Prof. Dr. Larry AdamsAcademic, Author & Researcher
How to Conduct Quantitative Research · Chapter 14 of 33

Part II — DESIGN, SAMPLING, ETHICS AND DATA

Chapter 10. Ethics, Data Protection and Research Quality

10.1 Ethics before data collection

Ethics approval must be obtained before recruitment or data collection where required. Retrospective approval is generally not a valid remedy for collecting data without required approval. Research involving employees, students, patients, vulnerable groups or identifiable organisational records may require special safeguards.

10.2 Consent

Informed consent should explain the study purpose, procedures, duration, foreseeable risks, potential benefits, voluntary nature, withdrawal conditions, confidentiality limits, data use and retention, contact details, and complaint route. Consent is a process, not merely a tick box. The researcher should not promise absolute anonymity if indirect identifiers or platform metadata may make re-identification possible.

10.3 Confidentiality, anonymity and data protection

Anonymous data cannot reasonably be linked to an individual by the research team; confidential data may be identifiable but protected. Use the correct term. Collect the minimum data needed, restrict access, encrypt files, store consent records separately, set retention and deletion rules, and follow the applicable law and university policy. Cross-border cloud storage and international collaborators require attention to data-transfer rules.

10.4 Research integrity

  • Do not fabricate participants, observations, analyses or references.
  • Do not falsify or selectively alter data to obtain a preferred result.
  • Do not omit outcomes or analyses merely because they are non-significant.
  • Disclose deviations from the approved plan and distinguish confirmatory from exploratory analyses.
  • Keep an audit trail of data cleaning, coding and analytic decisions.
  • Report funding, conflicts of interest, authorship contributions and relevant permissions.
  • Use plagiarism-detection tools as one aid, not as a substitute for proper citation and original synthesis.

10.5 Quality dimensions

Reliability concerns consistency of measurement. Validity concerns whether the evidence supports the intended interpretation. Internal validity concerns whether a causal interpretation is credible in the design. External validity concerns transfer or generalisation to other populations and settings. Statistical conclusion validity concerns whether statistical inferences are well supported. These concepts overlap but are not interchangeable.

10.6 Preregistration and analysis plans

Where feasible, record hypotheses, primary outcomes, sample-size rationale, exclusion rules, data collection stopping rules and planned analyses before examining outcome data. Preregistration is particularly valuable for confirmatory work and experiments. It does not prevent legitimate exploratory analysis; it makes the distinction transparent.