Experimentation

Marketing ResearchAaker, Kumar, Leone and Day Twelfth EditionInstructor’s Presentation Slides1Chapter Thirteen2ExperimentationMarketing Research 12th Edition ExperimentsStudies in which conditions are controlled so that one or more independent variable(s) can be manipulated to test a hypothesis about a dependent variable3Marketing Research 12th Edition Experimental vs. Descriptive Research 4Experimental ResearchManipulation of treatment variable (x), followed by observation of response variable (y)Experiment must be designed to control for other variables to establish causal relationshipDescriptive ResearchDescriptive research provides a snapshot of some aspect of market environment at a specific point in timeNo hint of a causal insight to be obtained from descriptive dataMarketing Research 12th Edition What Constitutes Causality?A change in one variable will produce a change in anotherConcept of a precondition influencing a variable of interestTime SequenceLack of association suggests absence of causation Attitude Behavior5Marketing Research 12th Edition Direction of Causation IssueApproach 1: Determining the direction of causationDraw on logic and previous theoryWhether one of the variables is relatively fixed and unalterableApproach 2: Consider the fact that there is usually a time lag exists between cause and effect The causal variable should have a positive association with the effect variable lagged in time6Marketing Research 12th Edition Conditions for Valid Causal Inference7Condition of Concomitant VariationEvidence that a strong association exists between an action and an observed outcomeCondition of Time Order of OccurrenceEvidence that the action preceded the outcomeAbsence of Competing Causal ExplanationsEvidence that there is no strong competing explanation for the relationship – that a high level of internal validity existsMarketing Research 12th Edition Laboratory and Field experiments8Marketing Research 12th Edition Experimental ResearchLaboratory Experiments Field ExperimentsLaboratory ExperimentsExperiments in which the experimental treatment is introduced in an artificial or laboratory settingTend to be artificialTesting effect exists as respondents are aware of being in a test and may not respond naturallyResults may not have external validityLeast costly and allow experimenter greater control over the experimentAlternative explanations of results are reduced, increasing internal validity9Marketing Research 12th Edition Field ExperimentsResearch study in which one or more independent variables are manipulated by the experimenter under carefully controlled conditions as the situation will permitExperimental treatment or intervention introduced in a completely natural settingResponse tends to be naturalTend to have much greater external validityDifficult to controlCompeting explanations for results exist10Marketing Research 12th Edition Threats to Experimental Validity11Marketing Research 12th Edition Threats to Internal validityHistoryMaturationTestingInstrumentationStatistical regressionSelection biasMortalitySelection–maturation interactionThreats to Experimental Validity (Contd.)12Marketing Research 12th Edition Threats to External validityThe reactive or interaction effect of testingThe interaction effect of selection biases and the experimental variableReactive effects of experimental arrangementsMultiple treatment interferenceIssues in Experimental ResearchWhat type of experimental design should be used?Should the experiment be performed in a "laboratory" setting or in the "field"?What are the internal and external threats to the validity of the experiment?13Marketing Research 12th Edition Basic Symbols and Notations O a formal observation or measurement X exposure of test units in the study to the experimental manipulation or treatment EG an experimental group of test units exposed to the experimental treatment CG a control group of test units participating in the experiment but not exposed to the experimental treatment R random assignment of test units and experimental treatments to groups. Randomization increases reliability M both the experimental group and the control group are matched on the basis of some relevant characteristics14Marketing Research 12th Edition Types of Experimental DesignsClassical Considers only one treatment level of an independent variable at a timeStatistical Allows for examining the impact of different treatment levels of an independent variable and the impact of two or more independent variables15Marketing Research 12th Edition 16Classification of Experimental DesignsMarketing Research 12th Edition Classical Designs - Pre-experimental DesignsOne Group, After-only DesignApply the experimental treatment to a subject or group and measure the results EG X O Leaves open the possibility that the results could be explained by events external to the design17Marketing Research 12th Edition Pre-experimental Designs (Contd.)Non-matched Control Group DeignIntroduce a control group to control for history and maturationEG X O1 - - - - - - - - - CG O2 18Marketing Research 12th Edition Pre-experimental Designs (Contd.)Matched Control Group DesignMatches experimental and control groups to reduce selection biasEG M X O1 - - - - - - - - - - - CG M O2 19Marketing Research 12th Edition Pre-experimental Designs (Contd.)One-group, Before - After DesignImprove control by adding before measureEG O1 X O2 Before measure adds sensitivity by adding another method to control for confounding variables20Marketing Research 12th Edition Pre-experimental Designs (Contd.)Threats to Experiment Validity in One-group, Before-After DesignBefore Measure EffectMay alert respondents to the fact that they are being studiedResults in more socially desirable behaviorMortality Effect Some subjects may stop participating in the experimentInstrumentation EffectResults from a change in the measuring instrument 21Marketing Research 12th Edition Classical Designs – True Experimental DesignsTrue experimental designs adopt random assignment procedure and use one or more control groupsRandom AssignmentFor any given assignment to a treatment, every member of the universe has an equal probability of being chosen 22Marketing Research 12th Edition True Experimental Designs (Contd.)Two Group, After-only DesignRandomization can match test and control groups on all dimensions simultaneously, given a sufficient sample size EG R X O1 - - - - - - - - - - - - - - CG R O2 There is no interaction effect of testing as there are no pretest requirements23Marketing Research 12th Edition True Experimental Designs (Contd.)Two-group, Before-after DesignAdds a control group to one-group, before - after designHelps control for history and maturationControls for reactive effect of O1 and O2 EG R O1 X O2 - - - - - - - - - - - - - - CG R O3 O424Marketing Research 12th Edition True Experimental Designs (Contd.)Solomon Four - Group DesignEG R O1 X O2 -------------------------------CG R O3 O4 ------------------------------EG R X O5 -------------------------------CG R O6 This design is often prohibitively expensiveProvides power to control for before measure effect of O1 on both X and O2 25Marketing Research 12th Edition Quasi-Experimental DesignsOffer some degree of control but there is no random assignment of variablesProvide more measurements and more information than pre-experimental designTime Series DesignsSeries of measurements are employed during which an experimental treatment occursEG O1 O2 O3 O4 X O5 O6 O7 O826Marketing Research 12th Edition Time Series Designs27Marketing Research 12th Edition Quasi-experimental Designs (Contd.)Trend StudiesMeasures over time come from succession of separate random samples from the same populationContinuous Panel StudiesCollect a series of measurements on the same sample of test units over an extended period of time28Marketing Research 12th Edition Statistical DesignsCompletely Randomized DesignAny number of treatments can be assigned to test units on a random basisEG1 R X1 O1 - - - - - - - - - - - - - - - - EG2 R X2 O2 - - - - - - - - - - - - - - - -EG3 R X3 O3 `29Marketing Research 12th Edition Statistical Designs (Contd.)Randomized Block DesignEmploys the randomization process for all variablesMatching ensures that there are no differences between test samples on matched variablesMatching and randomization are combined in randomized block designEG1 R X O1 - - - - - - - - - - - - - - - - CG1 R O2 ------------------------------ EG2 R X O3 - - - - - - - - - - - - - - - -CG2 R O430Marketing Research 12th Edition Statistical Designs (Contd.)Latin Square DesignReduces number of groups involved when interaction between the treatment levels and control variables are unimportant Requires same number of rows, columns, and treatment levelsCannot be used to determine interaction effects31Marketing Research 12th Edition Latin Square Design - Example32Marketing Research 12th Edition Statistical Designs (Contd.)Factorial DesignsTwo or more experimental variables are considered simultaneouslyEach combination of the experimental treatment levels applies to randomly selected groups EG1 R X1 (Hi Adv, Hi Price) O1 n 6 ....................................................................... EG2 R X2 (Hi Adv, Low Price) O2 n 6 ........................................................................ EG3 R X3 (Low Adv, Hi Price) O3 n 6 ........................................................................ EG4 R X4 (Low Adv, Low Price) O4 n 6 ........................................................................ EG5 R X5 (No Adv, Hi Price) O5 n 6 ........................................................................ EG6 R X6 (No Adv, Low Price) O3 n 6Provides the ability to determine interactive effects of pairs of experimental variables and the main effect33Marketing Research 12th Edition Interactive Effects34Marketing Research 12th Edition Guidelines for Conducting Experimental ResearchRecognition of and statement of the problem - should be clear and generally acceptableChoice of factors and levels – considering control and measurement Selection of the response variable - must provide useful information about process under studyChoice of experimental design – considering sample size, Selection of suitable run order for experimental trials, determination of whether or not blocking or other randomization restrictions are involvedPerforming the experiment – monitor carefully to avoid errorsData Analysis – use statistical methodsConclusion and Recommendations – draw practical conclusions and recommend a course of action35Marketing Research 12th Edition Limitations of ExperimentsCost InvolvedTime Considerations SecurityField experiment exposes marketing program in the marketplace Difficult to hide from competitors36Marketing Research 12th Edition Limitations of Experiments (Contd.)Implementation ProblemsDifficult to gain cooperation within the organizationContamination may occur in experiments involving market areas due to inability to confine the treatment to designated experimental areaVariability in behavior across test units can be so large that it is difficult to detect experimental effects37Marketing Research 12th Edition Uncertain Persistency of ResultsCauses:High rates of technological, economic, or social change in the market environmentAggressive competitive behavior38Marketing Research 12th Edition Simulated Test MarketingTo overcome limitations posed by experiments such as test marketing in terms of cost, time involved and sabotage or imitation by competition Enables firms to screen new ideas during the early phases of development and channel resources to maximize the potential for success Widely used by the packaged goods industry as an aid for effective business planning Examples of companies involved in STM research - BASES, ASSESSOR and Yankelovich. 39Marketing Research 12th Edition 40End of Chapter ThirteenMarketing Research 12th Edition