Method of recruitment and the scores of self-report measures: The example of worry in the elderly

Due to the difficulties in recruiting elderly people for psychological studies, a wide range of recruitment methods has been used. The underlying assumption that the recruitment method does not seriously affect the results has to be tested. Results in different worry measures are compared between three samples which were recruited with different strategies: (i) randomization sampling from the census of an urban area (random sample; n = 97); (ii) people involved in leisure activities in senior centers (convenience sample; n = 142); and (iii) people from the close environment of psychology students (snowball sample; n = 60). People from those samples, aged 65 and more, completed the Penn State Worry Questionnaire (trait-worry), and the Worry Scale (contents of worry). Differences were found for several demographic characteristics. There were marked differences between samples in the percentage of individuals endorsing cut-score criteria for generalized anxiety disorder (GAD), with a very high percentage for the snowball sample. For all of the worry measures, analyzes of variance revealed significantly higher total scores for the non-randomly selected samples (p < 0.01). Comparing the latter samples, higher trait worry (p < 0.01) and more worrying about health and personal competencies (p < 0.01) was found in the snowball sample. These results suggest that the sample recruitment method can specifically and seriously affect the outcomes of studies with elderly people, thus limiting the generalization of their results.