Since 2019, I am a researcher in environmental toxicology. During my PhD I investigated the effects of a pyrethroid insecticide, Esfenvalerate, increasingly complex ecosystems from individual, via the population, to the community. These works had one thing in common. They were characterized by complex, messy and multidimensional datasets that described various aspects of the ecosystems over time.
The most important insight I developed throughout my PhD was not be be afraid of statistics. When the results are complex and not straightforward there is a lot of value in the attempts to simulate the data from probability distributions connected by causal relationships. I realized that it is much more intuitive to think through a problem in forward direction rather than trying to guess its mechanisms from looking at the outcomes. Simulating outcomes from random numbers became a trusty companion in understanding many hard problems. From this insight the step to Bayesian inference is not a large one, which is based on sampling from probability distribution and gauging the explainatory model based on the data.
Today, I work a lot with modelling these systems and looking at them from a more mechanistic aspect, focusing on the mechanisms of chemical effects inside the organism. The explainatory models became more complicated but the principle remains the same. By formalizing mechanistic models we can hypothesize about simplified causal relationships in the real world, and by “sampling from the imaginary” (McElreath, 2015), and comparing it to the real numbers, we can test if our hypotheses are true.
On this site, I list some of my thoughts along the way. Whenever I encounter a problem that I do not quite understand, I try publish it here. Sometimes it is unfinished work, sometimes just a short note.