Experimental Design Details
Advisor Experiment: The following are the goals of this experiment: 1. Create the advice that will be given in the advisee experiment 2. Test whether advice is endogenous based on an advisee's gender and/or whether the test has a piece-rate structure vs. tournament structure. We are also interested in whether there are interactions between the two. We will explore heterogeneity by the gender of the advisor, whether the advisor knows their part 1 performance, whether the advisor perceives themselves as better at math, and whether the advisor perceives math as more interesting.
Advisors will take a test in which they will answer some math and verbal questions in a randomized order. A randomly chosen half of the advisors will learn about their score whereas the other half won't. Then, they will give advice to the other participants of the research study on whether they should choose to complete the math portion or the verbal portion of the test under 10 different scenarios. In 5 of these scenarios, they are randomly assigned to one of the following scenarios (1) male advisee, tournament condition; (2) male advisee, piece-rate condition, (3) female advisee, tournament condition; (4) female advisee, piece-rate condition. In the other 5 of the scenarios, they will not know the gender of the advisee or whether the advisee is in piece-rate vs tournament condition. These scenarios involve different combinations of correct answers in math and verbal tests in the test advisees will take. Advice answer choices are randomized across subjects.
Advisors' bonus payment will depend on either (1) the number of correct answers that they have in the test they will take in the first part of their survey, or (2) the number of correct answers that their advisee will have on the portion of the test that they will choose to take (math or verbal) after receiving their advice.
Advisee Experiment: The goal of this experiment is to test whether advisees are 1. more likely to follow the advice they received or 2. more likely to choose non-stereotypical test when there is a gender match between the advice and advisor. We will explore heterogeneity by gender of advisee, whether the advice received is stereotypical or non-stereotypical, whether the advisee is informed about the knowledge level of advisor, whether the advisee knows their own score, by payment scheme (tournament vs piece-rate), risk aversion of the advisee, and whether the advisee perceives themselves as better at math.
Advisees will answer 4 math and verbal questions. Then, they will be given a choice between choosing math or verbal portion of a new but similar test. Some of these advisees (randomly selected) will receive advice on what to do before they make their choice. Advisor characteristics will be orthogonal to the advice received. We will assign 1/6th of the subjects into the control group where they will not receive any advice. 1/6th of the subjects will be assigned to a treatment group where they will receive advice but not know the gender of the advisor. 2/6th of the subjects will be assigned to a a treatment group where they will receive advice from a male advisor and 2/6th of the subjects will be assigned to a treatment group where they will receive advice from a female advisor. Half of the subjects in all treatment groups will be randomly assigned to receive advice from a knowledgeable advisor and will know about this whereas the other half will not receive any information about the knowledgeability of the advisor. All of the subjects will be randomly assigned to either learn their part 1 performance or not learn their part 1 performance. All of the subjects will be randomly assigned to take the second test under a piece-rate scheme or a tournament scheme. Their choices (math or verbal) will be collected for both schemes. After subjects make their choices about which tests to take under what schemes, they will be randomly assigned to a scheme (piece-rate or tournament) and complete their choice of test (4 questions).
Advisees' bonus payment will depend on either (1) the number of correct answers that they have in the test they will take in the first part of their survey, or (2) the number of correct answers that they will have on the portion of the test that they will choose to take (math or verbal) and the payment scheme they are in (piece-rate or tournament). Advisees will also get paid if they correctly recall the characteristics of their advisor and the information provided to them about what the advisor knows and doesn't know.