Research Philosophy & Epistemology: Ontology, Epistemology, Methodologies & Ethics

TAKEAWAYS:

• Ontology determines whether social work research views reality as objective (fixed) or subjective (constructed), directly shaping research questions (e.g., "Do poverty levels objectively measure suffering?" vs. "How do slum dwellers subjectively experience poverty?").
• Epistemology divides research into positivist (quantifiable truths, e.g., NTC’s call-center efficiency metrics) and interpretivist (lived experiences, e.g., Pathao drivers’ stories of harassment), with Eastern philosophies (e.g., Dharma-based ethics) offering a third lens.
• Methodologies (quantitative/qualitative/mixed) are tools, not philosophies—choose based on the research question, not personal bias (e.g., measuring child labor rates vs. understanding child laborers’ agency).
• Ethics in social work research isn’t optional: it’s the raison d’être. Examples include NEPSE’s conflict-of-interest rules or eSewa’s GDPR-compliant data storage for vulnerable users.
• Politics of research hides in "neutral" questions (e.g., "Why do 80% of Kathmandu’s homeless sleep at Char Narayan?" could ignore systemic causes like land grabs).
• Research questions must align with philosophy: A positivist question ("What % of elderly abuse cases go unreported?") requires surveys; an interpretivist one ("How do caregivers frame elder abuse?") needs interviews.

The Philosophical Bedrock: Ontology and Epistemology

Social work research isn’t just about collecting data—it’s about believing what data means. These beliefs stem from ontology (what exists?) and epistemology (how do we know it?).

Ontology: The Nature of Reality

Ontology asks: Is the social world fixed and objective, or shaped by human interaction? Two dominant views collide in social work:

Eastern Ontology: Dharma and Relational Reality Western ontology often pits individual vs. society. Eastern traditions (e.g., Hinduism, Buddhism) see reality as interdependent (interbeing, Thich Nhat Hanh). For example:

• A Daraz delivery person’s "ontology" might view their route not as a fixed path (objectivist) but as a dynamic negotiation with traffic gods (Bhairav), customers’ moods, and their own karma.
• Worked Example: In a study on chhaupadi (menstrual exile), objectivist researchers might count cases; constructivist researchers might explore how women redefine purity during menstruation. An Eastern lens would ask: How does the village’s collective Dharma (duty) both enforce and challenge chhaupadi?

Epistemology: How We "Know" the Social World

Epistemology bridges ontology and methodology. Three key approaches:

1. Positivism (Scientific Realism)

   • Knowledge comes from empirical, measurable data.
   • Tools: Surveys, experiments, statistics.
   • Example: Ncell’s study on "customer satisfaction scores" uses Likert scales to quantify complaints.
   • Critique: Ignores power dynamics (e.g., a call-center agent’s scripted responses may not reflect real dissatisfaction).

2. Interpretivism (Social Constructionism)

   • Knowledge is co-created through interaction.
   • Tools: Interviews, participant observation, discourse analysis.
   • Example: A social worker studying kamlari (bonded child laborers) might find that "freedom" means different things to girls, parents, and employers.
   • Visual:

     flowchart TD
       A["Researcher"] -->|"Asks"| B["Kamlari Girl: 'Freedom is eating rice without fear'"]
       B -->|"Contrasts with"| C["Employer: 'Freedom is completing the 7-year term'"]
       B -->|"Shapes"| D["Researcher's Understanding"]

3. Eastern Epistemology: Pratyaksha (Direct Perception) and Anumana (Inference)

   • Knowledge comes from direct experience (pratyaksha) and holistic inference (anumana), not just data.
   • Example: A social worker in a tole (village) might "know" trust levels not by surveys but by observing who shares tea and who avoids eye contact.
   • Worked Example: Khalti’s "digital inclusion" reports use both quantitative data (transaction volumes) and qualitative insights (elders describing "fear of the app") to design user-friendly interfaces.

