Module 6 Quantitative Research II (Section 1) PDF

MKTG 330 Marketing Research Danny Kim Module 6: Quantitative Research II Schedule  Descriptive research Causal research Types of surveys Quasi-experiment CA3: FGD video (individual) CA4: FGD video II (individual)  Researc...

MKTG 330 Marketing Research Danny Kim Module 6: Quantitative Research II Schedule  Descriptive research Causal research Types of surveys Quasi-experiment CA3: FGD video (individual) CA4: FGD video II (individual)  Research methodology Qualitative research Quantitative research Exploratory research Descriptive research Causal research (observational research) (survey research) (experiment research) Indirect methods Cross-sectional study Panel study Projective techniques Continuous Association tests Discontinuous Word association tests Sentence completion tests Picture interpretation tests Observation studies Direct methods Depth interviews Focus group discussions Causal Research Causal research (Experiment research) Purpose Predict future event occurrences. Identify predictors of events. Causal: Nature Test if one event triggers another event to happen. Method Experiment Predictive: Function When an event happens, another event will soon follow. Causal Research: Methods An experiment must be conducted to test a cause-and-effect relationship. An experiment has two components: - The cause: Independent variable (X) – aka “the treatment.” - The outcome: Dependent variable (Y) – aka “the result.” Causal Research: Methods Suppose you want to test if smoking causes depression. - What is the independent variable (treatment)? - What is the dependent variable (outcome)? Causal Research: Methods Suppose you want to investigate if dancing to the sun causes sunny weather. - What is the independent variable (treatment)? - What is the dependent variable (outcome)? Causal Research: Methods Suppose you want to examine if offering a discount causes a spike in sales. - What is the independent variable (treatment)? - What is the dependent variable (outcome)? Experiment An experiment must meet three conditions. (1) Correlation (2) Temporal precedence (3) Random assignment Experiment: (1) Correlation A change in the treatment (X) should be accompanied by a change in the outcome (Y). To say that a discount (X) caused a spike in sales (Y), - The presence of the discount (X) should correlate with the spike in sales (Y). - The absence of the discount (X) should NOT correlate with the spike in sales (Y). Experiment: (1) Correlation – Caveat Correlation alone does not prove causation. - Scientists once claimed that smoking causes depression. - Later, scientists corrected that depression can also cause smoking. Correlation does not prove the direction of the causal effect. Experiment: (1) Correlation – Caveat Correlation alone does not prove causation. - A great sales performance may have caused more discounts to stimulate further sales. - A poor sales performance may have caused a “no discount” policy to avoid further losses. We need to see what happened FIRST. Experiment: (2) Temporal Precedence A change in the treatment (X) should occur before the outcome (Y). To say that a discount (X) caused a spike in sales (Y), the discount should occur first and then the spike in sales. Experiment: (2) Temporal Precedence – Caveat Temporal precedence alone does not prove causation. - Shamans once claimed that dancing causes sunny weather. - Later, scientists corrected that dancing has nothing to do with weather. Just because something happened before the outcome does not mean that it caused the outcome. Post hoc fallacy! Experiment: (2) Temporal Precedence – Caveat Beyond the discount, so many other things could have caused a spike in sales. All potential alternative reasons should be ruled out. Experiment: (3) Random Assignment All potential alternative reasons should be ruled out via random assignment. To say that a discount (X) caused a spike in sales (Y), - Stores should be randomly assigned to offer discounts. - A random assignment of stores rules out potential alternative reasons (location, foreign tourists). Experiment: (3) Random Assignment – Caveat Small sample: A random assignment cannot prevent bias. Bias due to Location! Bias due to travelers! Experiment: (3) Random Assignment – Caveat Random assignment is difficult to do. Ethicality issue: “Hypothesis: Class rewards prevent tardiness.” - Some students get rewards while others don’t? Practicality issue: “Hypothesis: GU students study more hours than UW students.” - Can you randomly assign and force people to attend GU and UW? When random assignment cannot be done, what should you do? Quasi-Experiment When random assignment cannot be done, a quasi-experiment can be conducted. Hypothesis: GU students study more hours than UW students - A quasi-experiment can relax the random assignment requirement.  Students who want to attend GU, attend GU. Students who want to attend UW, attend UW. - The quasi-experiment still satisfies the temporal precedence requirement.  Choice of university (GU vs. UW) precedes study hours. - The quasi-experiment still satisfies the correlation requirement.  Test if choice of university (GU vs. UW) varies with study hours. Now let’s see some real examples of experiments. Causal Research: Example I Research Question: Will people consume less ketchup if its color is blue (vs. red)? Hypothesis: People consume less ketchup if its color is blue (vs. red). Can we conduct an experiment? - What is the independent variable (treatment)? - What is the dependent variable (outcome)? Causal Research: Example I Does the experiment have three features to assess causality? - Correlation: Changes in the ketchup color should be accompanied by the changes in the ketchup consumption. - Temporal precedence: Participants are given one of the ketchup first, consumption is decided later. - Random assignment: Participants can be randomly assigned to the red (vs. blue) ketchup. Is a quasi-experiment necessary? Causal Research: Example II Research Question: Will people donate more in seeing a cat (vs. dog) on the poster? Hypothesis: People donate more in seeing a cat (vs. dog) on the poster. Can we conduct an experiment? - What is the independent variable (treatment)? - What is the dependent variable (outcome)? Causal Research: Example I Does the experiment have the three features to assess causality? - Correlation: Changes in the animal (cat vs. dog) should be accompanied by the changes in the donation amount. - Temporal precedence: Participants see one of the poster first, donation amount is decided later. - Random assignment: Participants can be randomly assigned to see cat (vs. dog). Is a quasi-experiment necessary?  Research methodology Qualitative research Quantitative research Exploratory research Descriptive research Causal research (observational research) (survey research) (experiment research) Indirect methods Cross-sectional study Panel study Experiments Projective techniques Continuous Quasi experiments Association tests Discontinuous Word association tests Sentence completion tests Picture interpretation tests Observation studies Direct methods Depth interviews Focus group discussions CA4: FGD Video II Download the CA4 worksheet (Canvas) Please watch a demo of a FGD titled “Focus Group Session 5 Demonstration” https://www.youtube.com/watch?v=scyTIRNDu_A Individual Assignment Deliverable: CA4 worksheet Due: Sep 22 (Sun), 2024, 11:59 PM PST