Why You Should Not Dichotomize Continuous Variables

QuantFish instructor Dr. Christian Geiser explains why you should not dichotomize continuous variables and what you can do instead.
#Mplus #statistics #geiser #quantfish #mplusforbeginners #statisticstutorials #statistician
FREE weekly stats tips: christiangeise...
FREE statistics workshops: www.goquantfis...
Get Dr. Geiser's FREE MPLUS SAMPLE SIZE PLANNING & SIMULATION MINI-COURSE: www.goquantfis...
FREE PATH ANALYSIS & MEDIATION MINI-COURSE:
www.goquantfis...
FREE INTRODUCTION TO lavaan MINI-COURSE:
www.goquantfis...
MPLUS FACTOR ANALYSIS COURSE: www.goquantfis...
CFA & SEM WITH MPLUS COURSE: www.goquantfis...
LONGITUDINAL SEM WITH MPLUS COURSE: www.goquantfis...
Dr. Geiser's BOOKS:
****Data Analysis with Mplus: amzn.to/3eg9vY4
****Longitudinal Structural Equation Modeling with Mplus: amzn.to/3ekOLOW
ON-DEMAND MPLUS COURSES:
****Mplus from Scratch: www.goquantfis...
****Multilevel Modeling with Mplus: www.goquantfis...
GET IN TOUCH with DR. GEISER:
****Facebook: / christiangeiserauthor
****Website: christiangeise...
ABOUT THIS INSTRUCTOR:
Dr. Christian Geiser is a quantitative psychologist, author of two books on Mplus, and a leader in the development of latent variable techniques for complex data. With his accessible books and sought-after workshops, he has helped thousands of researchers and students around the world to achieve their analytic goals. Sign up for his newsletter at christiangeiser.com.
ABOUT QUANTFISH:
We bring top-level stats training from the world's leading methodologists to researchers and students all over the world. Learn more at goquantfish.com.
Some affiliate links may be included at no additional cost to purchasers.
Get Dr. Geiser's FREE POWER ANALYSIS MINI-COURSE: www.goquantfis...
MPLUS FOR BEGINNERS COURSE: www.goquantfis...
Dr. Geiser's BOOKS:
****Longitudinal Structural Equation Modeling with Mplus: amzn.to/3ekOLOW
****Data Analysis with Mplus: amzn.to/3eg9vY4
ON-DEMAND MPLUS COURSES:
****MPlus from Scratch: www.goquantfis...
****Multilevel Modeling with Mplus: www.goquantfis...
****Latent Class Analysis with Mplus: www.goquantfis...
GET IN TOUCH with DR. GEISER:
****Facebook: / christiangeiserauthor
****Website: christiangeise...
ABOUT THIS INSTRUCTOR:
Dr. Christian Geiser is a quantitative psychologist, author of two books on Mplus, and a leader in the development of latent variable techniques for complex data. With his accessible books and sought-after workshops, he has helped thousands of researchers and students around the world to achieve their analytic goals. Sign up for his newsletter at christiangeiser.com.
ABOUT QUANTFISH:
We bring top-level stats training from the world's leading methodologists to researchers and students all over the world. Learn more at goquantfish.com.
Some affiliate links may be included at no additional cost to purchasers.

[@anasbit2]

Thank you.
Is there a situation where dichotomizing can be reasonable?
I heard about dichotomizing skewed variables. What do you think about that, or is it the same.

[@QuantFish]

Even with a skewed variable, you would loose information.
One reasonable situation that I can think of is individual selection where you have to make a binary decision (e.g., job selection, clinical diagnosis, college admission, etc.)
Best,
Christian Geiser

[@neiltalbert7091]

In the case of significantly skewed variables, you could try nonlinear transformation, e.g., taking the square root of the variable.

[@kurtpiron838]

Would this apply to established cutoff points for validated scales, eg a score above 7 suggests psychological distress level indicative of psychopathology?

[@QuantFish]

I think it is reasonable to define certain cut-off values for making, e.g., individual selection decisions (e.g., job selection) and/or for clinical diagnoses (e.g., clinically depressed vs. non-depressed). But for statistical analyses of inter-individual differences, it would generally be preferred to look at the variables in their original full metric (non-discretized continuum of scores).
Best,
Christian Geiser

[@kirasun4413]

Thanks, Dr. Geiser! I'm doing a bifactor SEM. Part of the indicators were measured on a 5-point Likert scale and the other part of the indicators were binary. In this case, do I need to dichotomize the 5-point variables and redo the SEM as a sensitivity analysis?

[@QuantFish]

No, you can use the variables as is. You should define them as CATEGORICAL in the Mplus VARIABLE command so they can be properly treated as ordinal (ordered categorical).
Best, Christian Geiser