Methodologies: Tools for Different Philosophies

Methodology is the how—but it’s shaped by ontology and epistemology. Below is a decision tree to match questions to methods:

flowchart TD
  A["Start: What is your research question?"] --> B{"Is reality fixed?"}
  B -->|"Yes"| C["Positivist\nQuantitative"]
  B -->|"No"| D{"Is reality socially constructed?"}
  D -->|"Yes"| E["Interpretivist\nQualitative"]
  D -->|"Yes, but holistic"| F["Eastern\nMixed/Participatory"]
  C --> G["Surveys\nExperiments\nStatistics"]
  E --> H["Interviews\nFocus Groups\nEthnography"]
  F --> I["Participatory Action Research\nNarrative Inquiry\nYoga-based therapy studies"]

Quantitative Methods: The Ncell Approach

Definition: Systematic, numerical data collection to test hypotheses. Steps (with Ncell Customer Complaint Analysis example):

1. Define Variables: Complaint type (billing, network, device), severity (1–5).
2. Data Collection: Surveys to 10,000 users (stratified by region/age).
3. Analysis: Chi-square test to see if complaints correlate with network towers’ age.
4. Conclusion: "Older towers in Province 2 have 3x more network complaints."

Advantages:

• Generalizable (e.g., "20% of Kathmandu users report drop calls").
• Objective (reduces researcher bias).

Disadvantages:

• Misses why users complain (e.g., a "5" rating might hide despair).
• Visual:

Qualitative Methods: The Pathao Driver’s Story

Definition: Explores meanings, not numbers. Used when "how" and "why" matter more than "how many." Tools:

• Interviews: "Tell me about a time you felt unsafe delivering in Kathmandu."
• Focus Groups: "How do drivers discuss harassment among themselves?"
• Participant Observation: Riding with Pathao drivers to map unsafe routes.

Worked Example: A study on traffic police corruption in Kathmandu might use:

• Quantitative: "X% of drivers pay bribes" (survey).
• Qualitative: Drivers’ narratives of how bribes are demanded (e.g., "The cop points to my bike’s scratch").

Advantages:

• Reveals power dynamics (e.g., why women drivers pay more).
• Flexible (questions evolve with participants).

Disadvantages:

• Not generalizable (can’t say "all drivers" based on 20 interviews).
• Visual:

Mixed Methods: The Daraz Delivery Algorithm

Definition: Combines both to answer complex questions (e.g., "Why are deliveries delayed in Lalitpur?"). Example: Daraz might:

1. Quantitative: Track delivery times by pincode (data shows Lalitpur takes 4 hours vs. 2 in Thapathali).
2. Qualitative: Interview drivers to find traffic jams at Tribhuvan Chowk are worse during Dashain.
3. Solution: Reroute algorithms + driver incentives for peak hours.

Eastern Mixed Methods: Participatory Action Research (PAR)

• Example: A social work team in a Newar community might:
  • Quantitative: Count families affected by heritage home demolitions.
  • Qualitative: Hold dhikri (community meetings) to co-design solutions.
  • PAR: Implement a pilot, then evaluate with the community.

The Politics of Research: Power, Ethics, and Hidden Agendas

Research is never neutral. Even "harmless" questions can reinforce oppression or ignore solutions.

Ethical Challenges in Social Work Research

Worked Example: eSewa’s Data Ethics

• Problem: eSewa collects biometric data (fingerprints) for transactions. A positivist study might quantify "fraud reduction," but an interpretivist lens would ask:
  • How do Dalit users feel about fingerprint scans? (Many associate them with caste-based surveillance.)
  • Who owns the data? (eSewa or the government?)
• Solution: eSewa now offers opt-out biometrics and publishes a Data Ethics Charter.

The Politics of Research Questions

Not all questions are equal. Consider:

1. Neutral-Seeming Question: "What are the causes of homelessness in Kathmandu?"
   • Hidden Bias: Blames individuals ("lazy") rather than systemic causes (land grabs, lack of affordable housing).
2. Critical Question: "How do real estate developers and municipal officials collaborate to displace squatters in Thapathali?"
   • Power Analysis: Exposes alliances between corporations and local governments.

Visual:

classDiagram
  class ResearchQuestion {
    +Is it value-neutral?
    +Does it center marginalized voices?
    +Who benefits from the answer?
  }
  class PowerStructure {
    <<abstract>>
    +Government
    +Corporations
    +Academia
    +Communities
  }
  ResearchQuestion "1" --> "0..*" PowerStructure : Reinforces or challenges
  note for ResearchQuestion "Example:\n'Why do slums exist?' vs.\n'How do slum dwellers resist eviction?'"

