The Reliability of Self-Report in Clinical Psychology: Evidence from Outpatient Clinics in Malawi

Abstract

Self-report measures are a common tool in clinical psychology for understanding a person’s mental health symptoms and deciding on treatment plans. But people often wonder how reliable these measures really are, especially when used in different healthcare settings. This particular study took a closer look at the reliability of some frequently used self-report measures. The researcher gathered information from 60 participants who were visiting outpatient mental health clinics in Malawi. These participants filled out standard measures about depression, anxiety, and stress, and then did the same set of questionnaires again two weeks later to see if their answers stayed consistent (this is called test-retest reliability). The researcher also checked how well the questions within each questionnaire went together (internal consistency) using a measure called Cronbach’s alpha, and looked at the stability of the results over time using something called intra-class correlation coefficients (ICCs). The results showed that the questionnaires were quite consistent internally across all of them (alpha scores were between .82 and .91). However, the consistency over the two weeks varied depending on what was being measured. The anxiety measures seemed to be the least stable (ICC = .68). When the researcher talked to the clinicians involved, discovered that there were often noticeable differences between what the self-report questionnaires suggested and the clinicians own professional opinions, particularly for clients with lower levels of literacy or those dealing with cultural stigma around mental health. These results highlight how important it is to adapt assessments based on the specific context and to include other types of evaluations, like those done by clinicians, to make diagnoses more accurate. The study goes on to discuss what these findings mean for how clinicians work and for developing better assessment tools in the future.