Central research question
Under what conditions has context social, has message or has architecture of choice modifie-t-it the judgments and the conduites of has population, and when is-it justified of parler of manipulation rather that of influence, of persuasion, of education, of advertising, of disagreement or of simple coïncidence?
Secondary questions:
- The mechanism is-it describes before the interpretation morale?
- The public visé, the exposure and the context are-ils établis?
- L’effet is-it observé, mesuré and distingué of has corrélation?
- The message laisse-t-it has possibility réaliste of understand, refuser and revenir on its choice?
- The people concernées have-they been informées, protégées and entendues?
- Of competing explanations and of against-évidences have-they been recherchées?
Definitions and terminology
| Term | Definition of work | This that the term not prouve not |
|---|---|---|
| Influence | Modification or orientation of has judgment or of has conduite under the effet of has context, of autrui or of has message. | HAS intention malveillante. |
| Persuasion | Communication visant to modifier has belief, has attitude or has decision by of raisons, of narratives or of appels affectifs. | That the auditoire has been trompé or private of choice. |
| Education | Transmission organisée of knowledge, skills and methods, with has finalité of learning. | That the teaching is neutre in all context. |
| Advertising | Communication promotionnelle identifiable or liée to has offers, has marque, has cause or has organisation. | That any emotion or repetition is manipulatoire. |
| Rhetoric | Organisation of the discours for informer, émouvoir, do adhérer or agir. | HAS fausseté or has coercion. |
| Nudging | Modification of the architecture of the choice that rend has option more saillante or more facile without supprimer formellement the alternatives. | That the choice is libre au sens fort or that the effet is bénéfique. |
| Coercion | Pressure, menace or sanction that reduces substantiellement the capacity of refuser. | That any pressure social ordinaire is coercive. |
| Deception | Présentation fausse or omission intentionnelle of has élément pertinent visant to induire in erreur. | HAS effect durable or has coordination of campagne. |
| Propaganda | Communication organisée visant to orienter of perceptions or behaviours au service of has project; the notion is descriptive but historically loaded. | That the content is faux or that its public is passif. |
| Misinformation | Information fausse or trompeuse partagée without evidence suffisante of has intention of be mistaken. | The intention of the émetteur. |
| Disinformation | Information fausse or trompeuse diffusée with has intention of be mistaken or of nuire, intention to document. | That any erreur becomes disinformation. |
| Pressure social | Cost, attente or signal social that rend has reply more probable or more acceptable. | HAS absence totale of choice. |
| Conformité | Ajustement of has judgment or of has behaviour to has norm or to has majorité. | HAS perte of autonomy or has erreur. |
| Polarisation | Éloignement of positions, durcissement of groups or accentuation of differences; the mechanism must be précisé. | HAS only cause algorithmique. |
| Identity of group | Sentiment of belonging and categories that organise the perception of soi and of other. | HAS homogénéité psychological of the group. |
| Capture of the attention | Procédé or event that increases the probabilité that a stimulus soit remarqué, consulted or répété. | HAS belief or has action. |
| Arousal émotionnel | Level of activation émotionnelle, distinct of the valence positive or négative. | HAS intention of manipuler. |
| Repetition | Réapparition of has message or of has motif. | The truth of the message or has effect identique for all. |
| Framing | Selection and placing in before of some aspects of has sujet for orienter its interpretation. | The falsification of the sujet. |
| Priming / amorçage | Effect possible of has stimulus préalable on has reply ultérieure, mesuré in has context precise. | HAS effect général and durable in the life real. |
| Agenda-setting | Influence possible of the saillance of has sujet on this that the public juge important, without déterminer nécessairement this that il must penser. | HAS control complet of the opinion. |
| Control coercive | Set répété of behaviours of domination, of surveillance, of isolement or of menaces in has relationship or has group; term to use with basis factuelle and prudence legal. | HAS diagnosis clinical or has evidence automatique of culpabilité. |
HAS same practice can relever of several categories, but the qualification must rester proportionate. By exemple, has interface can be has nudging; if she masque the refus, increases artificiellement the cost of sortie or dissimule the objectif, the hypothesis of manipulation becomes more forte. It must alors document the parcours real, not only the apparence visuelle.
