Now, one challenge with this approach is that many real-world examples are too entangled with existing political biases, media narratives, and emotional reactions. We then pored over existing definitions of exploitation and identified common properties that different thinkers had identified as components of exploitation. This modular design allowed us to isolate the exact impact, or “treatment effect,” of each distinct property on participants’ evaluations. For each vignette, participants answered the same question: “Based only on the information provided, how exploitative is this scenario in your opinion?” Importantly, the results are much the same when these fixed effects are excluded or observable characteristics are included instead.