The Gemini redesign proposal for the Consumer Expenditure Survey includes an emphasis on records to aid in the reporting of difficult-to-recall expenditures. Research was needed to understand how this new interview protocol would affect the length, burden, and respondent use of records during the interview. For this study, two new protocols were tested: 27 participants were assigned to a “respondent-track“ protocol and 25 participants to an “interviewer-track“ protocol. In the respondent-track protocol, participants determined the order of questioning and maintained control of records. In the interviewertrack protocol, interviewers followed a fixed order of questioning and controlled the records during the interview. Analyses show that participants in the respondent-track group reported significantly more of their expenses using records as compared to the participants in the interviewer-track group. This difference may be explained by two causes: an interviewer effect, whereby one interviewer had significantly less records usage than the other five interviewers; and more use of electronic records in the respondent-track group. And, qualitative feedback from participants emphasized the importance of advance communication for setting participant expectations and understanding of the interview content.