14 Is the experimental design any good Internal validity The manipulation of

14 is the experimental design any good internal

This preview shows page 14 - 17 out of 17 pages.

14Is the experimental design any good?Internal validity– The manipulation of the independent variable is the real cause of the observed effects – The degree to which the effects of extraneous variablesare controlledExternal validity – Can the effects found be generalized in reality?– What populations, settings, times, independent variables, and dependent variables can the results be projected?– It often comes down to lab vs. field experiments (more on this later)40Example of experiments with poor internal validity41• ‘Various pre-experimental designsVarious quasi-experimental designsExample of experiments with potentially low external validity42Which product would you choose?A researcher studies the effect of packaging design on consumerchoice in the behavioural labResearchers find consumers prefer the packaging on the right
Image of page 14
15What the choice looks like in the supermarket (field)?43Can the findings be extrapolated to the actual supermarket setting?Every else being equal, experiment designs with randomly assigned groups have higher internal validity than those without44Internal validityLowHighRandomassignmentof groupsWhy? The reason is explained on the next slideThe reason for this is that random assignment of testing units into EG and CG is one of the most important ways to control for extraneous variable45Experimental GroupControl GroupNew Marketing TacticOld Marketing TacticCustomer PoolRandomassignmentRecall the definition of experimentation – the process of manipulating theindependent variables and measuring their effects on the dependentvariable,while controlling for the effect of everything elseRandom assignment is particularly important because it solvesthe issue ofselection bias, which is the most common andserious issue with causal research
Image of page 15
16External validity can be enhanced by conducting experiments in the field (e.g., in a real supermarket or shopping website) with actual shoppers as subjects46Pros and cons of lab vs. field experiment47ArtificialEnvironmentRealisticHighControlLowHighDemand artifactsLowHighInternal validityLowLowExternal validityHighLowEffortHighLowCostHighLimitations of ExperimentationTime consumingExpensiveDifficult to administerCompetitors may deliberately contaminate the results of a field experiment.48
Image of page 16
17RecapCausal researchExperimental designReview Lecture 11 notes and suggested readings49Key ConceptsCausalityExperimentationMain elements of anexperimentExtraneous variableInternal and externalvalidityDifferent types ofexperimental designsLab vs. field experiment
Image of page 17

You've reached the end of your free preview.

Want to read all 17 pages?

  • Two '16
  • Causality, test units, static group design

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture