Return to the research questions
Decide which variables or qualitative material address each question. An analysis plan should explain what you want to learn, what evidence is available, and how each planned analysis contributes to an answer.
Understand how the data were generated
Record the sampling approach, collection tools, coding conventions, and units of measurement. Analysis cannot repair every limitation in the design or data collection process.
Document cleaning decisions
Check identifiers, missing values, coding consistency, plausible ranges, and duplicate records. Retain a protected original and record the changes made to the working dataset. Avoid silently removing inconvenient observations.
Match methods to the data
Choose methods based on the study design, type of data, research question, and relevant assumptions. For qualitative work, describe how material was coded and how interpretations were developed.
Interpret with care
Explain the size and meaning of a finding, the uncertainty around it, and the study limitations. A statistical relationship does not by itself establish a causal explanation. Include enough documentation for another person to understand the analytical choices.