Online tests often say you have one "attachment style," like "anxious" or "avoidant." But research shows people do not fit neatly into these boxes.

Instead, scientists use two scales. One measures worry about being left alone. The other measures how much you avoid depending on others. Everyone is somewhere on these scales. Also, your style is not fixed for life. Good relationships can change it.

Attachment theory began with studies of the bond between babies and the people who care for them. In 1987, two researchers applied the idea to adult love. Using answers to a newspaper survey, they suggested adults could be grouped into three styles called secure, avoidant, and anxious. About 56 percent of the people who answered chose the secure style.

Later research, however, found that attachment is better measured with two scales, anxiety and avoidance. A 2015 statistical study showed that the data fit continuous scales better than separate types, whether people were thinking about parents, friends, or partners. In addition, attachment is only moderately stable from childhood to adulthood. Experiences with trustworthy romantic partners may help people feel more secure.

Quizzes that label people as "anxious" or "avoidant" have become popular online, but research suggests such labels oversimplify human relationships. Attachment theory originated in studies of how infants seek comfort from caregivers, and in 1987 researchers extended it to romantic love, using a newspaper questionnaire to classify adults into secure, avoidant, and anxious styles, with roughly 56 percent describing themselves as secure. That early work relied on a single forced-choice item, and later studies moved toward multi-item scales measuring two dimensions: anxiety about rejection and avoidance of closeness.

In 2015, a taxometric analysis of two large samples tested whether attachment forms distinct categories or a continuum. Across relationships with parents, friends, and partners, the data were better explained by continuous dimensions, implying there is no sharp boundary between being "anxious" and being "secure." Furthermore, a meta-analysis found only moderate stability from infancy to about age 19, and theorists argue that responsive partners can gradually enhance security. That proposal, however, comes mainly from a review, so direct evidence from interventions is still limited. The most scientific use of a quiz is as a snapshot of current tendencies.