A new resource for behavioural science - Developing tools for understanding the relationship between behaviours

早睡和睡眠时间较长的人白天锻炼得更多吗？驾驶行为是否与人们回收利用或帮助他人有关？如果有人随身携带可重复使用的咖啡杯怎么办？它们是否更有可能在家中保护生物多样性(例如，放置一个鸟箱)？行为之间的一些关联是直观的(例如，身体活动较多的人可能睡得更长，要么是因为他们认为这两者都是“健康的”行为，要么是因为运动的人需要更多的休息)，而另一些关联则不是那么直观(例如，驾驶行为与保护生物多样性之间的关系)。然而，心理学家(和其他行为科学家)经常孤立地看待行为--例如，寻求改善睡眠或增加体力活动水平。有时，这类研究会考虑一种行为的改变在多大程度上“溢出”到另一种行为的改变(例如，增加体力活动导致人们也做出更健康的食物选择)或导致补偿(例如，增加体力活动导致人们消耗更多卡路里)。然而，这些分析通常仅限于少数行为，通常在同一领域内(例如健康或环境行为)。还有关于如何定义和实施行为的问题(例如，什么构成体力活动的增加？)，指出需要商定定义和/或允许在研究之间进行比较的框架。简而言之，鉴于日常生活的特点是各种各样的行为，理解行为之间的关系是至关重要的，无论是在领域内还是跨领域，这都是发展我们对行为的理解和为干预提供信息的关键。幸运的是，理解行为之间的关系所需的大量证据已经存在。任何衡量两种或两种以上行为并报告它们之间的相关性的研究，或者提供访问允许计算相关性的数据的任何研究，都可以提供对它们之间关系的估计，这种估计可以跨数据集合并。然而，到目前为止，审查缺乏定义行为的框架，往往依赖于相对复杂的方式来看待行为之间的关系(例如，聚类法和网络分析)，这可能使研究结果难以解释，并且一直局限于考虑领域内行为之间的关系(例如，健康)，从而难以理解例如，人们是否以及如何在领域之间进行权衡。我们的建议是开发工具，允许行为科学家通过创建结构化模型来定义行为及其相似和不同之处(例如，饮酒和吸食可卡因都是药物使用的例子，但只有使用可卡因是非法的)。然后，我们将开始整理关于行为之间关系的数据(例如，来自已发表的论文、大型次级数据集)，并开发一套工具--称为“协作工作台”--使研究人员能够输入自己的信息，从而轻松、快速和高效地生成新知识。最后，我们将开发可视化数据的方法，并允许用户(如学者、政策制定者、利益相关者)向社区提出问题，并查询知识库，以对行为如何相关的问题提供强有力的答案。

Do people who go to bed earlier and sleep for longer exercise more during the day? Is driving behaviour associated with whether people recycle or help other people? What if someone carries a reusable coffee cup? Are they more likely to conserve biodiversity at home (e.g. put up a bird box)? Some associations between behaviours are intuitive (e.g. people who are more physically active may sleep longer, either because they view both as 'healthy' behaviours or because active people need more rest), while others are less intuitive (e.g. relations between driving behaviour and efforts to conserve biodiversity). However, psychologists (and other behavioural scientists) often view behaviours in isolation - seeking to, for example, improve sleep or increase levels of physical activity. Sometimes these sorts of studies will consider the extent to which changes in one behaviour 'spillover' into changes in another (e.g. increasing physical activity leads people to also make more healthy food choices) or lead to compensation (e.g. increasing physical activity leads people to consume more calories). However, these analyses are typically only limited to a small number of behaviours, usually within the same domain (e.g. health or environmental behaviours). There are also questions about how behaviours are defined and operationalised (e.g. what constitutes an increase in physical activity?), pointing to the need for agreed definitions and/or a framework that permit comparisons between studies. In short, given that everyday life is characterized by a wide range of behaviours, it is crucial to understand how behaviours are related to one another, both within and across domains, both to develop our understanding of behaviour and to inform interventions.Fortunately, a lot of evidence needed to understand the relationships between behaviours already exists. Any study that measures two or more behaviours and reports the correlation between them, or that provides access to data that allows the correlation to be calculated, can provide an estimate of their relationship, which can be pooled across datasets. However, reviews to date lack a framework for defining behaviours and have tended to rely on relatively complex ways of looking at the relations between behaviours (e.g. cluster and network analysis), which can make the findings difficult to interpret and have been limited to considering the relations between behaviours within domains (e.g. health), making it difficult to understand whether and how, for example, people make tradeoffs between domains. Our proposal is to develop tools that will allow behavioural scientists to define behaviours, along with their similarities and differences, by creating a structured model (e.g. that drinking alcohol and taking cocaine are both examples of substance use, but only using cocaine is illegal). We will then start to collate data on the relationship between behaviours (e.g. from published papers, large secondary datasets) and develop a set of tools - termed a "collaborative workbench" - that will allow researchers to enter their own information to enable easy, rapid, and efficient generation of new knowledge. Finally, we will develop ways to visualize the data and allow users (e.g. academics, policy makers, stakeholders) to pose questions to the community and to query the knowledge base to provide robust answers to questions about how behaviours are related.