Background: High-quality requirements are considered crucial for successful software development endeavors as the requirements' purpose is to inform subsequent activities like implementation or testing. Requirements quality defects have been shown to incur significant costs for remediation, scaling up even to project failure. At the same time, the effort to improve the quality of requirements must be justified. Organizations developing software, therefore, need to understand when their requirements artifacts are of "good enough'' quality, i.e., they need to be able to identify the optimum between over- and under-engineering.

Problem: The body of knowledge in requirements quality does not yet offer solutions that would allow organizations to identify that optimum due to three shortcomings: (1) there is no generally accepted, theoretical foundation to describe requirements quality that can serve as a basis to coordinate distributed research efforts and the synthesis of evidence in the field, (2) the scientific practice currently applied in the field is of limited rigor to draw reliable conclusions from existing empirical contributions, and (3) the field lacks empirical evidence that can be aggregated to form a holistic view of requirements quality. These are potential causes for the lack of adoption of requirements quality research in practice.

Goal: In this cumulative, publication-based thesis, we address these three shortcomings and aim to contribute to a more evidence-based approach to requirements quality research grounded in scientific theory.

Method: First, we develop a theoretical foundation by adopting and integrating existing software engineering theories. Second, we evaluate the state of the art of data analysis and open science in the field and provide guidelines to improve these practices. Third, we demonstrate the application of these guidelines and conduct a controlled experiment to contribute additional empirical evidence to the field.

Results: The resulting set of analytical theories specifies requirements quality and provides a structure for future empirical contributions. Our evaluation of the state of the art shows both the need for a common theoretical foundation as well as support for applying rigorous research practices. Our empirical studies contribute to these needs and illustrate the complexity of the impact that requirements quality defects have on subsequent activities. Finally, we develop a method for the effective aggregation of empirical results.

Conclusion: Our theoretical, methodological, and empirical contributions help to coordinate a productive and constructive research agenda on requirements quality that is based on evidence and grounded in theory. This allows for rigorous and practically relevant research that ultimately informs organizations on how to engineer good-enough requirements.