The research team also wanted to see if survey responses translated to actual scientific output. They received permission from a portion of the participants to securely link their survey answers with their professional publication records.
The team utilized machine learning technology to analyze the text of the scientists’ published abstracts and article titles. The computer algorithms measured how closely the words and phrasing matched among different authors. They also built algorithms to map out who these scientists collaborated with and which older papers they cited as foundational literature.
The algorithms revealed that cognitive traits are associated with differences in real-world publishing activity. This remained true even when controlling for a researcher’s specific subfield and preferred tools. Two psychologists who study the exact same topic using identical methods are still more likely to cite the same reference materials if they happen to share similar internal thinking styles.
