Risky choice behaviour often deviates from the predictions of normative models. The information search process has been suggested as a source of some reported "biases". Specifically, gaze-dependent evidence accumulation models, where unfixated alternatives' signals are discounted, propose a mechanistic account of observed associations between eye movements, choices and response times, with longer fixated alternatives being chosen more frequently. It remains debated, however, whether gaze causally influences the choice process, or rather reflects emerging preferences. Furthermore, other aspects the information search process, like the order in which information is inspected, can be confounded with gaze duration, complicating the identification of their causal influences. In our preregistered study 179 participants made repeated incentivized choices between two sequentially presented risky gambles, allowing the experimental control of presentation duration, order, and format (i.e., alternative-wise or attribute-wise). Across presentation formats, we find evidence against an influence of presentation duration on choice. The order in which participants were shown stimulus information, however, causally affected choices, with alternatives shown last being chosen more frequently. Notably, while gaze-dependent accumulation models generally capture effects of gaze duration, causal effects of stimulus order are only predicted by some models, identifying potential for future theory development.