Data Collection Methods & Tools: Primary vs. Secondary

TAKEAWAYS:

• Primary data (e.g., surveys, interviews) is collected firsthand for your specific research, while secondary data (e.g., government reports, eSewa transaction logs) is pre-existing but must be critically evaluated.
• Observation (e.g., studying Pathao driver routes) is unobtrusive but risks bias; experiments (e.g., testing Khalti’s payment speed) require controlled variables.
• Questionnaires (structured or unstructured) are versatile but suffer from low response rates; focus groups (e.g., Daraz’s customer panels) reveal group dynamics but may lack anonymity.
• Sampling techniques (random, stratified, snowball) determine data representativeness—critical for Ncell’s market surveys or NEPSE’s stock trend analysis.
• Ethical tools (e.g., informed consent forms) protect participants, while unethical practices (e.g., fetal abnormality research without consent) violate human rights.
• Data presentation (tables, charts) must be clear—NTC’s traffic flow reports use pie charts for proportions and line graphs for trends over time.

1. Primary vs. Secondary Data: The Core Divide

Researchers collect data in two broad ways:

• Primary data: Original data gathered directly for your study (e.g., surveying Ncell customers about their billing preferences).
• Secondary data: Existing data from other sources (e.g., NTC’s annual report on mobile penetration rates).

mindmap
  root((Data Collection Methods))
    Primary Data
      - Surveys
      - Interviews
      - Observation
      - Experiments
    Secondary Data
      - Government Reports
      - Academic Journals
      - Corporate Databases (e.g., NEPSE)
      - Digital Archives (e.g., eSewa transaction logs)

Why it matters:

• Primary data is tailored to your research but time-consuming and costly.
• Secondary data is quick and cheap but may lack relevance or be outdated.

Worked Example: NEPSE’s stock market analysis relies on secondary data (historical price trends, company filings) to predict future movements. However, a study on investor sentiment during elections would need primary data (surveys of traders via WhatsApp polls).

2. Primary Data Collection Methods

A. Surveys and Questionnaires

Definition: Structured or unstructured questions to gather opinions or facts. Tools:

• Closed-ended (multiple-choice, Likert scales).
• Open-ended (free-response).

Advantages:

• Standardized responses for easy comparison.
• Can reach large samples (e.g., Daraz’s customer satisfaction survey).

Disadvantages:

• Low response rates (e.g., only 20% of Pathao riders reply to feedback forms).
• Social desirability bias (respondents lie about usage, e.g., "I use eSewa daily" when they don’t).

Example: A Likert scale (1–5) in a Khalti payment survey might ask: "How satisfied are you with Khalti’s transaction speed?" 1 (Very Dissatisfied) → 5 (Very Satisfied).

Visual:

flowchart TD
    A["Survey Question"] --> B["Closed-Ended (✓)"] --> C["Easy to Analyze"]
    A --> D["Open-Ended (____)"] --> E["Rich Insights but Hard to Code"]

B. Interviews

Types:

• Face-to-face (e.g., bank loan officers interviewing Nabil Bank customers).
• Telephone/Online (e.g., Ncell conducting usability tests via WhatsApp Voice Calls).

Advantages:

• Deep insights (e.g., why customers switch from NTC to Ncell).
• Flexibility to probe answers.

Disadvantages:

• Time-consuming and expensive.
• Interviewer bias (e.g., leading questions like "Don’t you think eSewa is slow?").

Worked Example: Chaudhary Group’s market expansion research used in-depth interviews with Himalayan Java farmers to understand supply chain bottlenecks.

C. Observation

Definition: Watching behavior without intervention (e.g., studying Kathmandu traffic patterns at Ring Road intersections).

Types:

• Participant observation (e.g., a researcher joins a Daraz warehouse team).
• Non-participant observation (e.g., filming Pathao drivers’ routes).

Advantages:

• Captures real behavior (e.g., how often Ncell users check data balances).
• Useful for sensitive topics (e.g., workplace productivity).

Disadvantages:

• Time-consuming.
• Observer effect (people act differently when watched).

Example: NTC’s traffic congestion study used hidden cameras at major junctions to analyze vehicle flow.

D. Experiments

Definition: Manipulating variables to test cause-and-effect (e.g., testing whether a 10% discount on Daraz Prime membership increases sign-ups).

Key Components:

• Independent variable (e.g., discount amount).
• Dependent variable (e.g., membership sign-ups).
• Control group (no discount).

Advantages:

• High internal validity (proves causation).
• Used in A/B testing (e.g., Google’s ad placement experiments).

Disadvantages:

• Artificial settings (e.g., lab-like surveys vs. real-world Khalti transactions).
• Ethical concerns (e.g., withholding a discount from a control group).

3. Secondary Data Collection Methods

Sources:

• Government databases (e.g., Nepal Rastra Bank’s inflation reports).
• Corporate reports (e.g., NEPSE’s annual reports).
• Digital archives (e.g., eSewa’s transaction logs).
• Academic journals (e.g., studies on digital payment adoption).

Advantages:

• Cost-effective and time-saving.
• Large datasets (e.g., NTC’s call detail records for network analysis).

