What is the problem, and who says

Content provided by the Faculty of the Harvard Medical School

What is the problem, and who says

In deciding how to evaluate the problems of their clients and patients, therapists inevitably rely largely on what they say about themselves in interviews, on questionnaires, and in standard psychological tests. But we don’t always know ourselves, or see ourselves as others see us. And our feeling, thinking, and behavior change with the time, the place, the situation, and the company. So it’s important to study how well reports agree or disagree when they come from different people in different settings.

It’s easiest to get reports from others about children because adults usually bring them for evaluation and treatment. Meta-analyses (pooled statistical analyses) of many studies show that children’s ratings of their own problems are correlated poorly with ratings by their parents and other adults in their lives. But the adults don’t agree with one another, either — at least not when they encounter children in different contexts, such as home, school, therapy, or a research lab. Adults who play similar roles in the child’s life — as parents, teachers, mental health workers, or research observers — are much more likely to agree about the nature of the child’s problems, although even then their agreement is imperfect.

But maybe that is to be expected. Children are often not good at talking about themselves, and their state of constant change may make their problems difficult to pin down even for people who know them well. Researchers at the University of Vermont have tried to find out whether adults present a more stable target. They conducted a meta-analysis of more than 100 studies that evaluated how well different informants (for example, self versus husband, wife, or clinician) agreed about an individual adult’s psychopathology.

They found somewhat greater agreement than in the studies of children, but most of the correlations were still not statistically significant. Agreement was particularly poor for “internalizing” conditions like depression and anxiety. It was somewhat better — no surprise — for problems like aggression and antisocial behavior, with symptoms that are more public and less subjective. But different observers agreed substantially only on drug abuse and alcoholism.

These findings imply that people function differently in different situations and social roles, and that evaluating their needs may require reports from a wider range of observers than is usually available to therapists.

But is it practical to get that information? The Vermont investigators cite the success of a national survey team that asked each participant to complete a self-report questionnaire and then nominate someone else to answer a parallel questionnaire about him or her. Most of the survey subjects were happy to oblige. Today, inexpensive and brief self-report questionnaires are becoming more popular, and the Vermont group suggests developing parallel versions for clients to pass on to friends or family members. The results could be combined and weighed using modern computer techniques, they say.

Comparing individual responses would be a way to flag areas of disagreement with implications for therapy. For example, is this person aggressive toward most people in most situations, or should treatment be more narrowly focused? Reports from self and others could also be weighed differently depending on the type of problem; for example, we might want to judge anxiety mainly by what people say about themselves and anger more by what others say about them.

Achenbach TM, et al. “Assessment of Adult Psychopathology: Meta-Analyses and Implications of Cross-Informant Correlations,” Psychological Bulletin (May 2005): Vol. 131, No. 3, pp. 361–82.

Last updated:	August 21, 2006