Psychology social: context and judgment individual
HAS decision depends of the person and of the situation. The context can fournir of norms, of attentes, has identity, of information or of sanctions. The person apporte its history, its objectifs, its knowledge and its contraintes. The psychology social seeks of régularités without effacer the diversity of individuals.
Trois safeguards are centrales. Premièrement, has mechanism means can have of effects hétérogènes: has same alerte can aider has person and distraire has other. Deuxièmement, the does that a person puisse expliquer its choice after coup is not sufficient to establish the cause real. Troisièmement, has association between exposure and behaviour can résulter of has auto-selection: the people already intéressées recherchent more has content.
HAS analysis must therefore noter: the situation initiale; the information disponibles; the power of refus; the intérêt of the émetteur; the resources of the public; the conséquences of has erreur; and the moment where the effet is mesuré. L’étiquette « biaisé » is trop vague if she not precise not the mechanism, the direction, the taille and the robustesse of the effet.
Conformité and norms sociales
The norms descriptives indiquent this that the other semblent do; the norms injonctives indiquent this that is approuvé or désapprouvé. HAS compteur, has commentaire, has file of attente or the behaviour of has collègue can devenir has signal social. The conformité can faciliter the coordination, the learning and the security. She can also reduce the expression of has doute when the cost of the disagreement is élevé.
HAS study of pressure social must distinguish the acceptation public of has opinion, the adhésion private and the behaviour observable. HAS person can itself conformer in public all in restant in disagreement. In has dossier, seek the rules explicit, the sanctions, the récompenses, the presence of has majorité real or apparente and the possibility of respond anonymement.
The nombre of participants visibles not is not always the nombre of people réellement convaincues. Of same, has consensus affiché can résulter of has tri of commentaires or of has sample sélectionné. The traces digital must therefore be accompagnées of informations on their mode of collecte.
Identity, belonging and polarisation
The identities sociales give of the sens, of the solidarity and of obligations. They can also accentuer the borders between « nous » and « eux », surtout when the status, the menace or the compétition are saillants. It must describe the practices and the discours observable without essentialiser has nationalité, has religion, has classe, has profession or has orientation political.
The polarisation can mean of positions more éloignées, has perception more négative of the other group, has déplacement of opinions towards the extrêmes or has ségrégation of networks. These indicators not are not équivalents. HAS community very homogène in its abonnements not is not nécessairement more extrême; has divergence of positions not is not nécessairement has hostilité.
For study the identity, code the auto-désignations, the borders of group, the norms of inclusion, the sanctions and the circulation of informations between groups. Protect the people: not not republier of pseudonymes, coordonnées or messages that permettraient of the retrouver, sauf need documented and basis légale.
Authority, expertise and trust
The authority can découler of has role, of has skill, of has réputation or of has capacity of sanction. L’expertise is has relationship between has person and has domaine; she not is not transférable automatiquement to all the sujets. The trust can be épistémique, institutional, affective or stratégique. HAS statement of expert can be pertinente without be infaillible.
HAS analysis must demander: the expertise is-she pertinente? The method is-she accessible? The interests are-ils déclarés? Existe-t-it has revue independent? The person distinguishes-t-she data, interpretation and recommandation? L’argument remains-t-it valable if the on retire the status of the orateur?
The appeal to the authority becomes préoccupant when the public not can not verify the source, when the dissentiment entraîne has sanction disproportionnée, or lorsqu’un role institutional is used for parler hors of its domaine without the report. Cela remains has hypothesis to document, not has conclusion on the caractère of the orateur.
Attention, emotion and repetition
L’attention is has ressource limitée. HAS title saillant, has alerte or has image choquante can increase the exposure. L’activation émotionnelle can faciliter the mémorisation or accélérer has decision; she can also reduce the temps consacré to the verification. The effects dépendent of the content, of the person, of the context and of the measure.
The repetition can make has document de travail more familière, but has repetition not garantit nor the belief nor the behaviour. For the study, distinguish nombre of impressions, nombre of consultations, durée of attention, rappel, belief declared and action. Verify if the people choisissent they-mêmes of revoir the content.
