Page 348 - CyberSecurityEssentialsEnglish

does not come from technology alone; it needs documentation of decisions,
assumptions, and changes.

● Remember: documentation does not slow research down

Good documentation reduces rework, eases collaboration, and protects you when the
tool changes, a member leaves, or a reviewer asks you to explain a step.

● Remember: Section summary
• Key idea: Reproducibility needs traceable files, versions, decisions, and AI uses.

• Practical decision: Create a log, an organized folder, a protocol, and a data dictionary
from the outset.
• Mistake to avoid: Leaving important steps inside conversations or in manual edits with
no record.

7.13 Practical Lab: From a Research Question to a Documented Evidence Map
◆ Key idea: an evidence map is an auditable package.
The output of this lab is not a complete article, but a small auditable package: a clear
question, a search strategy, screening criteria, an evidence matrix, a short synthesis, a
reference log, and an AI usage log. If that structure is sound, the writing that follows
becomes faster and more honest.

Choose a realistic, bounded question that can be searched in a short time. It might concern
the quality of a service, the effect of a teaching approach, the adoption of a technology, a

user experience, or a policy inside an organization. Do not begin with a question that needs
hundreds of studies or unavailable data. The aim is to practice the whole chain, not to
produce a final review.

Write the question brief: the problem, the audience or population, the decision or

knowledge required, the core definitions, and what you will not cover. Then use AI to
generate synonyms and alternative readings, but adopt the list only after reviewing it against
the sources of your field.

Build a search strategy across at least two sources suited to the question, perhaps one

academic database and one official site or index. Record every query and its date and the
number of results. Choose a manageable number of records for the exercise, such as twenty
to forty, and do not let the tool select for you whatever agrees with your expected result.

Set three to five inclusion and exclusion criteria. Screen the titles and abstracts, then read
the selected full texts. Record the reason for every exclusion. If the tool helped with

screening, test a sample yourself, and record the cases of disagreement and what you
learned from them.

347