem, using the same procedure as outlined above. - same paper as the citation abov: The Eye Movement Desensitization and Reprocessing Procedure Prevents Defensive Processing in Health Persuasion - Dijkstra & Asten 2013

-The main effect of condition was not significant. To assess the meaning of the interaction, the complete set of data was transformed to model a ‘high disengagement group’ and a ‘low disengagement group’, by adding and subtracting 1 standard deviation of the mean centred disengagement scores, respectively (Cohen, Cohen, West & Chaiken, 2003). - Disengagement beliefs in smokers: do they influence the effects of a tailored persuasive message advocating smoking cessation? Dijkstra, 2008