HAS accusation of « captation émotionnelle » must specify the levier: urgence artificielle, menace, récompense, honte, indignation, compassion or alternation. It must ensuite verify the intention documented, the targeting, the fréquence, the possibilités of pause and the effet comparative to has formulation neutre. L’émotion not is not in soi has faute: the témoignages, the art and the information of crise can légitimement émouvoir.
Framing, amorçage and placing to the agenda
The framing consists to select some aspects of has sujet and to the make more visibles: cost, justice, security, identity or efficacité. Two cadrages can be factuellement exacts all in orientant of priorités différentes. The analysis must indicate this that is included and this that is absent, without supposer that a omission is volontaire.
L’amorçage means has influence possible of has stimulus préalable on has reply ultérieure. The term is often used trop largement. It must specify the tâche, the delay, the réplication and the taille of the effet. The results of laboratoire not allow not of affirmer that a image or has mot control durablement the decisions quotidiennes.
The placing to the agenda concerns the saillance of sujets: the visibility of has thème can influencer this that the people jugent important. She not détermine not nécessairement their opinion. In has corpus media, code the volume, the position, the durée, the sources citées and the terms employés, in separating the disponibilité of has sujet and the attitude to its égard.
Persuasion, nudging and manipulation
The persuasion expose has message and seeks the adhésion. She can be transparente or discutable; she not is not automatiquement coercive. HAS nudging modifie the environment of decision. Its ethics depends of the transparency, of the facilité of sortie, of the proportionality, of the intérêt poursuivi and of the possibility of challenge.
L’hypothèse of manipulation is more plausible when several elements itself combinent: asymétrie of information; objectif caché; targeting of has vulnérabilité; fausse urgence; restriction practice of the refus; repetition no signalée; deception on the identity of the source; or sanction in cases of no-adhésion. These elements must be établis séparément.
HAS test minimal can take the form suivante:
| Élément to establish | Question | Trace attendue |
|---|---|---|
| Mechanism | Qu’est-this that is censé agir on the judgment? | message, interface, rule, interaction |
| Context | Where, when and with which contraintes? | chronologie, conditions of access |
| Target | Which public or which person, défini without stéréotype? | criteria of audience, sample |
| Exposure | That has effectivement vu, entendu or subi quoi? | logs agrégés, investigation, observation |
| Effect | Which changement observable and to which moment? | measure before/after, group comparateur |
| Alternatives | Which other causes are plausibles? | variables of control, against-evidence |
| Autonomy | The refus, the understanding and the withdrawal were-ils réalistes? | parcours, consentement, appeal |
Misinformation, disinformation and propaganda
HAS erreur partagée without intention known not is not automatiquement has campagne of disinformation. The disinformation exige has basis for the intention of be mistaken or of nuire. This intention can sometimes be documented by of instructions, has coordination, has model of financement or of repetitions malgré of corrections; she not itself déduit not of the only caractère faux of has claim.
The propaganda is has category more large and historically variable. She can contenir of facts vrais, sélectifs or faux, and be scope by has State, has mouvement, has company or has network. For avoid has label totale, describe the message, its émetteur déclaré, its but apparent, its canaux, its coordination, its public and the reply attendue.
The fact-checks must show their chaîne of verification. HAS claim can be fausse, trompeuse, invérifiable or sortie of its context; these categories not are not interchangeables. The readers must power remonter au document primary, connaître the date of évaluation and voir the corrections.
Platforms digital and systems of recommandation
HAS platform not is not has actor psychological unique. She can fournir has space of publication, classer of content, recommend of comptes, mesurer the engagement, vendre of the advertising, appliquer of rules and traiter of signalements. Each function has its propres mechanisms and data.
For has recommandation, distinguish: content disponible; content affiché; content réellement vu; content compris; content mémorisé; belief; partage; and behaviour hors platform. HAS ranking can increase the probabilité of exposure, but has exposure not is not has persuasion. The data of engagement agrègent of clics, of réponses, of curiosités and of réactions négatives; they not mesurent not directement the adhésion.
The audits must avoid of révéler of means of contourner of protections or of cibler of individuals. Préférer of agrégats, of comptes of test autorisés, of data public minimisées and has protocol reproducible. Document the changements of interface, the périodes, the countries and the rules of personnalisation. HAS comparison between users can be confused by their preferences initiales.
