Why Are There Now 20 Schemas Instead of 18?

Written by Jess O'Garr, Clinical Psychologist and Schema Therapy trainer at The Psych Collective.

If you learned schema therapy before 2022, using the original model developed by Jeffrey Young, you probably learned that there are 18 Early Maladaptive Schemas.

Then, at some point, you may have come across a newer schema questionnaire, workbook, or training that suddenly lists 20 schemas.

So, where did the extra two schemas come from?

The short answer is that nobody simply invented two new schemas. Instead, researchers went back and statistically analysed the Young Schema Questionnaire at the individual-item level and found that two of the original schemas were actually measuring two different constructs.

As a result, Emotional Inhibition was split into Emotional Constriction and Fear of Losing Control, while Punitiveness was split into Punitiveness-Self and Punitiveness-Other.

Eighteen schemas therefore became 20.

The research also resulted in a revised questionnaire, the Young Schema Questionnaire-Revised (YSQ-R), which reduced the original 232-item questionnaire to 116 items while retaining strong psychometric properties.

For clinicians, this is a good example of what should happen when a psychological model has been around for a while. We don't have to throw out a clinically useful model, but we also shouldn't keep using exactly the same measurement tools forever simply because that's what we've always used.

Sometimes we get better data. And when we get better data, we should probably use it.

Where did the original 18 schemas come from?

The traditional Schema Therapy model described by Jeffrey Young and colleagues identified 18 Early Maladaptive Schemas, grouped within five broad schema domains.

These 18 schemas became the framework that most clinicians learned and are still what you will see in many schema therapy books, websites, training programs and older resources.

The Young Schema Questionnaire, or YSQ, was developed to help measure these schemas.

The most comprehensive version, the Young Schema Questionnaire Long Form-3 (YSQ-L3), contains 232 questions assessing the original 18 schemas. A shorter 90-item version, the YSQ-S3, was also developed.

These questionnaires have been used extensively in both research and clinical practice.

That doesn't mean they can't be improved.

Over time, researchers examining the different versions of the YSQ produced mixed findings regarding aspects of its factor structure and construct validity. Yalcin and colleagues therefore took a different approach to examining the questionnaire.

Rather than asking whether we should throw out schema theory, they essentially asked:

How well are the individual questions actually measuring the schemas we think they are measuring?

That's where things became interesting.

What did Oz Yalcin's research find?

As part of his PhD research, Australian psychologist Dr Ozgur Yalcin, together with Ida Marais, Christopher Lee and Helen Correia, undertook a detailed psychometric analysis of the YSQ-L3.

Their study included 838 participants, comprising a large heterogeneous clinical sample and a smaller non-clinical sample.

Instead of relying solely on the types of factor analyses previously used to examine the YSQ, the researchers applied Rasch analysis.

You don't need to understand Rasch analysis to use the YSQ-R.

The easiest way I explain it clinically is that the researchers went back and looked at the questionnaire item by item to see whether each question was actually doing a good job of measuring the schema it was supposed to measure.

Of the 232 items in the YSQ-L3, 116 showed misfit across several statistical indicators. When they removed those poorly fitting items, the statistical fit of all schema subscales improved. Most revised subscales showed good to excellent reliability, although Enmeshment remained the notable exception.

Conveniently, that left 116 questions, which became the basis of the YSQ-R.

Why was Emotional Inhibition split into two schemas?

This is one of the changes that makes a lot of sense to me clinically.

The old Emotional Inhibition schema essentially captured the tendency to suppress or restrict emotions.

But think about two patients who both appear emotionally inhibited.

One patient might say:

"I don't show emotion. It's embarrassing. I hate people seeing me upset. I keep everything contained."

Another might say:

"I can't let myself get angry because if I start, I won't be able to stop. I'll completely lose control."

From the outside, both people may appear to suppress their emotions.

But why they are doing it is very different.

Yalcin and colleagues found that the Emotional Inhibition items fitted better when separated into two constructs: Emotional Constriction and Fear of Losing Control.

Emotional Constriction

Emotional Constriction describes excessive control or restriction of emotional expression, particularly where showing emotion is associated with shame or embarrassment.

The person learns to contain emotion, disconnect from it or avoid displaying it.

