The stages of a project, in order, with the links you’ll need at each one.
Before we first meet, or within your first week or two, work through the onboarding guide and get set up.
Start by picking a research area that interests you. We can then brainstorm topics and research questions together in our one-on-one meetings. Honours students can find a set of paradigms and starter references on the lab’s Honours page.
Both PhD and Honours students write a research plan. The length depends on your degree, but keep it succinct and follow the writing guide, which lists the sections a research plan needs. The plan is not marked, and you don’t need to respond to the feedback. It is simply a chance to get suggestions from another academic early in your program. Honours students should also read the proposal tips and meeting timeline on the Honours page.
We build and run experiments on Second Order, the lab’s experiment platform. There is no website to log in to and no code to write by hand. You work entirely in a chat with Claude or ChatGPT: you describe the study, and the AI builds it in jsPsych, tests it and hosts it for you. The steps below use Claude, but ChatGPT works the same way. If you haven’t connected Second Order yet, follow the steps in the onboarding guide first.
Claude writes the code, but you are responsible for piloting it thoroughly. Ask Claude to “show me the dashboard”. The dashboard lets you run the whole study, or any single part of it, and shows the data at the end without saving anything. Click through every part from the participant’s point of view, and check that everything you need for your analysis is being recorded. To fix bugs or make changes, simply describe them to Claude.
When you’re ready to collect data, ask Claude to “check the study is ready to launch” and work through anything it flags. Then ask for a participant link for Prolific or SONA. For Prolific, create the study there first so you can give Claude the completion link. Because the link always runs the live experiment, any change you make once data collection has started reaches participants immediately. Ask Claude to save a version before making a big change.
We collect data through two platforms, Prolific and SONA. When you set up your study, use the exact advertisement wording from the umbrella ethics application you are using. A video showing how to put a study on SONA is available on the recruitment page. Before collecting data, consider preregistering your study on AsPredicted.
Start with data from 10 participants. Add that data to your Claude project and ask Claude to check it, run some preliminary analyses, and confirm everything is working. You should also check the data yourself, including reading participants’ answers to any debug questions.
Because your data are on Second Order, Claude can read them directly and run the analysis for you in a chat in your project. Ask it to clean the data, apply your exclusions and run your planned analyses.
Treat Claude’s results as a first pass, not the final answer. You must reproduce every analysis yourself in R (or similar), check that you get the same numbers, and work out why if you don’t. Claude can help you write the R code. To get the data for R, ask Claude for a download link. By the end, you will need a clean data file and an R analysis script to publish alongside your paper or thesis.
Use the lab writing guide to write up your project.