A core aspiration of the ecological sciences is to determine how systems work, which implies the challenge of developing a causal understanding. Causal inference has long been approached from a statistical perspective, which can be limited and restrictive for a variety of reasons. Ecologists and other natural scientists have historically pursued mechanistic knowledge as an alternative approach to causal understanding, though without explicit reference to the requirements of causal statistics. In this paper, I describe the premises of an expanded paradigm for causal studies, the Integrative Causal Investigation Paradigm, that subsumes causal statistics and mechanistic investigation into a multi-evidence approach. This paradigm is distinct from the one articulated by causal statistics in that it (1) focuses its attention on the long-term goal of building causal knowledge across multiple studies and (2) recognizes the essential role of mechanistic investigations in establishing a causal understanding. The Integrative Paradigm, consequentially, proposes that there are multiple methodological routes to building causal knowledge and thus represents a pluralistic perspective. This paper begins by describing the crux of the problem faced by causal statistics. To understand this problem, it should be recognized that the word causal has multiple meanings and a variety of evidential standards. An expanded vocabulary is developed so as to reduce ambiguities and clarify critical issues. I further show by example that there is an important ingredient typically omitted from consideration in causal statistics, which is the known information related to the mechanisms underlying relationships being evaluated. To address this issue, I describe a procedure, Causal Knowledge Analysis, that involves an evaluation and compilation of existing evidence indicative of causal content and the features of mechanisms. Causal Knowledge Analysis is applied to three example situations to illustrate the process and its potential for contributing to the development of causal knowledge. The implications of adopting the proposed paradigm and associated procedures are discussed and include the potential for advancing ecology, the potential for clarifying causal methodology, and the potential for contributing to predictive forecasting.

中文翻译：

---

构建因果知识的综合范式


生态科学的一个核心愿望是确定系统如何运作，这意味着发展因果理解的挑战。长期以来，因果推理一直是从统计角度进行处理的，由于各种原因，统计角度可能是有限和限制的。生态学家和其他自然科学家历来追求机械知识作为因果理解的替代方法，尽管没有明确提及因果统计的要求。在本文中，我描述了因果研究的扩展范式的前提，即综合因果调查范式，它将因果统计和机理调查归入多证据方法。这种范式与因果统计所阐明的范式不同，因为它 （1） 将注意力集中在跨多项研究中构建因果知识的长期目标上，以及 （2） 认识到机械调查在建立因果理解中的重要作用。因此，综合范式提出有多种方法论途径来构建因果知识，因此代表了多元视角。本文首先描述了因果统计所面临的问题的关键。要理解这个问题，应该认识到因果关系这个词具有多种含义和各种证据标准。开发了扩展的词汇表，以减少歧义并澄清关键问题。我进一步通过示例表明，在因果统计中，通常会忽略一个重要的因素，即与正在评估的关系基础机制相关的已知信息。 为了解决这个问题，我描述了一个程序，即因果知识分析，它涉及对表明因果内容和机制特征的现有证据进行评估和汇编。因果知识分析应用于三个示例情况，以说明该过程及其促进因果知识发展的潜力。讨论了采用拟议范式和相关程序的意义，包括推进生态学的潜力、阐明因果方法的潜力以及促进预测的潜力。