Effectiveness, Ethics, and Deinstitutionalization

One person feeling better after therapy proves nothing on its own. People seek therapy when they are at their worst, and things tend to improve from there anyway, a pattern called regression to the mean. Life changes, social support, and simple time can all produce relief that has nothing to do with the treatment. To isolate the treatment's effect, researchers need a control or comparison condition, a group that is not getting the therapy, or getting a different one, so the researcher can see whether the treated group improves more than the comparison group would have improved regardless.

Any single study is small and noisy, so psychologists combine many studies using meta-analysis. Each study produces an effect size, a number describing how much better the treated group did than the control group. Because bigger, better-run studies give more reliable estimates, each effect size is weighted by its sample size before being averaged, so a study with 400 people counts for more than one with 20. The result is one pooled effect size, often displayed on a forest plot where each study is a horizontal line showing its effect size and confidence interval, and a diamond at the bottom shows the pooled estimate across all of them.

This same kind of evidence transformed psychiatric care. Before the 1950s, hospitals held large numbers of patients with severe disorders for years, often with no effective treatment. The arrival of psychotropic medications, especially antipsychotics, let many patients manage symptoms outside a hospital, driving deinstitutionalization, the mass release of patients into community-based care. Research now shows medication and psychotherapy combined often outperform either alone, which raises ethical questions about access, informed consent, and whether community support actually exists to catch people once released.