Research Design & Approaches: Types, Methods & Philosophies

TAKEAWAYS:

• Research design connects objectives to methods (e.g., experimental vs. survey designs).
• Quantitative research uses numbers and statistics (e.g., Ncell’s customer satisfaction surveys), while qualitative explores depth (e.g., Pathao driver interviews).
• Deductive logic starts with theory (e.g., "If X, then Y" in loan approval models) and inductive builds theory from data (e.g., Daraz’s pricing trends).
• Positivism (objective facts) vs. interpretivism (subjective meanings) shapes how research questions are framed.
• Mixed methods combine both approaches (e.g., eSewa’s transaction data + user feedback).
• Ethical considerations (e.g., anonymity in Khalti’s payment research) must guide all designs.

1. What Is Research Design?

Research design is the blueprint that links research questions to methods, ensuring logical flow and validity. It answers:

• What will be studied?
• How will data be collected?
• Why certain methods are chosen?

A well-designed study avoids bias, ensures reliability, and aligns with objectives.

2. Types of Research Designs

Research designs categorize into five main types, each suited to different research goals:

A. Descriptive Design

• Purpose: Describes phenomena (who, what, when, where, how).
• Methods: Surveys, observations, secondary data.
• Example: A study on "Customer demographics using eSewa" would collect data on age, income, and transaction frequency via a survey.

flowchart TD
    A["Define Research Question: 'Customer demographics using eSewa'"] --> B["Collect Data via Survey: Age, Income, Transaction Frequency"]
    B --> C["Analyze Demographics: SPSS/Excel"]
    C --> D["Report Findings: '60% of eSewa users are 18-35'"]

B. Exploratory Design

• Purpose: Gathers insights for further research (e.g., "Why do Pathao drivers quit?").
• Methods: Pilot studies, focus groups, literature reviews.
• Example: Before designing a large-scale study on driver burnout at Pathao, researchers might conduct 10 semi-structured interviews with drivers to identify key stressors (e.g., traffic delays, fare disputes).

C. Explanatory Design

• Purpose: Explains cause-and-effect relationships.
• Methods: Experiments, quasi-experiments, longitudinal studies.
• Example: A bank like Nabil Bank might test whether interest rate changes affect loan defaults by:
  1. Offering two groups of borrowers different rates.
  2. Tracking repayment rates over 6 months.
  3. Concluding: "Higher rates reduced defaults by 15%."

D. Correlational Design

• Purpose: Measures relationships between variables (no causation).
• Methods: Surveys, statistical analysis.
• Example: A study might find that "higher Daraz delivery fees correlate with lower repeat purchases" (but not prove causation).

E. Causal-Comparative Design

• Purpose: Compares groups to infer past causes (e.g., "Why do NTC users switch to Ncell?").
• Methods: Retrospective analysis, archival data.
• Example: Comparing NTC vs. Ncell customer churn rates over 5 years to identify patterns (e.g., pricing, network quality).

3. Research Approaches: Quantitative vs. Qualitative vs. Mixed

Approaches determine how data is collected and analyzed.

A. Quantitative Approach

• Focus: Numbers, patterns, generalizations.
• Methods: Surveys, experiments, statistical tools.
• Strengths:
  • Objective, measurable.
  • Large sample sizes possible.
• Weaknesses:
  • Lacks depth (e.g., "Why?" not explored).
  • Assumes variables are independent.
• Example: Ncell’s customer satisfaction survey uses a 5-point Likert scale (1=very dissatisfied to 5=very satisfied) to quantify user feedback across 10,000 respondents.

B. Qualitative Approach

• Focus: Meanings, experiences, context.
• Methods: Interviews, focus groups, case studies.
• Strengths:
  • Rich, contextual insights.
  • Explores "why" behind behaviors.
• Weaknesses:
  • Subjective, hard to generalize.
  • Time-consuming.
• Example: Pathao’s driver satisfaction study might interview 100 drivers to understand their work-life balance challenges, revealing themes like:
  • "Unpredictable fares make planning difficult."
  • "Lack of rest breaks increases stress."

C. Mixed Methods Approach

• Focus: Combines quantitative + qualitative for robustness.
• Example: eSewa’s fraud detection study:
  1. Quantitative: Analyze transaction data to find fraud patterns (e.g., high-frequency small transactions).
  2. Qualitative: Interview fraud victims to understand motivations (e.g., "I was tricked by a fake seller").

4. Research Philosophies: Positivism vs. Interpretivism

Philosophies shape how research questions are framed and answered.

A. Positivism

• Assumptions:
  • Reality is objective and measurable.
  • Research should be value-free.
• Approach: Quantitative, experimental.
• Example: NEPSE’s stock market analysis assumes:
  • Stock prices are influenced by factual data (e.g., earnings reports, macroeconomic indicators).
  • Researchers use statistical models to predict trends.

