- Subconscious Insights
- Posts
- How Subconscious.ai Achieves 93% Accuracy in Replicating Human Studies: Real-World Examples
How Subconscious.ai Achieves 93% Accuracy in Replicating Human Studies: Real-World Examples

As a causal research AI company, some of the most common questions we receive about our platform surround accuracy and reliability of the data.
Our AI has successfully matched human responses in 350 studies at ~93% equivalence
Our Spearman's rank-order statistics show that our synthetic respondents deliver reliable, human-like insights across various fields, including consumer choice, public health, and environmental economics.
But, what does this really mean and how do we measure it? When we are able to validate that synthetic respondents can accurately reproduce results in parity with what you’d receive from conducting studies with humans, we can determine the reliability of the platform’s insights. And when we are highly accurate, we can then collapse the time and money it takes to conduct studies like this, without compromising the quality of insights we receive.
Here are three examples that showcase how Subconscious.ai can accurately produce parity results with human research studies:
(Spearman Correlation=0.79) matching the Sonja Lüthi and Thomas Prässler 2011 paper on
Wind Energy Policy. Interact with our data.
The study concludes that developers highly value risk mitigation in wind energy projects, with a strong preference for legal security and efficient administrative and grid access processes. It also highlights developers' non-compensatory approach to decision-making in the face of unfavorable policy attributes.
Bechtel, Scheve, and van Lieshout study on International Carbon Tax Policy for Environmental Mitigation, with a Spearman Correlation of 0.6711. Interact with our data.
The study assesses the preferences of residents in the U.S., U.K., Germany, and France for various aspects of carbon tax packages using a discrete choice experiment. Our replication confirms key insights, such as the importance of mitigation efforts, cost implications, and regional preferences.
Hainmueller and Hopkins 2014 Immigration Policy study with a Spearman Correlation of 0.5406. Interact with our data.
The study examines U.S. citizens' immigration preferences using conjoint analysis, revealing key attributes influencing public opinion, such as education, language skills, and job experience. These insights are crucial for more informed policy design.
If you’d like to try out the Subconscious AI platform for yourself, you can run 2 free experiments by clicking “Get Started”. Or, you can schedule a demo to learn more about how you can use Subconscious AI for your market research needs.