Dan Benjamin, Mark Fontana and I Design an In-Depth Risk Aversion Survey

As an addendum to the title, let me mention that Michael Gideon also helped immensely with the conceptual design of this survey. Michael Gideon and Mark Fontana are both Ph.D. students in Economics at the University of Michigan. Dan Benjamin is an Assistant Professor at Cornell and a frequent coauthor of mine. 

Attitudes toward risk matter greatly for many economic decisions. Bob Barsky, Tom Juster, Matthew Shapiro and I got into the business of measuring risk aversion directly from hypothetical choices a while back:

Matthew and I continued that line of research along with Claudia Sahm:

I have several ongoing projects about risk aversion with various coauthors (including Matthew and Claudia). The one that envisions the most in-depth analysis of the nature of risk aversion is the research project Dan Benjamin, Mark Fontana and I are working on. We have a draft web survey that is in pretty good shape that we are trying to debug. Feel free to take a peek if you are curious to see what we are up to. But if any of you are willing to take this draft survey as if you were a real respondent, you could really help us in making sure the computer code is working as it should and in checking whether our procedures are working OK. Your answers will be anonymous and we will not use them directly in the research, but only as a pretest for checking our procedures.

The instructions are below, after the break. They are written as if you plan to do the whole thing. But if you are just taking a peek, still enter “miles” when it asks for “netid”. We will be able to tell you are just taking a peek from the fact that you stop part way through. However, if you make it the whole way, your answers will be much more valuable in our debugging. I am grateful for your help and your interest.

Instructions:

Thanks for agreeing to help test our investment preferences web survey. Your feedback is much appreciated. This survey is going to help us learn more about risk aversion, and how you think about your own decision making.

First, visit the following website:

https://survey.rand.org/research/hosting/risk/index.php

After clicking next on the first screen, you will be asked for your “netid”. Please enter “miles” as the answer to this question.

Next, you will be taken to the “Table of Contents” page. This section is only for purposes of receiving feedback and pilotting. The full survey will be one continuous stream of sections. It will feature several training modules, a calibration section (so the gambles you’ll face are tailored to your own risk aversion), and then the main body of the experiment, followed by a demographic survey. However, for the purposes of your feedback, we’re only asking you to go through two of the training modules and the main body. In total, this should not take more than 50 minutes or so to complete.

Select each of the following modules from the Table of Contents in the order listed; when you’ve finished one module, close out the browser, restart it, and reload the survey (again entering your “netid” as “miles), and start the next module.

1. Symbols training/quiz
2. Assumptions training/quiz
3. Phase 1 and 2 (this one is the main body)

For the first two modules, we’re curious whether everything is clear and concise.

For the main body, we’re curious if everything loads correctly. We’re also curious, once you get to "Phase 2” (which will be signaled by a slide with a sunset and a warning that the next slides might take a while to load), whether it takes too long to load slides. Our own testing has seen the longest wait time of about 4 minutes.

As you will see, the purpose of Phase 2 is to ask you how you feel about updating certain choices you made during Phase 1. We’re curious whether you end up being asked to resolve any intransitivities in your decision-making during phase 2 (i.e. cyclical preferences of the form A > B > C > A). **These slides would ask you to explicitly rank choices among three or more options.** Our guess is that our current algorithm for randomly asking these slides won’t have them appear very often. In other words, other phase 2 slides will have the effect of resolving any intransitivities in your decision-making, so you might never be explicitly asked to do so.

Thanks again, and happy choosing.