In my dissertation, I investigate the relationships between the non-contractibility in household labor, spousal bargaining power, and labor-intensive technological change. I extend the conjugal contract model introduced by Carter and Katz (1997) to understand (1) why non-contractibility in spousal labor matters and (2) how bargaining power may affect the efficiency loss due to non-contractibility in spousal labor. The model suggests that in households with greater bargaining power for the wife, the wife provides more labor on the farm, even if the husband controls the production. As a result, the household can take better advantage of a labor-intensive technological change that benefits the husband's sphere of control. I then discuss how to measure bargaining power and analyze the impact of a technological change based on spousal bargaining power. In addition, I propose a strategy to mitigate the non-adoption problem when assessing the effectiveness of agricultural technologies. I find that encouraging farmers to self-select into the sample yields more statistically powerful results than selecting from an experimental population. This sampling method is then applied to the randomized controlled trial study design to evaluate the impact of large-scale in-person agronomy training offered to smallholder coffee households in Uganda.