Low response rates for the Agricultural Resource Management Survey Phase III (ARMS III) are a growing concern. Beginning in 2011 and enhanced in 2012, data mining techniques were used to proactively identify operations that are likely to be nonrespondents to ARMS III. Once identified, a measure of the impact on the calibration targets of each likely nonrespondent was calculated. Likely nonrespondents were then divided into a treatment and control group to evaluate data collection methods. Likely nonrespondents in the treatment group were targeted with more intense data collection strategies in an effort to increase response rates. These strategies were evaluated to determine whether they could increase response to the survey but because these strategies were not uniformly applied, definitive conclusions about their effectiveness cannot be drawn.