Research Design & Approaches: Types, Criteria & Applications

---

Research Design: The Blueprint of Your Study

Research design is the structured framework that guides how you collect, analyze, and interpret data to answer your research question. It ensures your study is logical, valid, and replicable. Without a clear design, your research risks being unfocused, biased, or useless—like building a house without a blueprint.

Why Does Design Matter?

• Clarity: Defines what you will study and how.
• Validity: Ensures your findings are truthful (not misleading).
• Efficiency: Saves time and resources by avoiding wasted efforts.
• Replicability: Allows others to repeat your study and verify results.

---

Types of Research Designs

Research designs are categorized based on purpose, timing, and control. Here’s a breakdown with real-world examples from Nepal:

1. Exploratory Design

Purpose: To explore a problem when little is known (e.g., "Why do customers abandon Daraz carts?"). Methods:

• Secondary data (existing reports)
• Pilot studies
• Case studies
• Focus group discussions

Example:

• Daraz might use exploratory design to understand why users leave items in their cart. They could analyze past customer reviews or conduct focus groups with shoppers.

2. Descriptive Design

Purpose: To describe characteristics of a population or situation (e.g., "What is the average income of Kathmandu’s freelancers?"). Methods:

• Surveys
• Observations
• Case studies

Example:

• Nepal Rastra Bank (NRB) uses descriptive design to publish quarterly reports on inflation, unemployment, and GDP growth. These reports describe the current economic state but do not explain why trends occur.

3. Causal (Experimental) Design

Purpose: To test cause-and-effect relationships (e.g., "Does increasing ad spend on Facebook boost Nabil Bank’s loan applications?"). Methods:

• Controlled experiments
• Field experiments
• Quasi-experimental designs

Example:

• Ncell might run an A/B test: Group A sees ads for a new data plan, while Group B does not. If Group A’s data usage increases, Ncell can prove the ad caused the change.

---

Figure 1: Types of Research Designs and Their Methods

---

Criteria for a Good Research Design

A strong design must meet these five key criteria:

---

Deductive vs. Inductive Approaches

Research can follow two logical paths:

Worked Example (Deductive Approach): Problem: Does advertising on YouTube increase Pathao’s ride bookings? Hypothesis: "If Pathao runs video ads, then ride bookings will rise by 15%." Test: Run ads for 3 months, track bookings. Result: If bookings rise, the hypothesis is supported.

Worked Example (Inductive Approach): Observation: "Most Daraz customers who abandon carts do so after seeing shipping costs." Pattern: "High shipping costs = cart abandonment." Theory: "Simplifying checkout (e.g., free shipping over ₹500) reduces abandonment."

---

Research Design in Action: A Case Study

Company: Himalayan Java (Nepal’s largest coffee brand) Problem: Declining sales in Pokhara. Research Design Used:

1. Exploratory: Conducted focus groups with Pokhara customers to understand preferences.
2. Descriptive: Surveyed 500 customers on coffee habits (e.g., "Do you prefer instant or brewed coffee?").
3. Causal: Tested a discount campaign in half of Pokhara stores to see if sales increased.

Findings:

• Customers preferred brewed coffee but found it expensive.
• After a 20% discount, sales in test stores rose by 25%. Action: Himalayan Java reduced prices and expanded brewed coffee options in Pokhara.

---

In the Real World

1. eSewa’s Fraud Detection

   • Design Used: Causal (Experimental)
   • How? eSewa tests new fraud alerts (e.g., SMS notifications for unusual transactions) in a pilot group before rolling them out nationwide. If fraud drops, they scale the solution.

2. Ncell’s Customer Satisfaction Surveys

   • Design Used: Descriptive
   • How? Ncell sends monthly surveys to 1,000 users to track network quality, billing issues, and service speed. Results help them prioritize improvements (e.g., upgrading towers in low-signal areas).

3. Daraz’s A/B Testing for Product Pages

   • Design Used: Causal (Experimental)
   • How? Daraz tests two versions of a product page (e.g., red "Buy Now" button vs. green). If Version A gets more clicks, they keep it globally.

