The authors used program documentation and research literature to develop logic models. They used a  four-step process.

The four steps were:

1) The researchers identified programs for each intervention type. They started with the list of programs that met ESSA evidence standards according to this report.  They supplemented these programs with others highlighted in reviews of promising leadership programs. The authors also included those recommended by a panel of 12 experts. These experts included six RAND researchers and six external experts in leadership. Each intervention needed to have at least  one peer-reviewed study with positive outcomes and/or public documentation of evidence-based theory. The authors aimed to identify a sufficient number of programs to capture the range of approaches in an area. 

2) They collected or developed program-specific logic models. They searched public documentation to describe programs. This included research studies, program websites, user manuals, and other public documentation available online. Programs without sufficient information to develop logic models were excluded. Total counts of logic models included:

• 29 for principal preparation
• 11 for professional learning
• 8 each for leader evaluation, working conditions, and school improvement
• 3 for strategic staff management.

3) The authors aggregated program-specific logic models. They identified the common features across logic models. They then gathered additional information on implementation, outputs, and outcome measures to add detail to the logic models.

4) They gathered expert feedback to refine the logic models. Internal leadership experts provided feedback on several drafts of each logic model. Most of the information in the document comes from analysis of program data. One exception is that RAND largely generated the sample output measures and resource questions.