https://doi.org/10.4087/DEBS7207">
 

Papers from the International Association for Cross-Cultural Psychology Conferences

Publication Date

11-1-2025

Abstract

It is a challenging task to develop a scale to measure psychological constructs in a non-native language. In this study, we compared reliability and validity of a scale to measure five dimensions of Personal Beauty Values (PBV) generated by ChatGPT4 with a human-generated PBV scale. The original human-generated PBV scale in English consists of 14 items divided into five dimensions. The limitation of this scale developed by non-native English speakers is a restricted number of items, which will potentially limit the applicability of the scale to other cultural contexts. To complement this weakness, we generated additional items of the five PBV dimensions using ChatGPT4. Specifically, we asked ChatGPT4 to generate 5 additional English items for each dimension based on the existing human-generated items. Using the US sample (N = 340), we pruned the items into four items per dimension. Subsequently, we translated them into French and Japanese, collected data in France and Japan (N = 346 and N = 300, respectively) and compared reliability and validity of the ChatGPT4-generated scale with the human-generated scale in each country. The results of our study demonstrated the replicability of the five-dimensional PBV structure by using the ChatGPT4-generated scale in the US context. However, when these ChatGPT4-generated items were translated into French and Japanese, it was not possible to extract the same dimensional structure due to the lack of discriminant validity.

Download
Globalization in Context

Included in

Psychology Commons

Share

COinS