The first chapter of this thesis is motivated by a puzzle in consumer finance. In high-stakes financial decisions, people leave a substantial amount of money on the table, even when financial education is available. The ubiquity of financial choices makes understanding the effects of incentives and education on mistakes crucial. This chapter experimentally examines the impact of changes in incentives and educational availability on incentivized but hypothetical healthcare choices using Amazon Mechanical Turk. We find that increasing incentives are ineffective in increasing decision-making effort, even when these changes are made clear and salient to the subjects. Yet, surprisingly, despite this lack of effort response, subjects' choices improve when incentives are high. This result highlights an under-appreciated channel of incentives: when stakes become larger, often, the problems become simpler too. We next investigate the effect of available education. Overall, education leads to an increase in decision-making effort and an improvement in choice quality. However, this average effect masks significant heterogeneity across incentive treatments. Subjects are willing to put in the educational effort when either the problems are hard or mistakes are highly costly, but the return of the educational effort is zero for hard problems and positive for easy ones. Thus, only when stakes are high and the problem is easy does education have an effect. These findings suggest that people can be encouraged to get education for high-stakes decisions, and policy-makers have a role in simplifying problems to translate the extra effort into better choices. The second chapter dives deeper into the "easiness" channel of incentives. This chapter uses an experiment to disentangle "easiness" from the standard incentives on savings account choices, again using Amazon Mechanical Turk. We show that increasing the variance of the accounts improves choices without increasing time spent. This is true in both between-subject and within-subject analyses. Besides, we re-investigate the effects of incentives and education. We recover the usual effects of incentives, where paying subjects a higher rate motivates them to spend more time and do better. We also find that easiness and incentives complement education. Consistent with the literature, we show that the effectiveness of education diminishes with time, suggesting that education should be provided as and when people make decisions. In the third chapter finds experimental evidence for preference for flexibility (PFF). Although PFF is very intuitive, documenting PFF experimentally faces challenges from stochastic choices. Because there are random noises in decision-making, experimental data may over-estimate PFF due to such randomness. This chapter tackles stochastic choices by first deriving theoretical upper bounds for PFF. We then measure PFF against these upper bounds using menu choices presented in the Multiple Price Lists (MPLs) in a lab experiment. We find that subjects exhibit more PFF than what can be explained by random noises. Specifically, there are more PFF than two countervailing behaviors, indifference and preference for commitment. We then present two alternative models for PFF, which have different predictions for the effect of the probability of payment on PFF. We suggest a modified experiment to tell these two models apart for future research.