Dynamiques of group and communities in ligne
HAS community creates of rules of participation, of roles and of sanctions. The modérateurs can reduce the harassment; of membres can fournir has expertise or has soutien. The fermeture of has group can protect of people or limit the contradiction. It must document the rule and its application, not juger the community of after its label.
The dynamique of group can amplifier the norms, produce has auto-selection and récompenser the langage moral. The researcher must be transparent on its position, not not infiltrer of espaces privés without basis ethics and withdraw the identifiants. The captures of écran can be falsifiées, décontextualisées or create has risk of rediffusion; keep the original of way sécurisée and publish only this that is necessary.
Education, advertising and communication political
Education vise idéalement the understanding, the reasoning and the capacity of revision; she comporte however of choice of programme, of framing and of authority. Demander if the élèves can distinguish does, interpretation and value; if of sources contradictoires are presented; and if the contestation respectueuse is possible.
The advertising is explicitement persuasive, but the transparency of the sponsor, the clarté of conditions and the protection of mineurs remain essentielles. The targeting basé on has vulnérabilité, the dissimulation of the caractère commercial or the renchérissement artificiel of the sortie can justifier has examen renforcé.
The communication political combine information, narrative, mobilisation and identity. In the analyse, separate promesse, data, opinion, attaque, satire and incitement. Verify the provenance of images and of statistics. Not not déduire the fraude or the manipulation of has forte mobilisation, of has message efficace or of has ton émotionnel.
Peur, indignation morale and perception of the menace
The peur can orienter the attention towards has risk and accélérer has decision. L’indignation can report has transgression and soutenir has asks of réparation. These emotions have has function social and not are not of evidence of irrationalité. They become has objet of préoccupation when the risk is presented without basis, that the urgence bloque the verification, that a group is déshumanisé or that the refus entraîne has sanction.
Code séparément: description of the danger; probabilité announced; gravité; person désignée as responsible; solution proposée; delay; appeal to the action; possibility of correction. Compares the discours to the data of risk disponibles and recognise the uncertainty. Avoid the images or details that pourraient réactiver has traumatisme or servir to harceler.
Autonomy, consentement éclairé and audiences vulnérables
L’autonomie not is not only the absence of menace. She suppose has capacity practice to understand, compares, refuser and revenir on has choice. The consentement éclairé exige has information pertinente, compréhensible and no dissimulée derrière has interface or has jargon. HAS person can consentir to has action without consentir to has réutilisation of its data.
The audiences vulnérables not are not définis by has identity figée. The vulnérabilité can be situationnelle: âge, dependence, urgence, isolement, handicap, barrière linguistique, précarité, fatigue or manque of alternatives. The category must servir to strengthen the protections, never to withdraw has voix or to présumer has incapacité.
The situations help compare mechanisms and do not identify a person.
Freedom of expression and disagreement
The disagreement is has composante of the debate public. HAS person can changer of avis, be convaincue or relayer has erreur without be manipulée. HAS restriction to the parole must be examinée according to the right applicable, the need, the proportionality, the no-discrimination and the voies of appeal. The dossier not remplace not the analyse legal.
The editorial safeguards are compatibles with the pluralism: questions neutres and testable; séparation of facts, inférences and opinions; right of reply; correction visible; transparency of sources; statement of conflicts; and withdrawal temporary of has allégation when the evidence is insuffisante. HAS accusation of pressure is has objet to document, not has conclusion automatique.
Standards of evidence and inference causal
HAS evidence must respond to the question posée. HAS message can démontrer that a énoncé has been publié; it not démontre not that the has vu nor its effect. HAS investigation can mesurer has belief declared; she not démontre not always has behaviour. HAS corrélation between exposure and result can venir of has cause commune or of has auto-selection.
HAS inference causal solide asks has comparison pertinente, has temporalité cohérente, has measure fiable, has mechanism plausible and the examen of variables competing. The experiments randomisées can isoler has difference moyenne in the conditions of the protocol. The observations of terrain give of the ecology but are exposées to the confusion. The studies qualitatives can révéler of mechanisms vécus and of against-exemples without estimer has prévalence.
Avoid the faux seuils. The signification statistical not is not the importance practice; the taille of effet and the uncertainty importent. Seek the analyses préenregistrées, the réplications, the data and the code when c’est possible. HAS absence of evidence not is not has evidence of absence, but has accusation must not be publishede without basis proportionate.