B. Interpretivism

• Assumptions:
  • Reality is subjective and shaped by context.
  • Researcher’s perspective matters.
• Approach: Qualitative, exploratory.
• Example: Khalti’s payment delay study might use interviews with merchants to understand:
  • "Why do you think delays happen?" → "Bank processing is slow during peak hours."
  • "How does this affect your business?" → "I lose repeat customers."

5. Deductive vs. Inductive Approaches

These approaches determine how research moves from theory to data (or vice versa).

A. Deductive Approach

• Flow: Theory → Hypothesis → Data → Conclusion.
• Example: Bank loan approval model:
  1. Theory: "Borrowers with higher credit scores default less."
  2. Hypothesis: "If credit score > 700, default rate will be <5%."
  3. Data: Analyze 1,000 loan records at Nabil Bank.
  4. Conclusion: "Hypothesis confirmed: 4% default rate for score > 700."

B. Inductive Approach

• Flow: Data → Patterns → Theory.
• Example: Daraz’s pricing strategy:
  1. Data: Collect sales data for 100 products over 6 months.
  2. Patterns: Notice that "discounts >30% increase repeat purchases."
  3. Theory: "Daraz should offer 30% discounts during off-peak seasons."

flowchart TD
    subgraph Deductive["Deductive"]
        A["Theory"] --> B["Hypothesis"]
        B --> C["Data Collection"]
        C --> D["Test Hypothesis"]
    end
    subgraph Inductive["Inductive"]
        E["Data Collection"] --> F["Find Patterns"]
        F --> G["Develop Theory"]
    end

6. Applied vs. Fundamental Research

Research can be classified by its purpose:

In the Real World

1. eSewa’s Fraud Detection:

   • Idea Used: Mixed methods (quantitative + qualitative).
   • How: Combines transaction data analysis (quantitative) to detect anomalies with user interviews (qualitative) to understand fraudster motivations.

2. Pathao’s Driver Retention Program:

   • Idea Used: Exploratory design + interpretivism.
   • How: Initially conducted driver interviews (qualitative) to identify pain points (e.g., irregular payments), then designed a flexible pay structure based on findings.

3. NEPSE’s Stock Market Predictions:

   • Idea Used: Positivism + explanatory design.
   • How: Uses statistical models (quantitative) to predict stock trends based on objective data (e.g., GDP growth, interest rates), assuming markets react predictably to these factors.

Worked Example: Daraz’s Order Queue System

Scenario: Daraz wants to reduce delivery delays during peak seasons (e.g., Dashain).

Research Design:

1. Problem: "How can Daraz optimize order processing during high demand?"
2. Approach: Explanatory design (cause-effect) + quantitative (survey data).
3. Methods:
   • Survey 5,000 customers on preferred delivery times.
   • Test two queue systems:
     • System A: First-come-first-served.
     • System B: Priority based on order value (higher-value orders first).
4. Data Analysis:
   • System A: 30% of orders delayed >24 hours.
   • System B: 15% delayed (15% improvement).
5. Conclusion:
   • Priority-based queuing reduces delays by prioritizing high-value orders, which are more likely to be urgent.

Exam Tip

1. Compare and Contrast:

   • Always expect questions like "Compare quantitative and qualitative research designs." Use a table (as shown above) to highlight differences in methods, strengths, and examples.

2. Apply to Real Scenarios:

   • For case studies (e.g., ABC Bank), map the problem to a design type (e.g., "This is a causal-comparative study because it compares satisfied vs. dissatisfied customers").
   • Use Nepali examples (eSewa, Pathao, NEPSE) to show understanding.

3. Philosophy Questions:

   • If asked about positivism vs. interpretivism, link to methods:
     • "Positivism would use surveys to quantify customer satisfaction at Nabil Bank, while interpretivism would interview branch managers to understand internal processes."

4. Design Flowcharts:

   • Sketch a simple flowchart for any research design question (e.g., "How would you design a study on why Nepali students prefer online learning?").

5. Avoid Common Mistakes:

   • Don’t confuse "exploratory" with "descriptive": Exploratory is about discovering (e.g., "What are the reasons?"), while descriptive is about listing (e.g., "What are the reasons?" with predefined categories).
   • Don’t mix up deductive and inductive: Deductive starts with a theory; inductive builds one from data.

Final Note: Always connect theory to practice. Examiners love it when you say: "For example, if Ncell wanted to reduce customer complaints, they could use a descriptive design with a quantitative survey to quantify issues, followed by a qualitative interview to understand root causes—this is a mixed methods approach combining both philosophies."