A Naïve Bayes classifier distinguishes free-text definitions of curiosity and interest at above-chance accuracy on unseen data¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r2 · q3
Currency · newest study 2021 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Cross-validated classifier accuracy exceeded chance for both the main test data (79.29%) and additional test data (77.06%). → Donnellan 2021 q3 i? Feature selection identified 42 word stems that accurately distinguished curiosity and interest definitions in training data (87.13%). → Donnellan 2021 (2)

Evidence¶

Donnellan 2021¶

Donnellan, E., Aslan, S., Fastrich, G. M., & Murayama, K. (2021). How Are Curiosity and Interest Different? Naïve Bayes Classification of People's Beliefs. Educational Psychology Review. https://doi.org/10.1007/s10648-021-09622-9

q3 · i? · associational · r2

Pre-registered cross-validation study: a Naïve Bayes classifier trained on free-text definitions was tested on unseen main and additional test sets, with binomial tests against 50% chance. The article reports accuracy "significantly above chance" for both test sets; no effect size is printed.

"Importantly, the classifier was also highly accurate in predicting unseen data (Table 1), accurately predicting the correct definitio n type significantly above chance for both the main test data (79.29%; 95% CI: 72.98 –84.71%, p < .001) and additional test data (77.06%; 95% CI: 70.91 – 82.47%, p < .001)."

Donnellan 2021 (2)¶

Donnellan, E., Aslan, S., Fastrich, G. M., & Murayama, K. (2021). How Are Curiosity and Interest Different? Naïve Bayes Classification of People's Beliefs. Educational Psychology Review. https://doi.org/10.1007/s10648-021-09622-9

q3 · i? · associational · r2

Classifier training on the main dataset (Prolific Academic participants producing two free-text definitions each) identified 42 discriminative word stems, "accurately classifying 87.13% of definitions" in the training data.

"In training, feature selection identified 42 word stems which accurately distinguished curiosity and interest definitions in the training data (accurately classifying 87.13% of definitions)."

Discussion¶

Related Claims¶

• The non-expert-trained classifier also distinguishes expert definitions of curiosity and interest above chance — related
• Frequently used word stems such as \"can\", \"want\", \"thing\" and \"feel\" carry no discriminative value between curiosity and interest definitions — related
• Education and age of English acquisition moderated classifier accuracy by definition type in the main test data — related
• The consensus distinction characterises curiosity as active seeking of specific unknown information and interest as pleasurable, in-depth seeking in already-known domains — related
• Participants' ratings of how similar curiosity and interest are did not predict classifier accuracy — related