Quality of information and the ability of producers to act upon information are important factors influencing technological progress. The dissemination by extension agents of information on dairy management practices used to control mastitis and the reception and use of that information by producers is investigated. Producers are surveyed to determine current practices used. The relationship between milk yield, somatic cell count, management practices, and producer and production characteristics is estimated. Subjective probabilities are elicited from "experts", extension agents, and producers concerning the impact and cost of various management practices. Subjective marginal value products and marginal input costs are computed and compared for the respondent groups. Stochastic dominance is used to rank the relative importance of the practices as perceived by the respondents. A dual model is used to examine aggregate technical change and the dynamic structure of the U.S. dairy industry. Properties implied by the theory of the competitive firm and independent adjustment of two quasi-fixed inputs, labor and herd size, are tested and not rejected. Instantaneous adjustment, however, is soundly rejected for each quasi-fixed input. Input adjustment to optimal levels is estimated to take about two years for labor and ten for cows. Robustness of dynamic dual model results to choice of functional form is also examined. Modified generalized Leontief (GL) and normalized quadratic (NQ) functional forms are compared by examining their consistency with properties of the competitive firm, estimated rates of adjustment for labor and cows, tests on technological change, and elasticities. Homogeneity and symmetry is maintained in both models. Convexity is not rejected by the GL but is rejected by the NQ. Absence of technological change is rejected by both models, but the quality index on labor fully embodies the technological change occurring within labor in the NQ but not in the GL. Elasticities differ greatly between the functional forms. Dynamic dual models are not found to be robust to the choice of functional form