Data Collection & Presentation: Methods, Tables & Charts

TAKEAWAYS:

• Data collection must align with research objectives (surveys, experiments, or secondary sources) to ensure accuracy and relevance.
• Frequency tables organize raw data into meaningful categories, revealing patterns (e.g., age groups of Daraz customers).
• Graphs (bar, pie, histograms) visually communicate trends—e.g., NTC’s monthly call-drop rates or Pathao’s ride demand spikes.
• Misleading charts exploit axes, colors, or scales; always check data sources (e.g., NEPSE stock trends).
• Ethics matter: Privacy laws (e.g., Nepal’s 2018 Data Protection Act) restrict how companies like Khalti collect user data.
• Tech tools (Excel, Python’s pandas, Google Sheets) automate data cleaning and visualization for business reports.

1. Data Collection: Methods and Tools

Data is the raw material of statistics. Poor collection leads to flawed conclusions—like a Daraz survey on customer satisfaction using only urban respondents, ignoring rural preferences.

Primary vs. Secondary Data

How to Collect Primary Data:

1. Surveys/Questionnaires

   • Example: A bank like NMB collects loan repayment data via structured questionnaires.
   • Design tip: Use Likert scales (e.g., "1=Strongly Disagree" to "5=Strongly Agree") for subjective questions. Likert scale example for a Khalti user satisfaction survey:

     Q: How satisfied are you with Khalti’s customer service?
     1 (Very Dissatisfied) 2 3 4 5 (Very Satisfied)
     

2. Experiments

   • Example: Daraz tests a "Buy 1 Get 1 Free" promo on 1,000 users to measure sales impact.
   • Key: Randomize groups (control vs. test) to isolate variables.

3. Observation

   • Example: Counting foot traffic at a Kathmandu mall to predict sales (used by brands like Himalayan Xtreme).

4. Sampling Techniques

   • Simple Random: Every Ncell subscriber has an equal chance of being polled.
   • Stratified: Divide Daraz customers by region (Kathmandu, Pokhara, rural) for balanced insights.
   • Cluster: Survey all Pathao drivers in Chitwan district (if studying rural delivery challenges).

Worked Example 1: Sampling for a NEPSE Study Problem: A researcher wants to study investor behavior but can’t survey all 200,000+ NEPSE traders. How to sample? Solution: Use stratified random sampling by investor type (retail, institutional, foreign) and budget allocation.

graph TD
    A["Total NEPSE Traders\n(200,000)"] --> B["Stratum 1: Retail\n(150,000)"]
    A --> C["Stratum 2: Institutional\n(30,000)"]
    A --> D["Stratum 3: Foreign\n(20,000)"]
    B --> E["Sample: 1,500\n(Randomly selected)"]
    C --> F["Sample: 300"]
    D --> G["Sample: 200"]
    E --> H["Analyze\nTrends"]
    F --> H
    G --> H

2. Data Presentation: Tables and Graphs

Raw data is useless without organization. Presentation methods depend on the data type (categorical, numerical) and audience (e.g., NTC’s board vs. Pathao drivers).

A. Frequency Tables

Convert raw data into counts/frequencies. Example: Ages of 50 Daraz delivery agents. Raw data: 22, 25, 28, 30, 22, 24, 26, 29, 31, 23, ... (50 values) Frequency table:

Key terms:

• Class interval: Age groups (e.g., 20–24). Rule: Keep intervals equal (e.g., 5 years).
• Midpoint: (20+24)/2 = 22 (used in calculations like mean).

B. Graphs for Different Data Types

Worked Example 2: Histogram for NTC Call-Drops Data: Monthly call-drop percentages for 12 districts (assume values: 2.1, 1.8, 3.5, 2.9, ...). Steps:

1. Choose class intervals (e.g., 0–1, 1–2, 2–3, 3–4).
2. Count frequencies (e.g., 0–1: 3 districts).
3. Plot bars with height = frequency (or frequency density if intervals vary).

Real-World Tie-In: NTC uses histograms to identify districts with persistent call drops (e.g., >3%) and allocate repair crews. Example: If the 3–4% bin has 2 districts, NTC might investigate those first.

3. Common Pitfalls and Ethical Considerations

A. Misleading Graphs

Example: A Daraz ad shows "Sales ↑ 200%!" but uses a broken y-axis (starts at 100 instead of 0). Correct vs. Incorrect:

B. Ethical Data Collection

• Privacy: Nepal’s Data Protection Act (2018) requires consent for user data (e.g., Khalti’s transaction history).
• Bias: Avoid leading questions. Bad: "Don’t you hate Pathao’s late drivers?" Good: "How often do you experience delays with Pathao?"
• Transparency: Cite sources. Example: A study on Ncell’s network quality must disclose if data was provided by Ncell or an independent lab.

4. Tools for Data Presentation

Worked Example 3: Excel for a Bank Loan Analysis Task: Present loan default rates by region for a Nepalese bank.

1. Raw data:

   Region	Defaults	Total Loans
   Kathmandu	45	1,200
   Pokhara	12	800
   Chitwan	8	600

2. Excel steps:
   • Insert column chart (Defaults vs. Region).
   • Add data labels (% defaults).
   • Use conditional formatting to highlight >5% defaults (red).

Output:

In the Real World

1. eSewa’s Transaction Data

   • Idea: Frequency tables and histograms analyze how often users pay bills vs. transfer money.
   • Example: A histogram of transaction amounts reveals most users spend ₹1,000–₹5,000/month, helping eSewa design targeted promotions.

2. Daraz’s Inventory Management

   • Idea: Stratified sampling surveys customers in Kathmandu, Pokhara, and rural areas to stock products like solar panels (high demand in rural zones) vs. smartphones (urban).

3. NTC’s Network Optimization

   • Idea: Line graphs track call-drop rates over time to identify seasonal spikes (e.g., monsoon season in Chitwan).
   • Action: NTC uses this to deploy temporary towers in high-drop zones.

4. Pathao’s Driver Scheduling

   • Idea: Bar charts compare ride demand by hour (e.g., 7–9 AM vs. 12–2 PM) to optimize driver shifts.

Exam Tip

1. For short-answer questions:

   • Define terms precisely. Example:

     "Distinguish between primary and secondary data with a Nepalese business example." Answer: "Primary data is collected firsthand (e.g., eSewa surveying 1,000 users on app satisfaction), while secondary data uses existing sources (e.g., NTC’s published call-drop reports). Primary ensures relevance but is costly; secondary is faster but may lack specificity."

2. For numerical problems:

   • Show all steps for frequency tables/histograms. Example:

     "Given the data: 12, 15, 14, 18, 20, 16, construct a frequency table with class intervals 10–14, 15–19, 20+." Solution:

     Class Interval	Frequency
     10–14	2
     15–19	3
     20+	1

3. For graph interpretation:

   • Describe trends, not just shapes. Example:

     "Analyze this histogram of Khalti transaction amounts." Answer: "The data is right-skewed (most transactions are small, with a few large outliers). The modal class (highest bar) is ₹500–₹1,000, suggesting Khalti’s core users make frequent small payments (e.g., bill splits)."

4. Common exam traps:

   • Avoid: Saying a pie chart is best for trends (use line graphs for time series).
   • Do: Critique misleading graphs (e.g., "This bar chart exaggerates growth because the y-axis starts at 50% instead of 0.").

Final Checklist Before Submission:

• Did I label all axes in graphs?
• Did I cite real Nepalese examples (eSewa, NTC, etc.)?
• Did I compare methods (e.g., bar vs. pie charts) with pros/cons?
• Did I show calculations for frequency tables?