IR typically understands levels as levels of analysis that produce analytic/reductionist (rather than systemic/relational) explanations. Causes, separated by levels, are looked at as independent variables understood as distinct sources of explanation. Systemic explanations rely instead on related elements and levels of organization that are (understood to be) in the world (not just convenient epistemic devices). Systems approaches claim that parts on one level are organized into higher-level wholes that are themselves structured parts of still-higher-level wholes. (For example, subatomic particles, atoms, elements, chemical compounds.) The chapter concludes by examining the implications of a levels of organization framing for four important metatheoretical issues: micro–macro relations, the agent–structure problem, the natures of individual human beings and social groups, and the natures of individual and group identities.

We discuss the single embedded case study design in this chapter. We deliberate how this design is different from multiple and single holistic designs in terms of the levels of analysis and the nature of replication. The selection rationale and sampling are discussed next. Afterwards, we move on to the longitudinal and/or cross-sectional single embedded designs. The strengths and the weaknesses of the design in terms of internal validity, external validity, and the number of variables are discussed subsequently. This chapter also discusses the (mis)conception regarding longitudinal designs and temporal embedded units.

Here we discuss the multiple embedded case study design and the levels and cases involved in it. The design’s capacity to undertake between- and within-case comparative analyses are deliberated upon afterwards. Multiple embedded designs offer a variety of design possibilities, which we discuss in this chapter. The potential of the design for high internal validity is considered afterwards. Lastly, the design’s weaknesses, including the lack of contextual depth, the smaller number of explanatory variables, and the access difficulties to the cases/embedded units, are detailed.

In this chapter, we compare Routine Dynamics and Strategy as Practice based on an extended literature review. Routine Dynamics and Strategy as Practice are distinctive communities of thought in organization studies that exhibit a number of striking parallels: both subscribe to the overall “practice turn” in the social sciences, seek to bring the human being back in, and focus predominantly on the level of action involved in organizational routines and strategy, respectively. In our comparison of similarities and differences of Routine Dynamics and Strategy as Practice, we focus on their empirical domains, underlying theoretical perspectives, research frameworks, levels of analysis, and empirical methods employed. Based on that, we discuss what Routine Dynamics can learn from Strategy as Practice and vice versa. We conclude with some general reflections on the future relation between the two research communities and develop an agenda for future research that facilitates cross-fertilizations between the two research communities.

Identifying a good research question is a vital first step in any behavioural study because the question will focus the rest of the research cycle. Four logically distinct types of question can be asked about any behaviour. These concern its mechanisms, its development (or ontogeny), its function and its evolution (or phylogeny). The mechanisms underlying behaviour can be studied at many different levels, ranging from the social or physical environmental conditions that influence the behaviour down to the neural networks responsible for behavioural output. The nature of the research question will influence decisions about what species to study. Research questions are developed through a combination of approaches, including reading the literature, preliminary observations and exploratory data analysis. A research question leads to a set of hypotheses that need not be mutually exclusive but should all be testable. Each hypothesis should generate one or more specific predictions.