Marketing Research Design and Measurement Fundamentals
Research Problem Definition
- Define decision problem
- Specify research question
- Establish research objective
- Benefits of expected information = report
Types of Data
- Primary Data
- Secondary Data (Internal & External)
- Customer Knowledge
Qualitative Research (Primary Exploratory)
Qualitative research is primary exploratory research, subjective in nature.
Advantages of Qualitative Research
- Cheaper
- No better way to understand in-depth motivations and feelings
- Can improve efficiency of quantitative research
Limitations of Qualitative Research
- Many successes and failures are based on small differences
- Not necessarily representative of the population of interest
- A dominant individual can skew results
Quantitative Research
Used to quantify the problem by way of numerical data or data that can be transformed into usable statistics.
Fundamental Research Types
- Basic (Pure) Research: Attempts to expand the limits of knowledge.
- Applied Research: Conducted when a decision must be made about a specific real-life problem.
Research Design Frameworks
Exploratory Research
Problem Status: Ambiguous problem.
Objectives: Focus on gaining background information, defining terms, and narrowing the scope of the research topics. It is the initial stage of the research process and is not intended to provide conclusive evidence.
Exploratory Research Techniques
- Secondary data analysis
- Pilot studies
- Case studies
- Focus groups (Qualitative)
Descriptive Research
Problem Status: Partially defined problem.
Objectives: Focus on describing and measuring marketing phenomena at a particular point in time (who, what, when, where, how). Accuracy is important.
Method: Surveys/questionnaires (Quantitative) are the most common method (in-person, telephone, internet).
Causal Research
Problem Status: Sharply defined problem.
Focuses on establishing cause and effect relationships among variables (e.g., test marketing).
Variables and Relationships
- Independent Variable: Stands alone and does not change due to other variables you are trying to measure (e.g., a person’s age).
- Dependent Variable: Depends on other factors and is liable to change (e.g., eating habits can change this).
Measurement Scales (Levels of Measurement)
| Scale | Operation | Descriptive Statistics |
|---|---|---|
| Nominal | Counting | Frequency, Percentage, Mode |
| Ordinal | Rank ordering | Median, Range, Percentile ranking |
| Interval | Order and relative magnitude | Mean, Standard deviation, Variance |
| Ratio | Operations on actual quantities | Geometric mean, Coefficient of variation |
Test Market Approaches
Traditional Tests
- Advantages: Conducted in actual distribution channels; can determine both customer acceptance and trade support.
- Disadvantages: Cost, time, and exposure to competition.
Controlled Test Markets
- Advantages: Distribution is assured; costs are lower; competitive monitoring is difficult (for competitors).
- Disadvantages: Limited number of markets; trade support is unknown.
Simulated Test Markets
- Advantages: Cost and time saving; predicts trial and purchase cycle.
- Disadvantages: Isolation from the real world; broad-based customer reaction is difficult to measure.
Experimental Design Structures
One-Shot Design
A group of subjects is administered a treatment and then measured (or observed). No attempt is made to randomly assign subjects to the groups, nor does the design provide for any additional groups as comparisons.
Notation: GP – T – O
Note: The one-shot design may be useful as an inexpensive measure of a new treatment.
One-Group, Pre-Post Design
One group is given a pre-treatment measurement or observation, the experimental treatment, and a post-treatment measurement or observation. The post-treatment measures are compared with their pre-treatment measures.
Static Group Design
Two intact groups are used, but only one of them is given the experimental treatment. At the end of the treatment, both groups are observed or measured to see if there is a difference between them as a result of the treatment.
Notation:
- GP – T – O
- GP —– O
Note: Whether the groups were comparable or not is crucial in determining the extent of information yielded by this design.
Post-Test Only Control Group Design
Similar to the static group design, but attempts to ensure similarity of the groups before treatment. The design guarantees comparability between groups by assigning subjects to groups at random (R).
Notation:
- R – GP – T – O
- R – GP —— O
Pretest-Post Test Control Group Design
Adds a pre-test to the previous design as a check on the degree of comparability of the control and experimental groups before the treatment is given.
Notation:
- R – GP – O – T – O
- R – GP – O —— O
Note: This yields information on pre-treatment behavior and a comparison of post-treatment behavior between groups. Avoids most threats to internal validity because groups are randomized.
Solomon Four Group Design
Attempts to control for the possible “sensitizing” effect of the pre-test or measurement by adding two groups who have not been a part of the pre-test process.
Notation:
- R – GP – O – T – O
- R – GP – O —— O
- R – GP —– T – O
- R – GP ——— O
Note: Frequently used in behavioral, educational, and medical studies where the testing process allows the subject to “learn.”
Factorial Design
Assigns variations of the treatment. For example, a 2×2 factorial design tries two kinds of treatments varied in two ways. Some factorial designs include both assignment of subjects (blocking) and several types of experimental treatment in the same experiment.
Example (2×2):
- R – GP – T (A1 B1) – O
- R – GP – T (A1 B2) – O
- R – GP – T (A2 B1) – O
- R – GP – T (A2 B2) – O
Time Series Design
This design is used to assess the effects of a treatment with the same group or the same individual over a period of time. A measure or observation is made more than once to assess the effects of the treatment.