Methods expérimentales and observationnelles
Expérimentation. Define has manipulation limitée, randomiser if possible, provide has group comparateur, mesurer the état initial and the effet, and enregistrer the criteria before the analyse. Protect the participants of has exposure inutile, débriefer when the protocol uses has dissimulation autorisée and not not généraliser au-delà of the framework.
Investigation. Specify the population, the recrutement, the taux of reply, the formulation, the ordre of questions, the pondérations and the no-réponses. The réponses déclarées can be influencées by the désirabilité social and the memory.
Observation digital. Describe the API or the mode of échantillonnage, the period, the comptes audiences or privés, the changements of algorithme and the rules of collecte. Not not confuse viralité, visibility, engagement and persuasion.
Analysis of content. Establish has codebook, former several codeurs, mesurer the accord, keep of exemples no identifiants and publish the exclusions. Ajouter the content that contredisent the hypothesis.
Entretien and method qualitative. Distinguish experience rapportée, interpretation of the researcher and does externe. Protect the people against the ré-identification. HAS narrative can establish that a person has perçu has pressure; it is not sufficient to attribute the intention of the émetteur.
Safeguards ethical
- Questions neutres, testable, and langage descriptif.
- No data private or information permettant has localisation or has harassment.
- Not of diagnosis to distance, of label psychological or of generalisation identity-based.
- Consentement, right of withdrawal, minimisation and conservation limitée of data.
- Assessment independent when of people vulnérables, mineures or exposées to has risk are concernées.
- Statement of financements, affiliations, interests and limits.
- Publication of criteria of inclusion and of exclusion before the analyse when possible.
- Conservation of has audit trail: versions, decisions, sources consultées, corrections and raisons of retraits.
Case-study method
The cases ci-dessous are of démonstrations hypothetical, not of événements real.
- Délimiter. Describe the event, the period, the canal and the question without label accusatory.
- Cartographier the actors. Identifier roles, interests déclarés, powers and responsibilities; not not déduire has intention.
- Keep the trace primary. Capturer the URL, the date, the version, the context and the chaîne of conservation.
- Describe the mechanism. Framing, omission, repetition, pressure, ranking, deception, menace or other, with indicators observable.
- Establish the exposure. That has pu voir, that has vu, combien of temps, in which environment?
- Mesurer the effet. Understanding, rappel, attitude, choice or behaviour; before/after or comparison if possible.
- Seek the alternative. Preferences initiales, event extérieur, selection, erreur of measure, changement of rule.
- Entendre the réponses. Contacter the parties with has question precise and has delay raisonnable; distinguish absence of reply and contradiction.
- Decide. Publish, publish with réserves, différer, correct or withdraw temporairement. The evidence insuffisante not is not has échec: c’est has decision editorial responsible.
Comparative matrix of coding
| Dimension | Question of coding | Modalités possible | Evidence minimale |
|---|---|---|---|
| Source | That parle or agit? | person, institution, media, platform, group, unknown | trace primary and identification prudente |
| Intention | HAS but is-it documenté? | explicite, inféré with réserve, unknown | instruction, statement, financement or absence signalée |
| Mechanism | How the effet would be produces? | reason, emotion, repetition, norm, authority, interface, menace, deception | exemple daté and codebook |
| Transparency | The but and the sponsor are-ils visibles? | clair, partiel, masqué, unknown | message, label, conditions |
| Choice | The refus is-it praticable? | facile, possible but coûteux, bloqué, unknown | parcours real and alternatives |
| Public | That is ciblé or exposé? | général, segment explicite, vulnérabilité situationnelle, unknown | criteria of audience agrégés |
| Exposure | That has effectivement rencontré the content? | mesurée, estimée, only possible | data of usage or investigation |
| Effect | That changes-t-it? | rappel, belief, attitude, partage, behaviour, no effect observé | measure and fenêtre temporelle |
| Causality | Which force of inférence? | descriptive, associative, quasi-expérimentale, expérimentale | method, comparateur, limits |
| Reply | The people concernées have-they répondu? | oui, no, no contactées | courrier, delay, citation fidèle |
| Risk | Publication expose-t-she quelqu’un? | faible, modéré, élevé | analysis of ré-identification |
| Decision | That does the publisher? | retenir, nuancer, différer, withdraw, correct | decision datée and motivée |
Research hypotheses
These hypotheses are of research notes falsifiables, not conclusions:
- H1 — TO content constant, has sponsor clairement identifié increases the capacity declared to assess the message by report to has sponsor masqué; the effet can varier according to the trust initiale.
