Suppr超能文献

“机器人”如雨下:网络调查获取途径的便捷化如何引发了一场完美风暴。

It's raining bots: how easier access to internet surveys has created the perfect storm.

作者信息

Caven Isabelle, Yang Zhenxiao, Okrainec Karen

机构信息

University Health Network, Toronto, Ontario, Canada.

University Health Network, Toronto, Ontario, Canada

出版信息

BMJ Open Qual. 2025 Jun 1;14(2):e003208. doi: 10.1136/bmjoq-2024-003208.

Abstract

Online surveys are an increasingly common way to collect data from the public, with social media and financial incentives (e.g. gift cards) commonly used to increase participation rates. Anonymity, ease of response, and the potential to reach diverse demographics have also contributed to the popularity of online surveys. Health services research benefits from the increased accessibility that online survey-based data collection provides; however, fraudulent responses are of concern. The following article describes our team's experience with a national survey of Canadian healthcare providers being overrun with fraudulent responses and approach to ensure the validity of our survey data. We provide recommendations for research teams on how best to design their surveys, work with their institutions to implement safeguards within survey platforms, and screen completed responses. We also describe fraudulent open-text responses that we believe to have been produced with the help of artificial intelligence and are sounding the alarm for other researchers to be aware of this potential threat to data integrity. Informed by the learnings shared, researchers and research institutions can be better equipped to prevent and screen fraudulent responses to continue successfully engage the public in online research.

摘要

在线调查是一种越来越常见的从公众那里收集数据的方式,社交媒体和经济激励措施(如礼品卡)常被用来提高参与率。匿名性、易于回复以及能够接触到不同人口群体的可能性也促使在线调查受到欢迎。基于在线调查的数据收集所带来的更高可及性使卫生服务研究受益;然而,虚假回复令人担忧。以下文章描述了我们团队在一项针对加拿大医疗服务提供者的全国性调查中遭遇大量虚假回复的经历,以及确保我们调查数据有效性的方法。我们为研究团队提供建议,说明如何最好地设计他们的调查、与所在机构合作在调查平台内实施保障措施以及筛选已完成的回复。我们还描述了我们认为是在人工智能帮助下生成的虚假开放式文本回复,并向其他研究人员敲响警钟,让他们意识到这种对数据完整性的潜在威胁。基于所分享的经验教训,研究人员和研究机构能够更好地做好准备,预防和筛选虚假回复,从而继续成功地让公众参与在线研究。

https://cdn.ncbi.nlm.nih.gov/pmc/blobs/7c77/12142081/47f92760a3d2/bmjoq-14-2-g001.jpg

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验