What do I investigate?
Organizational and citizen communication, audience engagement and public accounts of human–AI relationships.
Public communication
and audience response.
I study public communication using computational text analysis, experiments and quantitative methods.
PhD · University of AmsterdamResearch Affiliate · Northeastern University
Question first.Method in service of evidence.
What can we measure and test?
Message features and audience responses in platform data; hypotheses in experiments; the performance of AI-assisted coding.
How do I teach these methods?
Guided notebooks, model comparison and prototype development, with revision through feedback.
02 / RESEARCH PIPELINE
Communication research.
AI as method and topic.
Two research streams connect through computational measurement, quantitative analysis and experimental designs selected for each research question.
THREE DISSERTATION CHAPTERS
Organizational communication on societal issues
Sender, framing and appeal
Context: Russia–Ukraine war.
Comparison: CEO versus brand; self versus other framing; informational versus emotional appeal.
Outcomes: engagement and digital activism.
Citizen discourse and brand strategy
Context: brand DEI decisions.
Comparison: pro- or anti-DEI political user-generated content (PUGC) × brand strategy.
Outcomes: engagement, motive attributions and brand responses.
Emoji symbolism
Context: brand communication about Sustainable Development Goals.
Message feature: symbolic and iconic emojis.
Outcome: consumer engagement.
CURRENT RESEARCH PIPELINE
Two streams · a shared methodological toolkit
Sender identity, political user content and message features in relation to audience engagement and responses.
- CEO versus brand voices during the Russia–Ukraine warPublished Journal of Public Policy & Marketing · JIF 5.3 ↗
- Political user generated content and brand sociopolitical decisionsRevise & resubmit Journal of Interactive Marketing · JIF 6.4 ↗
- Emoji symbolism in SDGs messagesReject & resubmit International Journal of Research in Marketing · JIF 7.5 ↗
AI empowerment framing, public accounts of human–GenAI interaction, and platform access in consumer AI.
- AI empowerment framing and humour in corporate videoTargeting International Journal of Research in Marketing · JIF 7.5 ↗
- Symbiosis of human–GenAI relationshipsTargeting Technovation · JIF 10.9 ↗
- Platform access and AI assistant recommendationsTargeting Marketing Science · 2024 JIF 4.9 ↗
SHARED METHODOLOGY
From constructs to measures and empirical tests
DATA
Platform observations
Corporate posts and videos · political user content · public accounts of human–AI interaction
Communication & public responseTHEORY-LED QUESTIONHuman–AI systems
MEASUREMENT
Computational coding
Theory-informed codebooks · LLM classification · BERT stance models · human-coded evaluation
ANALYSIS & DESIGN
Empirical tests
Statistical models of field associations · experiments testing effects · sensitivity analysis
03 / THE MESSY WORK OF RESEARCH
Which tasks belong
in which tier?
A research-task exercise adapted from Messy Jobs. Discuss task specification, contextual interpretation and responsibility; then compare your choices with a suggested mapping.
Separable tasks
Specified inputs, rules and outputs. Automation can be checked against explicit criteria.
The messy middle
Task definitions or interpretations need revision through contextual and methodological judgment.
Researcher responsibility
Researchers remain accountable for design, interpretation and reporting across all three tiers.
A CONVERSATION STARTER
Place nine research tasks
There is room for disagreement: the tier can change with the research question and the risk of an error.
Teaching pointAsk what can be specified, how the output will be evaluated, and who is responsible for the decision. A task’s tier depends on the study.
Inspired by Messy Jobs ↗ · Luis Garicano · Jin Li · Yanhui Wu
04 / THREE WORKING PLATFORMS
Infrastructure that
opens research questions.
Collection and publication workflows support research on AI communication, framing and political user content.
01 · SOURCE-BACKED OBSERVATION
AI Empowerment Observatory
Separate event and coverage records make AI developments and their reporting traceable.
- Collect event and coverage records with provenance and human review.
- Keep source evidence, classification and human review connected.
- Research use: sample AI coverage to study empowerment framing and compare representations across sources.
02 · AUDIENCE TESTBED
The Brief
Accessible, source-linked AI news provides material for audience research.
- Publish accessible AI news with links to supporting sources.
- Research design: compare loss versus gain and self versus others framing.
- Outcome: audience engagement with AI news; experimental comparisons require assigned framing conditions.
03 · SOCIAL LISTENING
European Security Monitor
A setting for studying naturally occurring political communication and audience responses.
- Research use: follow creators’ political posts and audience responses over time.
- Compare topic, stance and timing with government messages; define alignment through an explicit codebook.
- Link to my PUGC research: examine relations between citizen-created content and organizational communication in a security context.
05 / WHY I CHOOSE FHS · TOGETHER
Research · methods · shared capacity.
Communication and audience response
FHSInformation influence · strategic communication
I BRINGSender identity · framing · political user content
TogetherJoint studies · audience responses · moderators
Measurement and research design
FHSLLM · NLP · quantitative methods
I BRINGComputational coding · experiments · Python/R
TogetherLocal-model evaluation · validation · reproducibility
Researcher training and collaboration
FHSInformation Influence Lab · researcher training
I BRINGHuman-AI systems for research · source-backed platforms
TogetherWorkshops · paired analysis · reusable code
LOOKING AHEAD, TOGETHER
I look forward to working with you.