- H2 — HAS norm descriptive perçue modifie more the behaviour when the person juge the group pertinent; mesurer simultanément perception, intention and action.
- H3 — The repetition increases the familiarité more sûrement that elle not produces the adhésion; distinguish rappel, truth perçue and behaviour.
- H4 — L’activation émotionnelle increases the attention but reduces or increases the verification according to the temps disponible and the difficulté of the tâche; test these modérateurs.
- H5 — L’exposition to of content opposés not produces not has effect uniforme on the polarisation; the identity, the ton and the quality of the source must be mesurés.
- H6 — HAS interface with refus équivalent and explanation claire conduit to more of choice réversibles that a interface asymétrique; mesurer sortie, understanding and regret.
- H7 — The trust in has source is better prédite by the skill perçue, the cohérence and the transparency that by the only status; verify the contextes where this model échoue.
- H8 — HAS accusation of pressure will be less fiable when the exposure, the effet and the alternatives causales not are not separate in the questionnaire; prétester the formulation.
Press and institutional corpus
The corpus proposed in sources/bibliography-commentee.md combines of sources françaises and anglophones: Council of Europe and CEDH; OHCHR; UNESCO; Commission European; OCDE; OMS; APA; OSCE; Reuters Institute; articles of research; The Monde; The Figaro; Libération; France 24; BBC; Reuters; The Guardian; Financial Times; The New York Times; AP; Politico Europe; Deutsche Welle. The corpus does not claim équilibrer mécaniquement of opinions in opposant two articles: the équilibre sought is celui of methods, of perspectives and of evidence.
How to use this corpus
- Distinguish has study originale of has commentaire or of has article that the résume.
- Relever sample, recrutement, measure, period, countries, taille, uncertainty and limits lorsqu’ils are disponibles.
- Separate corrélation, temporalité and causality.
- Seek réplications, results nuls, critiques methodological and against-evidence.
- Not not psychologiser of individuals or of communities to partir of has citation, of has photo or of has behaviour political.
- Déclarer the paywalls, résumés only, pages inaccessibles and links to verify.
- Remonter of the press towards the document primary: decision, study, data, enregistrement or statement intégrale.
- Noter this that the source not allows not of conclure.
Editorial: right of reply, corrections and transparency
Before publication, adresser to the people or institutions mises in cause of questions factuelles précises, with has delay raisonnable. Reproduce their reply without the present as evidence independent. If they not répondent not, the indicate sobrement; not not turn the silence in aveu.
Hold has log public of corrections: date, passage, type of erreur, evidence of correction and impact on the conclusion. Ajouter has registre of rights and of the provenance: origine of each document, autorisation of republication, version archived and restrictions. Keep has audit trail internal of choice and of changements.
Provide has protocol of withdrawal temporary lorsqu’une source primary becomes indisponible, that a erreur important is signalée, that a risk of ré-identification appears or that a reply contradictoire modifie substantiellement the dossier. The withdrawal not efface not the historical: it must be daté, motivé and suivi of has réévaluation.
HAS revue independent must examine the method, the risks and the langage. The conflicts of interests must be déclarés. HAS date fixe of reassessment — by exemple to six mois or after has changement réglementaire or technique important — must be inscrite in the sheet of suivi. The links and métadonnées of this paquet remain to contrôler before any placing in ligne.
FAQ
HAS person that changes of avis has-t-she been manipulée?
No. The changement can résulter of has information fiable, of has réflexion, of has discussion or of has pressure. It must document the mechanism, the context and the effet, without confuse result and manipulation.
HAS content émotionnel is-it manipulateur?
No. L’émotion is fréquente in the art, the journalism, the education and the debate democratic. The question is celle of the transparency, of the véracité, of the targeting, of the contrainte and of the effet observé.
