The Four Attachment Styles: Bartholomew's Four-Category Model

Search for the types of attachment styles and you will find lists of three, four or sometimes five. The version most often used for adults today has four styles, and it comes from one paper: Kim Bartholomew and Leonard Horowitz’s 1991 test of a four-category model.^[1]

What are the four attachment styles?

Bartholomew and Horowitz started from John Bowlby’s idea of internal working models: the expectations about self and others that people build up through experience with attachment figures.^[4] They reduced these to two questions.

• Model of self: Am I worthy of love and support?
• Model of other: Are other people available and trustworthy when I need them?

Each answer can be more positive or more negative. Crossing the two gives four combinations, and each combination is one of the four attachment styles.^[1]

Style (1991 name)	Model of self	Model of other	Typical pattern	Name on this site
Secure	Positive	Positive	Comfortable with closeness and with independence	Secure
Preoccupied	Negative	Positive	Seeks closeness and approval, worries about being valued	Anxious
Dismissing	Positive	Negative	Values independence, plays down the need for others	Avoidant
Fearful	Negative	Negative	Wants closeness but expects rejection, so keeps distance	Fearful-avoidant

In everyday terms, a secure person tends to think: I am worth caring about, and people will usually be there. A preoccupied person tends to think: people could be there, but I am not sure I am enough for them to stay. A dismissing person tends to think: I am fine, and I would rather not count on anyone. A fearful person tends to think: I want people close, but I expect to be let down, and I may deserve it.

The descriptions in the table are short summaries. Real people are rarely pure types, and the styles describe tendencies in close relationships, not someone’s whole personality.

Why a fourth style was needed

Four years earlier, Cindy Hazan and Phillip Shaver had carried Ainsworth’s three infant categories into adult romantic relationships: secure, avoidant and anxious/ambivalent.^[3] Bartholomew noticed that the avoidant group mixed two quite different kinds of people.

Some avoid closeness from a position of apparent self-sufficiency. They feel fine about themselves, see little value in depending on others and keep emotional distance without much visible distress. Bartholomew called this style dismissing.

Others avoid closeness because they fear it. They would like intimacy, but they expect to be hurt or rejected, and they doubt their own worth. They pull away to protect themselves, often while longing for connection. Bartholomew called this style fearful.^[1]

The two groups look similar from the outside, since both keep people at a distance. The reasons and the inner experience differ, and so do the paths toward feeling more secure. The article on dismissive vs fearful-avoidant compares them in detail.

How were the four attachment styles measured?

The 1991 paper did not rely on a single questionnaire. In the first study, Bartholomew and Horowitz developed an interview about close relationships that gave each person both a category and a separate rating on each of the four styles.^[1] The way the four ratings correlated with each other fit the proposed model. The ratings also lined up with self-report measures of self-concept and interpersonal functioning, and each style showed its own typical interpersonal problems, according to both the participants and their friends.^[1] In a second study, interviews about the family of origin and about peers were rated independently, and the two sets of ratings were related.^[1]

Alongside the interviews, participants read four short paragraphs, each describing one style, and said which fit them best. This short self-report became known as the Relationship Questionnaire. It is quick, but a single choice among four paragraphs throws away information: someone who half fits two descriptions has to pick one.

Later work moved toward longer questionnaires with many items. In 2000, Chris Fraley, Niels Waller and Kelly Brennan analysed four widely used adult attachment measures, including one by Griffin and Bartholomew and the scales from Brennan, Clark and Shaver, with item response theory. They found that all of them could be improved and showed how to build scales with better measurement properties.^[7] These newer tools report scores on two dimensions rather than a category.

Types of attachment styles: how the models line up

Different research traditions use different lists, which is a common source of confusion.

Framework	Categories	What it describes
Ainsworth et al. (1978), Main & Solomon	Secure, avoidant, resistant, disorganized	Infant behavior toward a caregiver in the Strange Situation
Hazan & Shaver (1987)	Secure, avoidant, anxious/ambivalent	Adult feelings about romantic closeness
Bartholomew & Horowitz (1991)	Secure, preoccupied, dismissing, fearful	Adult models of self and other
Brennan, Clark & Shaver (1998)	Two scores: anxiety and avoidance	Adult attachment as a position on two dimensions

The infant categories and the adult styles share names and ideas, but they were measured in very different ways: one by watching one-year-olds, the others mostly by asking adults about their feelings. This is also why fearful-avoidant and disorganized attachment in infancy are related but not identical.

Mapping the four styles onto two dimensions

Bartholomew’s two models turned out to have a close relative. Dale Griffin and Kim Bartholomew showed that the model of self and the model of other correspond to two underlying dimensions found across adult attachment measures.^[2] When Brennan, Clark and Shaver later analysed a wide range of questionnaires, they found two factors they named attachment anxiety and attachment avoidance.^[5]

The correspondence is roughly this:

• A negative model of self goes with high attachment anxiety.
• A negative model of other goes with high attachment avoidance.

So the four attachment styles are four regions of a two-dimensional space. Secure is low on both, preoccupied is high on anxiety, dismissing is high on avoidance, and fearful is high on both.^[2][6] The article on attachment anxiety and avoidance explains why many researchers now prefer to report the two scores directly.

Are attachment styles categories or dimensions?

Whether people really fall into distinct types or simply differ by degree is an old question in adult attachment research. Fraley and colleagues tested it in 2015 with taxometric methods, statistical techniques designed to tell a true category from a continuum. In two samples, differences between people looked dimensional rather than categorical, both for attachment in general and for specific relationships such as those with parents and friends.^[8]

In practice this means the four styles are useful names for regions of a continuous space, not four boxes with walls. Picture two friends with almost the same scores, one just above the middle on anxiety and one just below. A category system calls one preoccupied and the other secure; a dimensional score shows how close they are. The same logic explains a common quiz experience: retaking a test a few weeks later and landing in a neighbouring style. Near a border, a small change in answers is enough to flip the label, even when the scores barely move.

Using the four styles wisely

Many people read the four styles and recognise themselves in more than one. That is not a sign of a bad model or of a confused reader. It follows directly from the dimensions underneath: someone in the middle of both scales shares features with all four quadrants.

The four-category model is useful because it is easy to grasp and it names a real difference that the three-style model missed. It has limits too:

• Borders are blurry. The evidence favors dimensions over types, so the labels mark regions, not separate groups.^[8]
• Styles can vary by relationship. A person may feel secure with a sibling and fearful with a new partner.^[6]
• Styles can change. Working models were never meant to be fixed, and adult attachment research treats them as open to revision.^[4][6]
• A style is not a diagnosis. It describes a pattern in how someone handles closeness, not a disorder.

Our attachment style quiz reports both dimension scores and the quadrant they fall into, and it names two styles when you sit near a border. How the quiz works explains the method and what it cannot tell you. If you are struggling with relationship patterns, getting support has pointers to professional help.