Experimental Design Details
To test the motivational effect of the upward bias on the belief in one’s present bias, we conduct a randomized control trial in a classroom setting. We ask students to take a 7-day challenge, which involves completing 3 exercise questions every day for 7 days. We repeat this challenge twice: once at the beginning of the semester and once at the end of the semester but before the final exam. The first challenge is unincentivized, helping students understand the rule and form beliefs about their present bias. The second challenge is incentivized by course credits, with students earning 0.1 points for each question completed. Before participating in the second challenge, students are asked to choose between receiving 1.5 points directly or participating in the challenge.
Students are randomized into three treatments: ex-ante perfect memory, ex-post perfect memory, and imperfect memory. Students in the ex-ante perfect memory treatment are aware before participating in the first challenge that they will be reminded of their completion status of the first challenge before they choose to participate in the second one. Students in the ex-post perfect memory treatment are not aware but receive reminders, nonetheless. Students in the imperfect memory treatment do not receive reminders about their completion status of the first task before they choose to participate in the second one.
Our first hypothesis is that students in the ex-ante perfect memory treatment will, on average, be more likely to complete the first challenge than those in the other two treatments because, when failures cannot be forgotten, students must persevere to maintain an optimistic belief about their present biases. Our second hypothesis is that students in the ex-post perfect memory treatment will be less likely to sign up for the second challenge than those in the imperfect memory treatment or the ex-ante perfect memory treatment because they have a less optimistic belief about their present bias.
Our randomization is done at the individual level, and we use stratified randomization based on Survey 1 answers and past GPA.
The experiment includes two surveys and two challenges. We explain each item in chronological order.
Survey 1:
Students are informed about the two challenges: the first challenge is not incentivized, but the second one is. They are also informed that if they decide not to take the second challenge, they will earn 0.5 points directly. Students are then given an example of a review question contained in Challenge 1 and asked to predict their performance in Challenge 1:
1. To achieve the best learning outcome, ideally, how many questions will you complete?
2. Predict how many questions you will actually complete.
3. Other course-related questions include how confident they are with this course and how much time they spend on this course weekly.
If their predictions are within two questions of their actual completion, they earn 0.1 points of course credit. Completion of this survey earns them 0.9 points.
Students in the ex-ante perfect memory treatment are told that they will be reminded about their ideal, predicted, and actual completion later this semester (in Survey 2).
Challenge 1:
Students are asked to complete 3 multiple-choice questions per day for 7 consecutive days. They are not rewarded with course credits for completion. To ensure that different students have a similar cost of completing this exercise, we use an interactive interface, showing students the correct answer along with explanations after they make an attempt. Students must select the correct answer before they can move on to the next question. To avoid random clicking, students are required to stay on the question and answer page for at least five minutes before moving on to the next.
Reminder:
In the email that sends out Survey 2, students in the ex-ante perfect memory and ex-post perfect memory treatments are reminded by email about their ideal, predicted, and actual completion of Challenge 1. Students in the imperfect memory treatment are not reminded.
Survey 2:
Students are informed again about the rule of Challenge 2.
1. To achieve the best learning outcome, ideally, how many questions will you complete?
2. Predict how many questions you will actually complete.
3. You will get 0.5 points directly if you choose not to participate in Challenge 2. Would you like to participate in Challenge 2?
If their predictions are within two questions of their actual completion, they earn 0.1 points. Completion of this survey earns them 0.9 points.
Challenge 2:
Challenge 2 is similar to Challenge 1, except that students earn 0.1 point per question.
Based on our theoretical predictions, we expect to observe the following treatment effects and heterogeneous effects.
Treatment effects:
Hypothesis 1:
Students in the ex-ante perfect memory treatment will have a smaller gap between their ideal and actual completion of Challenge 1 than those in the other two treatments.
Hypothesis 2:
Students in the ex-post perfect memory treatment will have a lower sign-up rate for Challenge 2 than those in the other two treatments.