The viralité prouve-t-she the disinformation?
No. The viralité measure has diffusion or has visibility according to has definition data. She not prouve nor the fausseté, nor the intention, nor the belief, nor the effet.
HAS algorithme manipule-t-it automatiquement?
No. It must describe the function, the ranking, the exposure, the alternatives, the interests and the effects. HAS recommandation can have has effect without that a intention humaine of be mistaken soit établie.
Can-on parler of « gourou », of narcissisme or of secte?
Not as diagnosis or raccourci. Use of terms descriptifs and document the acts, the rules, the pressions, the resources and the possibilités of sortie. The étiquettes psychological no étayées are excluded.
HAS pressure alléguée is-she has conclusion?
No. HAS pressure alléguée is has objet to document: paroles, rules, acts, temporalité, target, effect, alternatives and right of reply.
How protect the people citées?
Minimiser the data, anonymiser when possible, supprimer the coordonnées and details of localisation, verify the risk of ré-identification and give has right of reply. Not publish that this that is necessary to the question.
That do lorsqu’un link is inaccessible?
The report. Seek has copie official or has document primary accessible, verify by has seconde source and not not do rest has accusation on has article inaccessible and no corroboré.
This dossier establishes-it has result empirical?
No. It proposes has framework of research, of definitions, of hypotheses, has matrix and has corpus to verify. It not present not of collecte nor of analysis originale completed.
Conclusion
The psychology social allows of understand how the context oriente the judgment without reduce the people to of marionnettes nor the groups to of essences. The manipulation cognitive is has hypothesis demanding: she appelle has description of the mechanism, of the public, of the exposure, of the effet, of the autonomy and of evidence competing. The prudence not protects not only the people accusées; she améliore also the quality of the investigation.
The debate democratic has besoin of has vocabulaire precise. Persuader not is not contraindre; cadrer not is not falsifier; partager has erreur not is not prouver has campagne; be ému not is not perdre its autonomy. When of practices opaques or coercitives are documentées, they must be décrites with rigueur and soumises to of safeguards of right, of proportionality and of appeal.
Prochaines steps of research
- Verify each URL, title, auteur, date, status of access and annotation of the bibliography.
- Select trois cases real documentés by of sources primary accessibles, after analysis of risks.
- Préenregistrer the question, the codebook, the exclusions and the indicators before coding.
- Do relire the method and the langage by has person independent, if possible in the two langues.
- Put in place the registre of corrections, the registre of rights, the audit trail and the protocol of withdrawal.
- Réévaluer the dossier to date fixe and after changement majeur of platform, of réglementation or of corpus.
Bibliography and sources
Publication note
Version/access date: 16 August 2026. External links not have not pu be consulted in the environment hors ligne. No déploiement not is revendiqué.
Fichiers of sources: bibliography commentée · README of sources
Warning and scope
The situations help compare mechanisms and do not identify a person.
The psychology social étudie how the judgments, the emotions and the behaviours can be influencés by the presence real, imaginée or supposée of autrui, by the norms, the roles, the institutions, the groups and the environments matériels or digital. This definition describes of conditions of influence; she not allows not, to she only, of inférer has intention cachée, has personnalité or has pathologie.
The mot « manipulation » is loaded. In this dossier, it means has hypothesis opérationnelle: has configuration where has actor or has system seeks, or produces by its design, has modification of judgment or of conduite in réduisant the access to has information pertinente, in exploitant has asymétrie, in masquant the but, in imposant has cost indu au refus, or in altérant the conditions of has choice informé. This definition must be testée, not présumée. HAS content persuasive, has advertising, has discours political, has correction or has disagreement not are not in eux-mêmes of evidence of manipulation.
The périmètre covers the mechanisms sociaux and cognitifs, the media, the platforms, the education, the advertising and the communication political. It not covers not the évaluation clinical of individus, the attribution of intentions on the basis of their style, nor the surveillance of groups. The data private, the accusations no verified, the generalisations identity-based and the details that faciliteraient the harassment are excluded.
The announced knowledge basis and the document cadre éditorial not were not présents in the directory of production. This dossier uses only the explicit editorial requirements of the brief; it does not claim reproduce of pages of the site « Struggle For Free Speech ». The present localement has been écarté as demandé.