Finding Love in the Black Box: Algorithm Awareness on Dating Apps - Data Files

The dataset included here belongs to the project ‘Finding Love in the Black Box: Algorithm Awareness on Dating Apps’ by Bukman, Sharabi and Timmermans (2024). The aim was to assess the level of algorithm awareness among Dutch and Belgian users of the dating apps Breeze (Dutch) and Tinder, and test whether it is associated with high algorithm trust and satisfaction with matches.<b>BACKGROUND </b>Most dating apps incorporate artificial intelligence (AI) to match users to one another but are not fully transparent about how their recommender systems work. In turn, users have varying degrees of awareness of matching algorithms. Unfortunately, little is known about how insight into algorithmic functioning affects the online dating experience. This research examined the relationship between algorithm awareness, trust in the recommender system, and satisfaction with the quality of matches.<b>METHOD </b>A survey was conducted among 216 users of U.S.-based Tinder and 500 users of the Dutch dating app Breeze. Relationships were estimated with structural equation modeling, and effect sizes were compared between these two groups of users to assess variability based on dating app affordances. Higher algorithm awareness was associated with more trust in the recommender system, which in turn positively related to the ease by which users could find matches. The dataset is a composite of two separate collections. The data among Breeze users was collected in collaboration with Breeze itself. Users received a popup in the app with the link to the survey. 688 people participated, which resulted in 500 complete responses in two days. The data for Tinder users was collected in a period of several weeks on various unrelated social media pages. After 500 respondents started the survey, 216 complete responses remained. All participants had to be over 18 years old, and had to have used the app in the previous 30 days.<b>RESULT </b>The presence of this fully mediated effect between awareness and satisfaction with matches suggested that knowledge of matching algorithms can affect both the perception of the recommender system as well as the results of using dating apps. In other words, users with higher algorithm awareness are more content with the quality of the profiles they match with, mediated by a higher trust in the abilities of matching algorithms. No significant differences were found between groups.<b>CONCLUSION </b>Overall, the results indicate that increasing the awareness that users have of algorithm functioning could improve the online dating experience. This research contributes to the limited literature on human-AI interaction in the context of dating apps.

本数据集隶属于Bukman、Sharabi与Timmermans于2024年发布的研究项目《黑箱中的寻爱：约会应用中的算法认知》（*Finding Love in the Black Box: Algorithm Awareness on Dating Apps*）。本研究旨在评估荷兰及比利时用户对约会应用Breeze（荷兰本土专属）与Tinder的算法认知水平，并检验该认知水平是否与算法信任度及匹配满意度存在关联。

一、研究背景
多数约会应用均引入人工智能（Artificial Intelligence, AI）以实现用户间的匹配，但并未完全公开其推荐系统的运作逻辑。由此，不同用户对匹配算法的认知程度存在差异。遗憾的是，目前学界对算法运作认知如何影响在线约会体验的相关研究仍较为匮乏。本研究旨在探究算法认知、推荐系统信任度及匹配质量满意度三者间的关联关系。

二、研究方法
本研究通过问卷调查收集数据，受访对象分为两类：美国版Tinder用户216名，荷兰约会应用Breeze用户500名。研究采用结构方程模型估算变量间的关系，并对比两组用户的效应量，以评估不同约会应用特性带来的群体差异。
分析结果显示，算法认知水平越高，用户对推荐系统的信任度也越高，而信任度又正向影响用户的匹配便捷性。
本数据集为两份独立数据集的整合。其中Breeze用户的数据由Breeze官方合作收集：用户在应用内收到指向问卷的弹窗提示，共计688人参与，两天内回收500份有效问卷。Tinder用户的数据则通过数周时间在多个无关社交媒体页面征集完成：共500人启动问卷，最终回收216份有效问卷。所有参与者均需年满18岁，且在过去30天内使用过对应约会应用。

三、研究结果
算法认知与匹配满意度间存在完全中介效应，这表明对匹配算法的了解既会影响用户对推荐系统的感知，也会影响约会应用的使用效果。换言之，算法认知水平更高的用户，其对匹配对象的质量满意度也更高，这一关系通过提升对匹配算法能力的信任度实现中介传导。两组用户间未发现显著差异。

四、研究结论
总体而言，本研究结果显示，提升用户对算法运作机制的认知水平，可有效改善在线约会体验。本研究为约会应用场景下的人机交互相关有限学术文献库提供了新的补充。