Cooperation in a social dilemma is noble but risky because one is not sheltered against selfish individuals. This “social uncertainty” is essentially captured in the linear public good game, where it is socially optimal to contribute everything to the public good, while it is privately optimal to keep everything for oneself. Many real-world social dilemmas have an additional source of “strategic uncertainty”, as socially and privately optimal strategies tend to depend on actions of others. Here, we compare the determinants of cooperation in a linear public good game and a threshold game, where individuals are challenged to guess the contributions of the partners to determine an appropriate investment that aligns with private and collective interest. We combine elicited risk preferences and cooperative attitudes with information from a survey on social capital and demographics to analyse what explains cooperation. Our experiments are carried out with farmers in Cambodia who are exposed to various social dilemmas on a daily basis. We find that risk and social capital explain cooperation in the linear public good game, but not in the threshold game. These findings call for a more careful examination of real world social dilemmas that typically comprise coordination and cooperation elements.