Citizen science approaches in ecology have recently become increasingly popular. Although many advantages, such as the cost-effective collection of vast amounts of data, outweigh the disadvantages, most projects face difficulties, such as non-random sampling, pseudo-absences or various biases, such as detection/reporting biases or participant-related biases. To unravel some of the environmental and socio-economic factors underlying data provision occurring non-randomly, we analysed the geographically tractable record-level databases of 17 separate citizen science projects in ecology and conservation in Hungary. We matched the records to an independent administrative dataset to identify those environmental and socio-economic predictors that are expected to shape participant activity, which varies widely according to the purpose, subject and other characteristics of the projects. Despite the projects' variation, we were able to identify general patterns linking population density of a given municipality and the proportion of protected areas with participant activity. Both variables were significantly associated with the number of observations. If the most urbanised and densely populated capital was left out of the analysis, both the level of education and the proportion of elderly people were positively associated with the number of observations a project received. However, the relationship between a population's socio-economic status and participant activity varied greatly across particular citizen science projects. Our results highlight that citizen science participation is shaped by both environmental context and socio-economic characteristics, revealing systematic spatial biases in data provision. Our results thus provide new insights into the methodology and design of future citizen science projects. Read the free Plain Language Summary for this article on the Journal blog.