BACKGROUND: Adolescents spend a majority of time at school, making the classroom a natural venue not only for risk behavior interventions, but for administration of surveys assessing the effectiveness of these interventions. Computerized approaches provide valuable advantages over self-administered questionnaires (SAQ); however, few comparative studies have included the personal digital assistant (PDA) or the PDA with audio-enhancement (APDA). Even less is known about the role of cognitive burden (e.g. reading ability and preferences, language mastery, attentiveness) on survey completion by different mode. PURPOSE: This study has four aims: (1) to understand the impact of three data collection modes (SAQ, PDA, APDA) on the number of survey questions answered, missing data, data consistency, and student evaluation of the survey experience; (2) examine associations between cognitive burden and outcomes; (3) assess moderating effects of cognitive burden on relationships between mode and outcomes; (4) assess the role (mediating or moderating) of perceptions of confidentiality/privacy on relationships between mode and outcomes. METHODS: Two-hundred seventy-five students were recruited from seven urban K-8 schools in the Midwest. Consented participants were stratified based on reading scores and randomized to complete a survey by SAQ, PDA, or APDA. Upon completion, students completed a paper-based debriefing survey, assessing student survey experience. Academic and behavioral assessments were completed by each student’s teacher. RESULTS: APDA respondents completed significantly more questions compared to SAQ and PDA. Both PDA and APDA had significantly less missing data than SAQ. No differences were found for student evaluation by mode. Several measures of cognitive burden were related to outcomes. While data inconsistency (e.g., factorial variance) was found across modes, data from students at varying levels of cognitive burden was not significantly different. One significant moderator was found with the relationship between mode and number of questions varying by perceived reading difficulty. Finally, neither perceived confidentiality nor privacy acted as a moderator or mediator. CONCLUSIONS: This study indicates strong benefits to be gained by the use of APDA for school-based data collection with adolescents, particularly those with reading, language or attention issues. While students with greater cognitive burden answered fewer questions, their data was found to reliable.