Page 329 - CyberSecurityEssentialsEnglish

TABLE 7-23: DISTINGUISHING LEGITIMATE USE FROM FABRICATION

The case Why

Permitted if its source and purpose are described
Synthetic data used to test code
and it is not presented as real evidence.

Generated code that was Permitted with documentation and verification of the

reviewed and tested assumptions.

Interviews translated with a Requires disclosure, retention of the original, and
tool review of meaning.

Creating values or participants
Not acceptable; it alters the research record.
to fill gaps

Deleting data points with no
Not acceptable; it conceals an unwanted result.
recorded criterion

Generating quotations or Not acceptable; it manufactures support that does
findings not exist.

✕ Common misconception Correction

Test the output against a small known
Treating code that runs without error as example, check the assumptions, and read

code that is correct. what each step actually does before relying
on the result.

⚠ Warning: never let a tool modify the original data file
Use a copy for processing, keep a log of the transformations, and write new outputs that
can be compared against the original.

▣ Practical example: analyzing a satisfaction survey

Context: analyzing a satisfaction survey.
Task: clean the data and compute descriptive indicators.
Approach: the researcher writes the data dictionary, requests documented code, tests it
on ten cases, then compares the totals manually.
Why this design: it raises speed while the rules stay auditable.
Verification: a log of excluded values, a description of the missing data, and an

independent review of the figure and the table.

328