Notation: GP – T – O – T – O – T – O or GP – O – O – O – T – O – O – O
Note: There is no randomization of test units to treatments. Example: Assessing the effect of ad spending on sales or the plain packaging law on cigarette sales.
Probability Sampling Methods
Simple Random Sampling (SRS)
Researchers use a table of random numbers, a random number generator, or some random selection procedure that ensures each sampling unit in the target population has an equal chance of selection.
Systematic Random Sampling
More efficient than SRS. Involves selecting samples based on a fixed periodic interval.
Stratified Random Sampling
Used when the defined target population is believed to have a non-normal (skewed) distribution for one or more distinguishing characteristics (age, sex, income, etc.).
- Researchers must identify sub-populations referred to as strata.
- After strata are segmented, a simple random sample is drawn for each stratum.
Cluster Sampling
Requires the defined target population to be segmented into geographic areas, each of which is considered to be very similar to the others.
- Researchers randomly select a few areas (clusters).
- A census of the elements in each selected cluster is then conducted.
Non-Probability Sampling Methods
Convenience Sampling
Samples drawn at the convenience of the researcher or interviewer, often as the study is being conducted. This method has the potential for significant bias.
Judgment Sampling
Participants are selected according to the researcher’s or some other experienced individual’s belief that they will meet the requirements of the study. Used when looking for a specific target market.
Quota Sampling
Selection of prospective participants according to pre-specified quotas regarding demographic characteristics or specific attitudes/behaviors.
Underlying Purpose: To provide assurance that pre-specified subgroups of the defined target population are represented on pertinent sampling factors (cross-section).
Snowball Sampling
Identifying and qualifying a set of initial prospective respondents who can help the researcher identify additional people to be included in the study (referrals).
Measurement and Construct Validity
Goal: To obtain high-quality data.
Construct Development
The process of precisely identifying and defining what is to be measured. A construct is a hypothetical variable comprised of responses or behaviors that are thought to be related.
- Construct Abstractness: Concrete vs. subjective properties; the more subjective, the more abstract.
- Construct Dimensionality: Identifiable and measurable components that constitute the domain of observables.
- Construct Validity: The process of establishing that the construct is valid, by testing for content, convergent, discriminant, and nomological validity.
- Construct Operationalization: Explaining a construct’s meaning in measurement terms by specifying the activities necessary to measure it.
Key Measurement Concepts
- Concept: A generalized idea.
- Conceptual Definition: Verbal explanation of the meaning of a concept (what it is and what it is not).
- Operational Definition: Gives meaning to a concept by specifying the activities or operations necessary to measure it.
Scale Reliability and Errors
Criteria for Good Measurement
Reliability: The degree to which measures are free from random error.
Split-Half Method: Used to determine constancy by checking one half of a set of results against the other half.
Types of Research Errors
Random Sampling Error
Caused by choosing a sample that does not perfectly represent the population (e.g., choosing only 50 respondents enrolled in MKTG202 to represent the population of all MQ students).
Note: Unavoidable but can be estimated (calculating confidence intervals) or reduced (increasing sample size).
Systematic Error (Non-Sampling Error)
Results from some imperfect aspect of the research design or from a mistake in the execution of the research. Can be managed.
Sample Bias (Respondent Error)
A persistent tendency to deviate in one direction from the true value of the population parameter. This error results from some respondent action or inaction.
Non-Response Error
The statistical difference between a survey that includes only those who responded and a perfect survey that would also include those who failed to respond.
- Non-Respondent: A person who is not contacted or who refuses to cooperate in the research.
- No Contact: A person who cannot be reached (e.g., target population members not online).
- Refusal: A person unwilling to participate in research.
- Self-Selection Bias: People who feel strongly about a subject are more likely to respond to survey questions than people who feel indifferent about it (e.g., volunteers).
Response Bias
Respondents tend to answer questions with a certain bias that consciously or unconsciously misrepresents the truth.
- Deliberate Falsification: People deliberately give false answers to appear intelligent or conceal personal information (common when interviewing children and politicians).
- Unconscious Misrepresentations: Response bias arising from question format or content, even when the respondent is trying to be truthful.
- Acquiescence Bias: Some individuals tend to agree with all questions (e.g., an obedient respondent choosing “yes”/“agree” to all questions).
- Extremity Bias: Some individuals tend to use extremes when responding to questions (ticking the lowest or highest marks for all questions).
- Auspices Bias: Respondents are influenced by the organization conducting the study (e.g., questions about general shopping experiences asked within a Woolworths store).
- Social Desirability Bias: Respondents’ desire, either conscious or unconscious, to gain prestige or appear in a different social role (e.g., questions related to personal or moral behavior).
Question Classification in Surveys
Surveys may be classified according to structure and disguise:
- Structured Question: A question that imposes a limit on the number of allowable responses (closed-ended).
- Unstructured Question: A question that does not restrict the respondents’ answers; open-ended, allowing considerable freedom.
- Undisguised Question: Straightforward; assumes that the respondent is willing to answer.
- Disguised Question: Indirect; assumes that the purpose of the study must be hidden from the respondent.
Limitations of Classification:
- The degree of structure and the degree of disguise vary; they are not clear-cut categories.
- Most surveys are hybrids, asking both structured and unstructured questions.