Disadvantages:

• Outdated (e.g., 2019 data for 2024 trends).
• Bias (e.g., Ncell’s reports may overstate customer satisfaction).
• Lack of context (e.g., why a Daraz product was discontinued).

Worked Example: Analyzing Nepal’s GDP growth uses secondary data from the Central Bureau of Statistics, but a study on digital payment adoption would need primary data (surveys of eSewa/Khalti users).

Comparison Table:

4. Choosing the Right Method: A Decision Framework

flowchart TD
    A["Research Question"] --> B{"Is data pre-existing?"}
    B -->|"Yes"| C["Use Secondary Data\n(e.g., NEPSE reports)"]
    B -->|"No"| D{"Can you observe behavior?"}
    D -->|"Yes"| E["Use Observation\n(e.g., Pathao routes)"]
    D -->|"No"| F{"Is causation needed?"}
    F -->|"Yes"| G["Use Experiment\n(e.g., Daraz discount test)"]
    F -->|"No"| H["Use Survey/Interview\n(e.g., Khalti feedback)"]

When to Use Observation:

• Circumstances: Studying natural behavior (e.g., how often Ncell users top up at 3 AM).
• Justification: Unobtrusive, captures real-time actions (e.g., traffic patterns at Thapathali).

Unethical Scenario Analysis: Case: Students plan to research fetal abnormalities via ultrasound scans without pregnant women’s consent. Issues:

1. Violation of autonomy (participants cannot refuse).
2. Risk of harm (emotional distress).
3. Lack of informed consent.

Ethical Fix:

• Obtain written consent from participants.
• Ensure confidentiality (anonymize data).
• Use ethics review boards (like TU’s IRB).

5. Data Collection Tools: A Practical Guide

Worked Example: Pathao’s driver performance review uses:

1. Primary data: GPS logs (observation) + driver surveys (questionnaires).
2. Secondary data: NTC’s traffic reports (to correlate delays).

6. Real-World Applications

In the Real World

1. eSewa’s User Feedback Loop:

   • Tool: Online questionnaires (primary data) + transaction logs (secondary data).
   • Idea: Uses survey data to improve app UX (e.g., adding "Pay Later" option after users complained about transaction fees).

2. NEPSE’s Stock Market Predictions:

   • Tool: Secondary data (historical prices, earnings reports).
   • Idea: Applies time-series analysis to forecast trends, but supplements with primary data (investor sentiment polls).

3. Daraz’s Inventory Management:

   • Tool: Observation (warehouse scans) + customer reviews (primary data).
   • Idea: Uses real-time stock levels to auto-reorder popular items (e.g., Himalayan Java coffee).

Case Study: Nabil Bank’s Loan Approval Process

flowchart TD
    A["Customer Applies"] --> B["Primary Data: Credit Score\n(Collected via interview/questionnaire)"]
    B --> C["Secondary Data: Market Trends\n(Nepal Rastra Bank reports)"]
    C --> D["Decision: Approve/Reject\n(Based on risk analysis)"]
    D --> E["Ethical Check: No Discrimination\n(Complies with banking laws)"]

• Primary Data: Customer’s income, employment history (survey).
• Secondary Data: Current interest rates, economic forecasts.
• Ethical Consideration: No bias against gender/region (Nepal’s banking law requires fairness).

7. Data Presentation: Tables and Beyond

Why it matters: Raw data is useless; tables and charts reveal patterns. Rules:

• Tables: Use for precise numbers (e.g., Ncell’s monthly data usage).
• Charts:
  • Bar graphs: Compare categories (e.g., eSewa vs. Khalti user demographics).
  • Line graphs: Show trends (e.g., NEPSE’s stock prices over 5 years).
  • Pie charts: Proportions (e.g., NTC’s revenue sources).

Example: NTC’s Traffic Flow Data presented as a table (raw) and line graph (trend):

Exam Tip: Always label axes, use clear titles, and avoid 3D charts (they distort data).

Exam Tip: How to Score Full Marks

1. Define clearly:

   • For face-to-face interviews, say:

     "A structured conversation where the researcher asks predefined questions to participants in person, allowing for real-time probing (e.g., Nabil Bank’s loan officer interviews)."

2. Compare methods:

   • For secondary vs. primary data, use a table (as above) and explain trade-offs (cost, relevance, time).

3. Apply to cases:

   • If given a scenario (e.g., fetal abnormality research), highlight ethical violations (consent, harm) and suggest fixes (IRB approval, anonymization).

4. Show real-world links:

   • Tie answers to Nepali companies (e.g., "Khalti uses surveys to improve its app, while NEPSE relies on secondary data for market analysis").

5. Visualize:

   • Draw flowcharts for processes (e.g., data collection steps) or tables for comparisons.

6. Avoid:

   • Vague terms like "important" or "useful"—explain why (e.g., "Observation is unethical for sensitive topics like medical research because it lacks consent").

Final Note: This unit is 50% practical. Expect questions on:

• Choosing the right tool (e.g., "Why would you use observation for Pathao’s driver routes?").
• Ethical dilemmas (e.g., "Is it ethical to use Ncell’s old data for a 2024 study?").
• Presenting data (e.g., "How would you display NEPSE’s stock trends?").

Practice: Recreate the Nabil Bank flowchart or NTC traffic table from memory—this is how you’ll be tested.