4. Nepal Rastra Bank’s Inflation Reports

   • Design Used: Descriptive
   • How? NRB collects monthly data on prices of 500+ items (rice, fuel, vegetables) to describe inflation trends. Policymakers use this to adjust interest rates.

5. NEPSE’s Market Trend Analysis

   • Design Used: Exploratory → Descriptive
   • How? NEPSE first explores why stock prices fluctuate (e.g., global oil prices, political news). Then, it describes trends (e.g., "Microfinance stocks rose 10% this quarter") to guide investors.

---

Common Mistakes in Research Design

Even experienced researchers make errors. Here are five pitfalls to avoid:

1. Poor Sampling

   • Mistake: Surveying only Kathmandu university students to generalize about all Nepali youth.
   • Fix: Use stratified sampling (e.g., students, professionals, farmers).

2. Leading Questions

   • Mistake: Asking, "Don’t you think NTC’s service is terrible?"
   • Fix: Use neutral questions: "How would you rate NTC’s service on a scale of 1-10?"

3. Ignoring External Factors

   • Mistake: Blaming low Khalti transactions only on poor marketing, ignoring COVID-19 lockdowns.
   • Fix: Control for external variables (e.g., economic crises, holidays).

4. Overgeneralizing from Small Data

   • Mistake: Concluding "All Nepalis hate Pathao" based on 10 complaints.
   • Fix: Use large, representative samples.

5. Not Piloting the Study

   • Mistake: Launching a 50-question survey without testing it first.
   • Fix: Run a pilot test with 5-10 people to check for confusing questions.

---

Research Design Process: Step-by-Step

Figure 2: The Research Design Process

---

Exam Tip

This unit is heavily tested in TU exams. Here’s how to score full marks:

1. Definition + Classification (50% of questions)

• Always start with a precise definition (e.g., "Research design is a structured plan that outlines how data will be collected, analyzed, and interpreted to answer a research question.").
• Classify designs clearly (e.g., "Exploratory, descriptive, and causal are the three main types based on purpose.").
• Use examples from Nepal (e.g., NRB for descriptive, Daraz for causal).

2. Criteria Questions (Common in Short Notes)

• List all 5 criteria (validity, reliability, generalizability, objectivity, feasibility).
• Explain with a real example (e.g., "For generalizability, a study on Pokhara traffic should not only survey Thapathali; it must include Lalitpur and Kageshwari too.").

3. Deductive vs. Inductive (Tricky but High Marks)

• Deductive: "Theory → Hypothesis → Test → Conclusion" (e.g., Nabil Bank’s loan interest study).
• Inductive: "Observations → Pattern → Theory" (e.g., eSewa’s fraud detection).
• Compare in a table (as shown above) to stand out.

4. Case Study Questions (Long Answer)

• Structure your answer like this:
  1. Problem (e.g., Himalayan Java’s declining sales).
  2. Design Used (exploratory → descriptive → causal).
  3. Methods (focus groups, surveys, A/B testing).
  4. Findings & Actions (e.g., "Discounts increased sales by 25%").

5. Avoid These Mistakes

• ❌ Vague answers: Don’t say "Research design is important." → Say how (e.g., "It ensures validity by controlling biases like leading questions.").
• ❌ Ignoring Nepal examples: Examiners love real-world ties (e.g., Ncell, Daraz, NRB).
• ❌ Mixing up deductive/inductive: Always link to theory → data (deductive) or data → theory (inductive).

---

Quick Revision Checklist

✅ Can you define research design and list its three main types? ✅ Do you know the 5 criteria for a good design and one Nepal example for each? ✅ Can you distinguish deductive vs. inductive with a business case? ✅ Are you ready to write a 10-mark case study on a Nepali company (e.g., Daraz, Ncell, NRB)? ✅ Do you know how to avoid common design errors (sampling, leading questions, etc.)?

---

Final Thought:

"A research design is like a recipe—if you skip ingredients (methods) or steps (sampling), your dish (findings) will be ruined. Master this unit, and you’ll never write a weak research proposal again!"