They may be uncomfortable crying in front of someone. They may struggle to talk about feelings. They may believe emotional expression makes them look weak, embarrassing or out of control.

The emotional system becomes tightly constrained.

Fear of Losing Control

Fear of Losing Control is different.

Here, the concern isn't simply I shouldn't show emotion.

It is closer to:

If I allow this emotion out, something bad will happen.

The person may believe that once they start crying they won't stop, that anger could become uncontrollable, or that allowing themselves to experience intense emotion will result in catastrophic consequences.

Yalcin and colleagues described this construct as involving anxiety about the consequences of emotions not being contained.

Clinically, that distinction can be really useful.

Both patients might suppress emotion, but the belief driving that suppression is different. And if the underlying belief is different, the conversation we have in therapy may also need to be different.

Why was Punitiveness split into Punitiveness-Self and Punitiveness-Other?

The same problem existed with the original Punitiveness schema.

Punitiveness broadly describes the belief that mistakes or wrongdoing deserve punishment and that people should face harsh consequences for getting things wrong.

But who deserves the punishment?

Some people are extraordinarily punitive towards themselves.

Others are punitive towards other people.

And although somebody can certainly do both, they aren't necessarily the same psychological pattern.

The statistical analysis supported separating these into Punitiveness-Self and Punitiveness-Other.

Punitiveness-Self

Punitiveness-Self turns the punishment inward.

If I make a mistake, I deserve to suffer for it.

If I fail, I should pay for it.

I don't deserve compassion because I should have known better.

These patients can be incredibly harsh towards themselves. Mistakes aren't simply opportunities to learn or things that human beings inevitably make. They become evidence that punishment is warranted.

Punitiveness-Other

Punitiveness-Other directs the same basic principle outwards.

If somebody does the wrong thing, they deserve what they get.

People should be punished for their mistakes.

There may be very little room for context, fallibility, forgiveness or compassion when another person violates an expectation or standard.

Again, you can immediately see why separating these constructs can be clinically useful.

A patient who mercilessly attacks themselves every time they make a mistake isn't necessarily the same person who demands punishment when somebody else gets something wrong.

Putting both under one Punitiveness score can obscure that distinction.

Does this mean the original 18 schemas were wrong?

No.

I think this distinction matters because psychological models can sometimes be treated as though a newer version makes everything that came before it suddenly invalid.

That's not what happened here.

There is nothing inherently wrong with using the traditional list of 18 schemas, and there is an enormous amount of schema therapy literature, training and clinical knowledge built around that framework.

Similarly, clinicians aren't suddenly practising schema therapy incorrectly because they have used the YSQ-L3 or YSQ-S3.

The revision is better understood as a refinement.

We had a clinically useful model.

Then researchers accumulated more data and developed better ways of examining how the questionnaire was functioning.

When Yalcin and colleagues examined the individual items using Rasch analysis, the evidence suggested that some items could be removed and that two of the broader schemas were better represented as separate constructs.

So the model became more precise.

Which list of schemas should clinicians use now?

I teach the revised list of 20 schemas and recommend the YSQ-R.

That doesn't mean you need to throw every old schema therapy book in the bin.

Most of the schema model remains exactly as you know it. The important difference is recognising that when you see Emotional Inhibition in older resources, the revised framework separates that into Emotional Constriction and Fear of Losing Control.

Likewise, when you see Punitiveness in older material, consider whether you are actually looking at Punitiveness-Self, Punitiveness-Other, or both.

For me, these aren't annoying semantic changes.

They actually make clinical sense.

Knowing that somebody suppresses emotion is useful.

Knowing why they suppress emotion is considerably more useful.

Knowing that somebody has punitive beliefs is useful.

Knowing whether those beliefs are primarily directed towards themselves, other people, or both gives us a clearer picture of what is happening.

That is ultimately what good psychometrics should do.

It should make our clinical formulation more precise.

Reference

Yalcin, O., Marais, I., Lee, C., & Correia, H. M. (2022). Revisions to the Young Schema Questionnaire using Rasch analysis: The YSQ-R. Australian Psychologist, 57(1), 8–20. DOI: 10.1080/00050067.2021.1979885

Read the YSQ-R research paper