Methodology: Academic Foundations & Limitations

This document outlines the theoretical frameworks, methodological choices, and acknowledged limitations of this speculative futures study.

---

1. Futures Studies Frameworks

1.1 Causal Layered Analysis (CLA)

Source: Inayatullah, S. (1998). "Causal layered analysis: Poststructuralism as method." Futures, 30(8), 815-829.

Description: CLA is a futures research method that examines issues at four levels of depth:

Layer	Description	Example Application
Litany	Surface-level facts, media reports	"Chega won 18% of the vote"
Systemic Causes	Social, economic, political drivers	Housing crisis, corruption scandals, brain drain
Worldview	Underlying beliefs, assumptions	"Portugal is losing its identity" vs "Portugal is becoming more European"
Myth/Metaphor	Deep narratives, archetypes	"The Nation as Family" vs "The Nation as Community"

Application in this study:

• Used to analyze candidate supporter motivations beyond surface positions
• Particularly applied to understanding Chega voters' deep fears and hopes
• Informs persona value profiles

Key insight: Political positions emerge from deep worldviews and myths, not just rational policy preferences.

---

1.2 Three Horizons Framework

Source: Sharpe, B., Hodgson, A., Leicester, G., Lyon, A., & Fazey, I. (2016). "Three horizons: A pathways practice for transformation." Ecology and Society, 21(2).

Description: Three Horizons provides a structure for thinking about system change:

• H1 (Horizon 1): The dominant present system, "business as usual"
• H2 (Horizon 2): Transitional innovations, disruptions, experiments
• H3 (Horizon 3): The emerging future system

Visual representation:

Dominance
    │    H1 ╲
    │        ╲     H2
    │         ╲   /  ╲
    │          ╲ /    ╲  H3
    │           ╳      ╱ ╱
    │          / ╲    ╱ ╱
    │         /   ╲  ╱ ╱
    └─────────────────────────────► Time
         Now       Transition    Future

Application in this study:

• H1: Portugal in January 2026 (current challenges, institutions, trends)
• H2: What each candidate's presidency enables or blocks
• H3: Portugal 2030 under each candidate's influence

---

1.3 2×2 Scenario Matrix

Source: Standard foresight methodology, widely used in strategic planning (Shell scenarios, Global Business Network tradition).

Description: A 2×2 matrix is created by selecting two critical uncertainties as axes, generating four distinct scenario quadrants.

Our axes:

• X-Axis: Economic Trajectory (Development/Growth ↔ Stagnation/Crisis)
• Y-Axis: Social Openness (Open/Cosmopolitan/Inclusive ↔ Closed/Backwards)

Rationale for axis selection:

1. Economic trajectory captures the material conditions debate
2. Social openness captures the cultural/identity debate
3. These two dimensions are somewhat independent (high growth can be exclusive or inclusive)
4. They align with Portugal's key political divides

---

2. Political Psychology Foundations

2.1 Lakoff's Moral Politics

Source: Lakoff, G. (1996, 2002). Moral Politics: How Liberals and Conservatives Think. University of Chicago Press.

Core theory: Political attitudes derive from deep cognitive frames rooted in family metaphors:

Frame	Metaphor	Values	Political Expression
Strict Father	Nation as family with authoritarian father	Discipline, self-reliance, moral order	Conservative/right
Nurturant Parent	Nation as family with caring parents	Empathy, mutual support, fairness	Progressive/left
Biconceptual	Mixed frames, context-dependent	Variable	Moderate/swing

Application in this study:

• Personas are assigned frame tendencies
• Helps explain why voters support candidates whose policies may not maximize their material interests
• Informs dialogue simulations between different worldviews

Critical note: Lakoff's model emerged from US context; Portuguese political culture has distinct characteristics (Catholic social teaching influence, post-colonial context, European integration) that may modify these frames.

---

2.2 Schwartz Human Values Theory

Source: Schwartz, S. H. (1992). "Universals in the content and structure of values: Theoretical advances and empirical tests in 20 countries." Advances in Experimental Social Psychology, 25, 1-65.

Core theory: Ten universal human values organized in a circular structure:

                    Openness to Change
                           │
            Self-Direction │ Stimulation
                    ╲      │      ╱
          Universalism ╲   │   ╱ Hedonism
                        ╲  │  ╱
        Benevolence ─────  ●  ───── Achievement
                        ╱  │  ╲
            Tradition ╱    │    ╲ Power
                    ╱      │      ╲
              Conformity   │   Security
                           │
                    Conservation

Four higher-order values:

1. Self-Transcendence: Universalism + Benevolence
2. Self-Enhancement: Power + Achievement
3. Openness to Change: Self-Direction + Stimulation
4. Conservation: Tradition + Conformity + Security

Political correlations (from cross-cultural research):

• Higher Self-Transcendence and Openness → Political left
• Higher Self-Enhancement and Conservation → Political right

Application in this study:

• Each persona has a Schwartz value profile
• Helps predict which proposals resonate emotionally
• Adds depth beyond simple left-right positioning

---

3. Synthetic Persona Methodology

3.1 "Silicon Sampling" Research

Source: Argyle, L. P., et al. (2023). "Out of One, Many: Using Language Models to Simulate Human Samples." Political Analysis, 31(3), 337-351.

Approach: Condition LLMs on demographic backstories derived from real survey data to generate "silicon samples" that approximate human response patterns.

