Increased interest in managing seeding rates within soybean fields is being driven by the advances in technologies and the need to increase productivity and economic returns. A wealth of previous research was focused on studying how different seeding rates affect soybean yields at small-plot scales. However, little is known how different site-specific factors influence the responsiveness of soybean to higher or lower plant population densities at field levels, especially across geographic areas with similar soils, weather, and management conditions. In addition, there is no system that farmers can use to evaluate various recommendations for variable rate seeding. The objective of this study was to use on-farm observations to identify major factors that affect yield response of soybean to seeding rates that are slightly above or below the planting rates currently used by farmers. Between 2009 and 2011, farmers conducted 83 field-scale replicated strip trials across Iowa with two soybean seeding rates, high, about 395 K seed ha^-1 and the low, about 340 K seed ha^-1. The two seeding rates were replicated at least four times in each trial. Yield responses to the higher seeding rates were estimated at 30-m grid patterns within each field. Hierarchical modeling and Bayesian analysis were used to identify field and within field-level factors that had significant effect on yield response to the higher seeding rate. For the field-level factors, we considered soybean row spacing, soybean planting dates, monthly and cumulative growing season rainfall. For the within field-level variables, we used relative elevation, slope, soil drainage class, crop suitability rating index, and soil organic matter levels. The Bayesian analyses helped to quantify the uncertainty in the parameters of observed yield response distributions and make predictions for potential yield responses to higher or lower seeding rates at field and within-field areas not studies but assume to have similar crop management and weather conditions. Based on estimated predictive posterior probabilities of profitable yield response (a yield increase above the marginal cost for the seeds) to higher soybean seeding rates, a decision management system was developed that would help farmers and agronomists make economic decisions where to increase or decrease soybean seeding rates within and across fields.