Marketing Research Process and Secondary Data Analysis

Course Logistics and Community Engagement

• Activity Two Completion Status: Approximately 12% of the class has completed the first marked activity. The average completion time was recorded as just under 40 minutes, though students can exit and return, which may inflate this average.

• Learning Objectives: The course aims to cover roughly 11 to 15 concepts per week to avoid overwhelming students, selecting only those most essential for project success.

• Complementary Nature of Lectures: Some concepts included in the weekly activities may not be explicitly covered during physical lectures due to time constraints, but they are considered vital complementary information.

• Community Check-in: A specific practice to build "belongingness" through short interactions with peers and positive reinforcement mindset exercises, such as reciting: "I am capable of learning new things today."

Circumstances Where Marketing Research is Unnecessary

Marketing research is not always the optimal decision for a firm. Key factors that preclude research include:

• Inaccessibility of Target Audience: If the target cannot be reached (e.g., top-level managers with packed schedules), research is unfeasible.

  • Proxies: In cases where the target cannot express themselves (e.g., infants), proxies such as parents must be consulted.

• Resource Constraints: When there is a lack of time or money. High-speed decision-making environments often leave no room for proper research.

• Cost Outweighs Value: Research should be avoided if the findings will only serve a minor purpose or if the investment exceeds the potential benefits.

• Nature of the Decision: "Nominal decisions" (e.g., deciding whether to sell in Woolworths vs. Coles) may not require expensive research. However, strategic decisions (e.g., entering a new cultural market like Singapore or China) require thorough investigation.

• Niche Markets: Small, well-defined markets with consistent consumer behavior (e.g., vegans) may be understood well enough without formal research.

• Unpredictable Market Conditions: During systemic crises such as the COVID-19 pandemic or the Middle East war, consumer responses may reflect temporary survival/panic modes rather than long-term trends, rendering data unreliable post-crisis.

The Marketing Research Process

The process consists of 11 steps, with a focus for practical student projects often residing in steps 8 through 11 (analysis and reporting).

Step 1: Identify and Clarify Information Needs

• Problem Definition: Clarifying the core issue is the most critical step. Researchers must perform a "situation analysis" and distinguish between causes and symptoms.

  • Symptoms: Measurable indicators (e.g., declining brand endorsement, interest in digital platforms).

  • Causes: The root driver (e.g., the threat of AI and academic integrity concerns leading to the discouragement of online quizzes).

  • Case Study: Bud Light (Spotlight): The brand observed declining sales among younger generations. They defined the problem as being "old-fashioned" and attempted to modernize via a transgender influencer, Dylan Mulvaney. This resulted in a backlash/boycott because the modernization strategy offended the "core audience" segment (traditionally represented by characters similar to those in the film The Shawshank Redemption).

• Unit of Analysis: Determining who or what is measured (e.g., individual students, a specific unit, or a whole faculty).

• Relevant Variables: Translating concerns (e.g., user experience, learning outcomes, confidence, system glitches) into measurable questions.

Step 2: Determine Research Design and Objectives

• Objectives: These draw the boundaries of the study to prevent budget exhaustion.

• Data Types:

  • Secondary Data: Previously collected data for another purpose ("secondhand").

  • Big Data: Massive, unstructured datasets beyond human processing capacity. Example statistics regarding elasticity (the percentage change in sales resulting from a 1-unit increase in a variable):

    • Distribution elasticity: $0.368$

    • Price elasticity: $-0.4$

    • Line length elasticity: $0.45$

  • Primary Data: Firsthand data collected specifically for current research needs.

Step 3: Research Design Categorization

1. Exploratory Research: Seeks to understand "what" is happening or "why" something might be important. Can be qualitative or quantitative.

2. Conclusive Research:

   • Descriptive: Quantitative focus on "who," "what," and "where."

     • Cross-sectional: A one-off measurement (snapshot).

     • Longitudinal: Surveys the same group continuously over a period to see changes over time.

   • Predictive: Identifies factors that indicate the occurrence of an event. Example: Specific cloud shapes (predictor) indicating a thunderstorm (event), even if the cloud is not the technical cause.

   • Causal: Demonstrates that factor A specifically causes factor B.

Step 4: Sampling and Methodology

• Population vs. Sample: If an entire population is surveyed, it is a Census (e.g., the Australian population census). Usually, researchers use a representative Sample.

• Sampling Methods:

  • Probability Sampling: Every member has a random, equal chance of selection (e.g., names drawn from a hat).

  • Non-probability Sampling: Often based on Convenience (e.g., surveying only the students who attend an 8 AM lecture). This can introduce bias as early-morning attendees may differ fundamentally from those who watch recordings.

Step 5: Data Collection Instruments

• Measurement Types:

  • Single-item: For simple categories (e.g., gender).

  • Multi-item: To capture complex constructs (e.g., "university satisfaction," which includes commute, social life, food, and academics).

• Pretesting: Vital for identifying wording issues, ambiguity, or sequence flaws before the full launch.

• Ethics and Sequence: Screeners (e.g., verifying age is 17+) appear first. Personal demographics typically appear at the end to avoid scaring participants away.

• Tools: Google Forms (simple), Qualtrics (complex branching/logic), and QuestionPro (non-research surveys).

Secondary Data Analysis

Secondary data is widely used to solve problems directly or to refine primary research designs.

Sources of Secondary Data

• Internal Sources: Sales invoices (names, ABN, staff), customer service reports (recorded calls), internal student satisfaction data (individual unit levels).

• External Sources:

  • Popular Sources: Bloomberg, Forbes, Harvard Business Review.

  • Scholarly Sources: Peer-reviewed journals, OneSearch, Google Scholar. Australia uses the ABDC list (rankings of $A^*$, $A$, $B$, $C$, and $D$).

  • Government Sources: Australian Bureau of Statistics (ABS).

  • Syndicated Data: Commercial services where professional companies sell data to merchants (e.g., consumer panels or store audits).

Evaluation Criteria for Secondary Data

Data must be evaluated on:

• Purpose (why was it collected?)

• Accuracy and Consistency

• Credibility and Methodology

• Potential Bias (e.g., leading questions like "Do you agree this is a great product?" vs. "Is this product bad?").

Principles of Hypothesis Testing

Core Definitions

• Variable: Concrete and observable (e.g., height, weight).

• Construct: Abstract or blurred concepts (e.g., service quality). Service quality often includes five dimensions: Reliability, Assurance, Tangibles, Empathy, and Responsiveness.

• Independent Variable (IV): The factor that predicts or precedes the change.

• Dependent Variable (DV): The outcome that depends on the state of the IV.

Relationships and Hypotheses

• Directional Relationships:

  • Positive Relationship: Variables move in the same direction ($+$).

  • Negative Relationship: Variables move in opposite directions ($-$).

• Null Hypothesis ($H_0$): States there is no relationship or no difference. We aim to reject the null hypothesis based on a sample.

• Alternative Hypothesis ($H_1$): States that a relationship or difference exists.

• Tail Types:

  • One-tail: Predictive/Directional (e.g., "A is larger than B").

  • Two-tail: General difference (e.g., "A is different from B").

Errors in Testing

We specify a Significance Level ($\alpha$), traditionally set at $0.05$.

• Note: An extremely low significance level decreases the chance of Type I errors but increases the chance of Type II errors.