Evidence of citation bias in the pesticide ecotoxicology literature

As scientists, we are tasked with letting evidence guide our conclusions. In the world of pesticides this takes on added importance as the data can influence ecological and human health outcomes and regulations, and even the manner in which we grow food. Yet, there seems to be a reticence to engage with the totality of the pesticide ecotoxicology literature, especially papers that report few or no effects or low risk to non-target organisms. We suspected that these studies would have fewer citations than studies that report significant effects or risk for the same compound, and this would be unrelated to the strength of the study, e.g., high quality studies with few or no effects would be cited less frequently than weaker studies that reported effects. To investigate this, we examined a subset of literature around the herbicide atrazine. We found that papers reporting an effect had significantly more citations per year than those that did not (p < 0.05). There was no significant relationship between the strength of the study and number of citations, but a general trend for weaker studies to have greater number of citations. The impact factor of journals was not positively correlated with the strength of the study methods, but studies that reported effects were published in journals with a greater mean impact factor than those that reported no effects (p < 0.05). This analysis reveals evidence of citation bias within the pesticide ecotoxicology literature, as well as bias by journals to publish studies that report effects, regardless of study quality.