Participation of Mexican American Female Adolescents in a Longitudinal Panel Survey

The sources of systematic sample attrition are examined for a community-based panel survey of 1,023 Mexican American and non-Hispanic white female adolescents, 874 (85.4%) of whom were reinterviewed after two years. There were few differences between Mexican Americans born in the United States and non-Hispanic whites, but Mexican Americans born in Mexico were distinctively different. The characteristics of the original interview shape the respondent's concurrent attitude toward repeating the experience but do not affect directly subsequent behavior. The results demonstrate that the validity of panel data may be problematic even when indicators of the survey content do not appear to be associated directly with loss to follow-up. Attrition exerted a greater impact upon the external than internal validity of the panel data.