With the rise of smartphone ownership, some respondents find it convenient to take surveys on a mobile device despite researchers’ expectation that survey completion will occur on a computer. This "unintentional” mobile data collection can produce survey rendering problems that make it difficult for respondents to complete surveys. This presentation offers results of an experiment to determine how data quality and other survey measures differ based upon the device used to complete the survey and whether steps were taken to optimize the survey for the data collection mode. Importantly, this research involved random assignment of respondents to different data collection devices, allowing any observed differences to be attributable largely to the data collection method itself, and not to characteristics of respondents choosing a particular data collection device.