FRENAR
methodology

How the thesis is conducted

This page makes the research design visible without publishing its data. It describes what is being studied and how, the corpus, the experimental trial, the interviews, and the way the materials will be opened after the defence.

Piece
Open dossier
épistémiquemethodologyMay · 2026
Overall design

The thesis adopts a mixed-methods approach. Three components are articulated together: a forensic corpus analysis to characterise the threat, a randomised controlled trial to test a response, and a series of expert interviews to situate both within the institutional reality. None of these strands suffices on its own; their triangulation is the central methodological argument.

A

Forensic corpus analysis

Volume
21 documented campaigns, totalling 52 analytical units.
Period
2019, 2025.
Inclusion sources
Viginum, the EEAS (FIMI reports), the Microsoft Threat Analysis Center, EU DisinfoLab, the DFRLab and the specialised press. Only events that are the subject of public documentation are retained.
Coding
Each event is coded with the DISARM framework, an open taxonomy of information-manipulation tactics and techniques.

The exhaustive list of the 52 analytical units and the detail of the coding are not published here; they appear in the manuscript and will be opened after the defence.

B

Randomised controlled trial (RCT)

Design
A between-subjects trial, with randomisation between a treatment condition (prebunking) and a control condition.
Target audience
Adults, recruited online; participation is voluntary and anonymous.
Duration
A short session, of around five minutes.
Instrument
Primary measure: the per-stimulus credibility ratings. The FBSS (FIMI Belief Susceptibility Scale), a first-generation instrument, is its associated susceptibility scale.
Sample
502 valid completers (253 control, 249 inoculated); the pre-registered target was reached.
Pre-registration
Hypotheses and analysis plan pre-registered on Zenodo (DOI 10.5281/zenodo.20403018)
Ethics
The protocol was approved by a research ethics committee before recruitment opened. The file reference is provided on request to peers and institutions.

Data collection is complete. On the primary endpoint, the mean credibility of the five FIMI stimuli, the inoculated arm scored slightly lower than the control arm, a small and statistically non-significant difference (d ≈ −0.10, p = .29); sharing intention did not differ between the arms. The thesis reads this null primary result as support for structural and environmental measures rather than reliance on individual inoculation alone. Full effect sizes, statistical tests and per-stimulus analyses are reported in the manuscript (Chapter 5).

C

Expert interviews

Volume
One completed and interviewee-validated interview, with further interviews scheduled.
Profiles
Five milieus: intelligence and threat monitoring, platform governance, media regulation, specialist journalism, civil society.
Anonymisation
Interviewees are cited by role and milieu, with no identifying element or detail liable to re-identify them. Verbatim quotes are checked to that end before any use.

The content of the interviews and the conclusions drawn from them belong to the manuscript.

Opening of the data

An open-data policy after the defence

The choice to reserve the data before the defence is not a choice of closure, but of sequencing. Once the defence has taken place, the thesis plans:

  • the deposit of the shareable materials in a permanent archive (Zenodo or equivalent), with a persistent identifier;
  • the opening of the DISARM-coded forensic corpus, in a documented and reusable version;
  • the release of the interviews in anonymised form, in keeping with the commitments made to the interviewees.

The experimental component is presented in detail, with its ethical framework and the invitation to take part, on the Experimental study page.