In the Real World

1. Ncell’s Customer Satisfaction Surveys

   • Philosophy: Positivist (quantifies complaints).
   • Ethics: Uses randomized response techniques to reduce bias (e.g., "Did you complain? Yes/No/Unsure" instead of direct questions).
   • Impact: Led to Ncell’s "Complaint Escalation Team" for severe issues.

2. Pathao’s Driver Safety Study

   • Philosophy: Mixed methods.
     • Quantitative: Maps accident hotspots using GPS data.
     • Qualitative: Interviews drivers on harassment (e.g., "Police ask for sex in exchange for safe routes").
   • Outcome: Pathao now pays drivers hazard allowances in high-risk areas and lobbies for women-only delivery slots.

3. Daraz’s Algorithm Bias

   • Philosophy: Initially positivist ("Optimize delivery routes for speed").
   • Problem: Data showed deliveries to "informal" addresses (e.g., squatter colonies) were delayed. Qualitative interviews revealed geocoding errors (Daraz’s system didn’t recognize non-pincode areas).
   • Solution: Partnered with local NGOs to map informal addresses, then retrained the algorithm.

4. NEPSE’s Insider Trading Investigations

   • Philosophy: Legal positivism (follows rules) but often fails to ask why rules are broken.
   • Critique: A social work lens would explore:
     • How do brokers pressure analysts to leak tips? (Interpretivist)
     • Which NEPSE employees benefit from weak enforcement? (Critical theory)

Generating Research Questions: From Philosophy to Practice

A strong research question aligns with your ontology, epistemology, and ethical stance. Follow this 5-step framework:

1. Identify the Problem
   • Example: "Child labor in brick kilns persists despite laws."
2. Choose an Ontology
   • Objectivist: "What % of child laborers are under 14?"
   • Constructivist: "How do child laborers define work vs. play?"
3. Select Epistemology
   • Positivist: Use labor inspectorate records.
   • Interpretivist: Conduct play sessions to observe children’s work routines.
4. Ethical Safeguards
   • For interpretivist work: Child-friendly consent, non-judgmental language ("Tell me about your day").
5. Pilot Test
   • Try questions with 2–3 participants. Example:
     • Weak: "Do you work in the kiln?"
     • Strong: "Show me how you spend a day. What parts feel like work? What parts feel like helping?"

Worked Example: Traffic Congestion in Kathmandu

Exam Tip: How to Score Full Marks

This unit is conceptual but applied. Examiners want:

1. Definitions with Examples

   • ❌ "Ontology is the study of being."
   • ✅ "Ontology examines whether social reality is objective (e.g., NTC’s internet access stats) or subjective (e.g., a Newari elder’s definition of ‘connected’ as having a landline)."

2. Critical Analysis

   • Weak: "Quantitative methods are better."
   • Strong:

     "Quantitative methods excel in measuring NEPSE’s trading volume trends, but they fail to explain why insider trading persists—qualitative interviews with brokers might reveal cultural norms like guthi (rotational fund) pressures. A mixed-methods approach would triangulate data for deeper insights."

3. Real-World Links

   • Do: "Like Daraz’s delivery algorithms, social work research must account for informal systems (e.g., guthi networks) that quantitative data often ignores."
   • Don’t: Generic examples like "schools" or "hospitals."

4. Ethics as a Separate Paragraph

   • Always end with:

     "This study adheres to ethical guidelines by [anonymizing data/using assent for minors/avoiding conflict of interest], ensuring participant safety and validity."

5. Diagrams > Text

   • For methodology: Use the decision tree above.
   • For ontology/epistemology: Draw a Venn diagram (but label it clearly as "Western vs. Eastern Research Assumptions").
   • For ethics: A table (like the one above) scores more than bullet points.

Common Pitfalls:

• Confusing ontology with epistemology: Ontology = what exists; epistemology = how we know it.
• Ignoring power: Always ask, "Who benefits from this research?"
• Overlooking Eastern perspectives: Even in positivist questions, note cultural nuances (e.g., "In Nepal, ‘trust’ in surveys may reflect santosh (contentment) rather than truth-telling").