Figure 1
Free association user interface of the Small World of Words study (https://smallworldofwords.org) showing the cue word ‘intelligence’ at the top. Below are three lines for respondents to enter free associations, with two already populated by the response words ‘knowledge’ and ‘insight.’
Figure 2
Overview of the working example with associatoR functions and class ‘associatoR’ object.
Figure 3
GPT-4-Turbo-generated free associations to the cue word ‘intelligence’. (A) A UMAP projection of the response-based embedding determined word positions. Word and point size show response frequency; colors show membership to clusters. (B) Word clouds for each cluster, with font size indicating word frequency among the responses.
Figure 4
Comparisons between demographic groups: (A) Comparison of cluster proportions between male and female (high school- and university-educated) simulated subjects. (B) Top 50 high-frequency words among the responses ordered by phi coefficient between male vs. female (high school-vs. university-educated) simulated subjects from more frequent among males (high school-educated) on the left to more frequent among females (university-educated) on the right.
