In this paper, we employ a framework that allows us to jointly model the response to the stated choice component as well as the response to the attribute processing questions for organic product attributes among consumers in Nigeria. The model allows us to make use of respondent reported information on processing strategies and conditions attribute parameters on underlying latent attribute importance ratings, while avoiding the potential endogeneity bias and measurement error problems arising with traditional methods. Using between-sample design, we compare the welfare estimates from respondents under cheap talk and honesty priming treatments. Our results suggest that the HP task leads to lower WTP values by a factor of two relative to cheap talk task, for three of the four attributes identified. Our findings also reveal some modest impacts on implied WTP patterns, with a more realistic difference between the valuations for the attributes, and lower overall heterogeneity relative to the commonly used mixed multinomial logit model.