Quantifying the unintended effects of research on farmers’ practices and social networks
Code used for the analyses in the paper titled "Quantifying the unintended effects of research on farmers’ practices and social networks". For a full description of the hypotheses, methods used and results, please refer to the paper. In short, we quantified the effects of an agroecological research project on farming practices and the social network in a village in Papua New Guinea. To track changes in farming practices and the community’s social network, we administered two waves of interviews, one at the beginning of the research project and one at the end. We show that the inferences that people draw from researchers' activities may be very different to those researchers are aiming to test. We also show that social networks may not always help spread information, and that research interventions can actively shape relationships between people, even if this is not the aim. Here we present the R scripts used to analyse (1) changes in farming behaviour using logistic regressions with household number as a random effect, (2) changes in farming behaviour using logistic regressions *without* household number as a random effect, (3) the joint evolution of farming behaviour and social networks using stochastic actor-oriented modelling, and (4) the evolution of social networks only using stochastic actor-oriented modelling.