To explore the psychology of voting secrecy, we conducted a field experiment to examine voter sensitivity to arrangements for ballot paper selection under the French ballot system (i.e., multiple ballot papers). Working closely with Swedish election authorities, we randomly assigned participants to vote in a fictional election under low, medium-high and high privacy conditions with a follow up paper-and-pen survey to record perceived voting secrecy. Results show that participants perceived an arrangement for selecting ballot papers behind a closed screen as clearly more secret than one where selection was public, and that a third and even more private arrangement, where voters did not have to walk a few meters with ballot papers in hand, potentially visible to onlookers, provided even higher levels of perceived secrecy. The study demonstrates that voters are sensitive even to small changes in the voting environment. Accordingly, election administrators should be aware that details matter for the experience of voting secretly.