Variables in Research: Types, Roles & Measurement

TAKEAWAYS:

• Variables are the measurable traits or characteristics that researchers manipulate, measure, or control in a study.
• Independent variables are the "causes" (e.g., advertising spend), while dependent variables are the "effects" (e.g., sales revenue).
• Extraneous variables (e.g., competitor actions) can distort results unless controlled or measured.
• Measurement scales (nominal, ordinal, interval, ratio) determine how data can be analyzed statistically.
• Real-world examples like eSewa’s transaction success rates (dependent variable) influenced by user interface design (independent variable) illustrate variable relationships.
• Ethical considerations arise when manipulating variables (e.g., A/B testing in ads must avoid deception).

1. What Are Variables?

Variables are the fundamental units of research—they represent the elements that change or are measured in a study. Without variables, research would lack direction or measurable outcomes.

Types of Variables

Variables are classified based on their role in the research process. The three primary types are:

mindmap
  root((Variables in Research))
    Independent Variable
      "Cause (e.g., training program, ad campaign)"
      "Manipulated by researcher"
    Dependent Variable
      "Effect (e.g., employee productivity, sales)"
      "Measured for change"
    Extraneous Variable
      "Confounding factor (e.g., weather, competitor actions)"
      "Must be controlled or measured"

Worked Example: Daraz’s Delivery Time Study

• Independent Variable (IV): Delivery service type (standard vs. express).
• Dependent Variable (DV): Customer satisfaction scores.
• Extraneous Variable (EV): Traffic conditions in Kathmandu (must be accounted for).

2. Independent vs. Dependent Variables

3. Extraneous Variables: The Hidden Distorters

Extraneous variables are unwanted influences that can skew research results if ignored. They can be:

• Controlled (e.g., holding store location constant in a sales study).
• Measured (e.g., recording temperature in a product testing experiment).
• Randomized (e.g., assigning participants randomly to groups to balance effects).

Real-World Example: NTC’s Internet Speed Study

• IV: Type of internet plan (fiber vs. DSL).
• DV: User-reported speed.
• EV: Distance from NTC exchange (must be controlled or measured to avoid bias).

How to Handle Extraneous Variables?

flowchart TD
  A["Identify Extraneous Variables"] --> B["Decide: Control/Measure/Randomize"]
  B --> C["Control: Keep constant (e.g., same time of day, same testing location)"]
  B --> D["Measure: Record as additional data (e.g., distance from NTC exchange)"]
  B --> E["Randomize: Use random assignment (e.g., random sample of users)"]
  C --> F["Reduce Confounding Effect"]
  D --> F
  E --> F
  F --> G["Ensure Valid Results"]

4. Measurement Scales: How Data is Quantified

Variables must be measured using appropriate scales of measurement, which determine the statistical analyses that can be applied. The four scales are:

Worked Example: Kathmandu Traffic Congestion Study

• Variable: Traffic density (vehicles per hour).
• Scale: Ratio (zero means "no traffic"; intervals are equal).
• Analysis: Can calculate the mean traffic density and compare it across different times.

5. Variables in Real-World Business Research

Case Study 1: eSewa’s Transaction Success Rates

• IV: Payment method (credit card vs. mobile wallet).
• DV: Transaction success rate (%).
• EV: Network connectivity (must be controlled for).
• Measurement Scale:
  • IV: Nominal (categorical).
  • DV: Ratio (0% to 100%).
• Finding: Mobile wallets had a 15% higher success rate than credit cards, controlling for network issues.

Case Study 2: Nabil Bank’s Loan Default Prediction

• IV: Customer credit score (continuous).
• DV: Loan default (Yes/No).
• EV: Economic conditions (measured via inflation rate).
• Measurement Scale:
  • IV: Ratio (credit score ranges from 300–850).
  • DV: Nominal (binary outcome).
• Analysis: Used logistic regression (appropriate for nominal DV) to predict defaults.

6. Ethical Considerations with Variables

Manipulating variables raises ethical questions, especially in human subjects research:

• Deception: A/B testing ads without user consent can be unethical.
• Harm: Testing a new drug (IV) on participants without clear benefits may cause harm.
• Informed Consent: Participants must know if they’re in the control or experimental group.