Validation: European Parliament voting simulation achieved F1 score of ~0.793 (Ponizovskiy et al., 2024).

3.2 Critical Limitations

Source: Bisbee, J., et al. (2024). "Synthetic Replacements for Human Survey Data? The Perils of Large Language Models." Political Analysis, 32(4), 401-416.

Key findings:

1. Reduced variation: LLM responses show less variance than real human surveys
2. Coefficient differences: 48% of regression coefficients differ significantly from human data
3. Prompt sensitivity: Minor prompt changes produce different results
4. Temporal instability: Same prompt yields different results over time

3.3 Our Methodological Position

We adopt a "generative, not predictive" approach:

What we claim	What we do NOT claim
Personas represent plausible perspective types	Personas represent actual individuals
Scenarios explore possibility space	Scenarios predict likely outcomes
Dialogues surface potential tensions	Dialogues predict real conversations
The study aids reflection	The study provides evidence

Key safeguards:

1. All personas explicitly marked as synthetic
2. No quantitative claims about population distributions
3. Focus on qualitative exploration of perspectives
4. Transparent about LLM limitations

---

4. Experiential Futures

4.1 Conceptual Framework

Source: Candy, S. (2010). "The Futures of Everyday Life." PhD Dissertation, University of Hawaii.

The Experiential Futures Ladder:

Level	Description	Immersion
Setting	General future context	Low
Scenario	Narrative description	Medium
Situation	Specific moment in time	High (our focus)
Stuff	Tangible artifacts	Highest

"Day in the Future" vignettes operate at the Situation level:

• Specific time and place
• Sensory details (sight, sound, smell)
• Emotional experience
• Daily routine touchpoints

4.2 Design Principles for Vignettes

1. Grounded specificity: Real places, realistic details
2. Multiple perspectives: Same scenario, different experiences
3. Mundane futures: Focus on everyday life, not spectacular events
4. Emotional truth: What does it feel like to live there?
5. Implicit policy: Show consequences through experience, not exposition

---

5. Known Biases & Mitigations

5.1 LLM Biases

Bias	Evidence	Mitigation
Left-leaning default	Multiple studies show ChatGPT defaults center-left	Explicit conservative persona conditioning; careful prompt design
WEIRD overrepresentation	Training data skews Western, Educated, Industrialized, Rich, Democratic	Deliberate inclusion of rural, low-education, non-Western immigrant personas
Stereotyping	LLMs may reproduce stereotypes	Review personas for nuance; avoid caricature
English-language bias	Most training data in English	Seek Portuguese-language sources where available

5.2 Researcher Biases

Bias	Description	Mitigation
Value position	Researcher holds progressive values	Transparent critical framing; genuine effort to inhabit conservative perspectives
Urban perspective	Project developed in urban context	Deliberate rural and interior personas
Academic framing	Academic language may not resonate with all	Accessible language in vignettes and dialogues

5.3 Structural Biases

Bias	Description	Mitigation
Voter focus	Study centers on election	Include affected non-voters (immigrants)
National frame	Focus on Portugal may miss transnational dynamics	Include diaspora and geopolitical context
Present-anchored	Scenarios start from 2026 conditions	Acknowledge uncertainty; multiple scenarios

---

6. Ethical Framework

6.1 Empathetic Inclusion

All political perspectives are explored with genuine effort to understand their internal logic, including:

• Far-right voters' fears about cultural change
• Conservative concerns about social stability
• Progressive hopes for equality
• Liberal emphasis on individual freedom
• Socialist concerns about workers' rights

6.2 Critical Framing

Empathy does not mean equivalence. When perspectives include:

• Anti-democratic elements
• Discrimination against minorities
• Human rights violations

...these are analyzed with equal depth but with explicit critical context.

6.3 Representation Ethics

• Personas are marked as synthetic constructions, not real people
• No claims to speak for any community
• Goal is to surface perspectives, not represent populations
• Immigrant and minority personas created with awareness of our positionality

---

7. Academic Integrity Commitments

1. All scenarios presented as speculative, never as predictions
2. Limitations section included in every public output
3. Sources cited throughout with hyperlinks
4. Method clearly explained for replication and critique
5. Personas explicitly synthetic, never presented as real individuals
6. No electoral recommendation - study does not endorse any candidate

---

8. References

Futures Studies

• Inayatullah, S. (1998). Causal layered analysis: Poststructuralism as method. Futures, 30(8), 815-829.
• Sharpe, B., et al. (2016). Three horizons: A pathways practice for transformation. Ecology and Society, 21(2).
• Candy, S. (2010). The Futures of Everyday Life. PhD Dissertation, University of Hawaii.

Political Psychology

• Lakoff, G. (2002). Moral Politics: How Liberals and Conservatives Think. University of Chicago Press.
• Schwartz, S. H. (1992). Universals in the content and structure of values. Advances in Experimental Social Psychology, 25, 1-65.

LLM Methodology

• Argyle, L. P., et al. (2023). Out of One, Many: Using Language Models to Simulate Human Samples. Political Analysis, 31(3), 337-351.
• Bisbee, J., et al. (2024). Synthetic Replacements for Human Survey Data? Political Analysis, 32(4), 401-416.

Portugal Context

• See research documents for specific sources on Portuguese politics, demographics, and trends.