Turning self-worth into a number
Self-esteem is a person’s overall evaluation of their own worth. Because it cannot be observed directly, researchers rely on self-esteem scales: short questionnaires in which people rate how far they agree with statements about themselves, such as feeling satisfied with themselves or having qualities worth having. The answers are scored and summed, giving a number that can be compared across people or across time.
This is a practical approach, and several widely used scales have been tested repeatedly for consistency. But the step from a questionnaire score to a claim about social media involves several assumptions worth understanding. This article is educational and offers no advice about anyone’s wellbeing.
What the scales capture and what they do not
- Overall versus specific. Many scales measure a general sense of worth, while people also hold views about particular areas such as appearance, ability or popularity. A general score can hide a change in one area.
- Trait versus state. Self-esteem has a fairly stable component, which changes slowly, and a momentary component, which rises and falls with events and mood. A single survey mixes the two.
- Self-report. People answer according to how they see themselves and how they wish to appear. Social desirability and question wording can shape the result.
- Cultural differences. Norms about expressing pride or modesty vary, so the same score may not mean the same thing in different places.
- Rough scoring. A small difference between two scores may be within the range of ordinary measurement error.
Because of these features, a number such as “average self-esteem was lower among heavy users” does not by itself say how large the difference is in everyday terms.
Social comparison theory
A common explanation is social comparison theory: people evaluate themselves partly by comparing with others. Platforms make comparison easy by displaying others’ achievements, appearance and social lives, often in selected and polished form. If people compare upward and feel they fall short, self-esteem might fall; downward or affirming comparisons might raise it.
The theory is useful for generating hypotheses, but evidence is mixed. Whether comparison happens depends on the person, the content and the reason for browsing. Some people compare often and others rarely. Also, receiving positive feedback, such as supportive comments, could have the opposite effect. Researchers sometimes measure comparison tendency directly with a separate questionnaire, which then raises its own self-report issues. The measures used for engagement metrics, such as likes, do not tell us how a person interpreted them.
Why a one-off score can mislead
Imagine a survey that finds higher self-esteem among people who use a platform less. Several explanations fit equally well.
- Heavy use might lower self-esteem.
- Lower self-esteem might lead to heavier use, an instance of reverse causation.
- A third factor, such as stress, sleep or life events, might affect both, which would be a confounding variable.
- The sample might not represent the wider population.
Such rival explanations cannot be separated by a single snapshot. Even when two surveys give the same average, the underlying story can differ.
Designs that help, and their limits
- Longitudinal surveys measure the same people several times. They can show whether changes in use tend to come before or after changes in self-esteem, though outside events remain possible.
- Diary studies capture self-esteem repeatedly during the day, linking brief shifts to what someone just did. They are better at short-term effects, but the burden on participants limits size and duration.
- Experiments ask people to reduce use or to view particular content, then measure self-esteem. They provide more control but are usually short, and participants know they are in a study.
- Natural experiments take advantage of events such as a service outage or rollout, which can offer cleaner comparisons but are rare and depend on circumstance.
Each design offers a different compromise. Combining them gives a fuller picture than any one alone.
Who and what is averaged
Study results typically describe averages, and averages conceal variation. A small average difference could reflect a large change for a few people and none for most. This is why group results from studies should not be read as predictions for individuals.
When reading a headline, it helps to ask which scale was used, whether self-esteem was measured at several points, what the size of the difference was, and whether the article treats association as cause. For concerns about self-worth or mood, a qualified health professional is a better source of help than a general overview of study design.