Example: Pathao’s Ride Pricing Experiment

• IV: Dynamic pricing algorithm (surge pricing).
• DV: Rider satisfaction.
• Ethical Issue: Riders were not informed about the experiment, leading to complaints.
• Solution: Pathao later disclosed the test and offered discounts to affected riders.

7. Common Mistakes in Variable Handling

Students often make these errors in exams and research projects:

1. Confusing IV and DV: Treating the effect as the cause (e.g., saying "high sales cause more ads" instead of "ads cause high sales").
2. Ignoring Extraneous Variables: Forgetting to control for factors like seasonality in sales data.
3. Wrong Measurement Scale: Using an ordinal scale (e.g., Likert ratings) for parametric tests (e.g., t-tests).
4. Overlooking Ethics: Manipulating variables without considering participant welfare.

8. How Variables Are Used in Research Designs

Variables are the backbone of quantitative research designs:

• Experimental Design: IV is manipulated; DV is measured (e.g., testing a new training program’s effect on productivity).
• Correlational Design: Relationships between variables are studied (e.g., does social media use correlate with brand loyalty?).
• Survey Design: Variables are measured via questionnaires (e.g., measuring job satisfaction as a DV).

Mermaid Diagram: Research Designs and Variables

flowchart LR
  A["Research Design"] --> B["Experimental"]
  A --> C["Correlational"]
  A --> D["Survey"]
  B --> E[IV Manipulated
DV Measured
(e.g., training program → productivity)]
  C --> F[Relationship
Between Variables
(e.g., social media use ↔ brand loyalty)]
  D --> G[Variables Measured
Via Surveys
(e.g., job satisfaction → satisfaction scale)]
  E --> H[Cause-Effect
Inference]
  F --> I[No Manipulation,
Just Observation]
  G --> J[Descriptive
or Predictive]

In the Real World

1. eSewa’s Fraud Detection System

   • IV: Transaction behavior (e.g., unusual login times).
   • DV: Fraud flag (Yes/No).
   • How it works: Uses ratio-scale data (transaction amounts) and nominal outcomes (fraud/no fraud) to train machine learning models. Extraneous variables like network delays are filtered out.

2. Daraz’s Order Fulfillment

   • IV: Warehouse location (central vs. peripheral).
   • DV: Delivery time (hours).
   • EV: Traffic conditions (measured via GPS data).
   • Insight: Central warehouses reduced delivery time by 24% on average, controlling for traffic.

3. Nepal Rastra Bank’s Inflation Reports

   • IV: Monetary policy changes (e.g., interest rate hikes).
   • DV: Inflation rate (%).
   • EV: Global oil prices (measured as a confounder).
   • Scale: Both IV and DV are ratio (continuous numerical data), allowing for regression analysis.

Exam Tip

1. Always Define Variables Clearly:

   • In exam questions like "Analyze the determinants of customer satisfaction at ABC Bank," explicitly state:
     • IV: Determinants (e.g., staff training, branch location).
     • DV: Customer satisfaction score.
     • EV: Economic conditions (controlled for).

2. Use Real-World Examples:

   • Link variables to Nepali businesses (e.g., "In Pathao’s case, the IV was ride pricing, and the DV was rider complaints").

3. Watch for Measurement Scale Questions:

   • If asked to "discuss how variables are measured," describe the four scales and give a business example for each (e.g., "Brand loyalty (nominal), employee performance ratings (ordinal), temperature in a cold chain (interval), revenue (ratio)").

4. Ethics is a Common Trap:

   • For questions on "ethical issues in business research," mention:
     • Variable manipulation without consent (e.g., A/B testing ads).
     • Harm to participants (e.g., stress from high-stakes experiments).
     • Confidentiality (e.g., leaking customer data in surveys).

5. Practice Variable Classification:

   • Given a scenario, classify variables as IV, DV, or EV. Example:
     • "A study on how study hours affect exam scores."
       • IV: Study hours (ratio).
       • DV: Exam scores (ratio).
       • EV: Teacher quality (must be controlled).

Final Note: Variables are the DNA of research—they define what you measure, manipulate, and analyze. Mastering their classification, measurement, and ethical handling will set you apart in exams and real-world research projects. Always tie your answers to Nepali business cases (e.g., banks, e-commerce, telecom) to score full marks!