We conducted this experiment to test modifications to some of the materials in the replacement mail questionnaire package. Our goals were to reduce the complexity of the paper questionnaire package by removing some of the inserts, and to soften the response mode choice (paper or Internet) in the messages included in the package. We evaluated return rates, section completion rates, item nonresponse rates, and response distributions to assess the impact of the changes on respondent behavior. Our findings are documented in this report.