Business Research Methods William G. Zikmund Chapter 12: Experimental Research
Experiment A research investigation in which conditions are controlled One independent variable is manipulated (sometimes more than one) Its effect on a dependent variable is measured To test a hypothesis
Basic Issues of Experimental Design Manipulation of the Independent Variable Selection of Dependent Variable Assignment of Subjects (or other Test Units) Control Over Extraneous Variables
The experimenter has some degree of control over the independent variable. The variable is independent because its value can be manipulated by the experimenter to whatever he or she wishes it to be.
Experiment Treatment Alternative manipulations of the independent variable being investigated
Independent Variable The experimenter controls independent variable. The variable’s value can be manipulated by the experimenters to whatever they wish it to be.
Manipulation of Independent Variable Classificatory Vs. continuous variables Experimental and control groups Treatment levels More than one independent variable
Experimental Treatments The alternative manipulations of the independent variable being investigated
Dependent Variable Its value is expected to be dependent on the experimenter’s manipulation Criterion or standard by which the results are judged
Dependent Variable Selection Measurement e.g... sales volume, awareness, recall, Measurement
Test Units Subjects or entities whose response to the experimental treatment are measured or observed.
Two Types of Experimental Error Constant errors Random errors
Field versus Laboratory Experiments
Controlling Extraneous Variables Elimination of extraneous variables Constancy of conditions Order of presentation Blinding Random assignment
How May an Experimenter control for Extraneous Variation? Eliminate Extraneous Variables Hold Conditions Constant Randomization Matching Subjects
Establishing Control
Demand Characteristics Experimental procedures that intentionally hint to subjects something about the experimenter’s hypothesis
Demand Characteristics Guinea pig effect Hawthorne effect
Field Vs. Laboratory Experiment
Laboratory Experiment Field Experiment Artificial-Low Realism Natural-High Realism Few Extraneous Variables Many Extraneous Variables High control Low control Low Cost High Cost Short Duration Long Duration Subjects Aware of Participation Subjects Unaware of Participation
Control Groups Isolate extraneous variation
When does an Experiment have Internal Validity? Internal Validity - The ability of an experiment to answer the question whether the experimental treatment was the sole cause of changes in a dependent variable Did the manipulation do what it was supposed to do?
Factors Influencing Internal Validity History Maturation Testing Instrumentation Selection Mortality
Isolating Extraneous Variation with a Control Group History Effects Maturation Effects Mortality Effects
Type of Extraneous Variable Example History - Specific events in the environment between the Before and After measurement that are beyond the experimenter’s control Maturation - Subjects change during the course of the experiment Testing - The Before measure alerts or sensitizes subject to nature of experiment or second measure. A major employer closes its plant in test market area Subjects become tired Questionnaire about the traditional role of women triggers enhanced awareness of women in an experiment.
Instrument - Changes in instrument result in response bias Selection - Sample selection error because of differential selection comparison groups Mortality - Sample attrition; some subjects withdraw from experiment New questions about women are interpreted differently from earlier questions. Control group and experimental group is self-selected group based on preference for soft drinks Subjects in one group of a hair dying study marry rich widows and move to Florida
How can Internal Validity Increase?
Increasing Internal Validity Control group Random assignment Pretesting and posttesting Posttest only
What are the Different Basic Experimental Designs?
Quasi-Experimental Designs One Shot Design (After Only) One Group Pretest-Posttest Static Group Design
One Shot Design (After Only) X O1
One Group Pretest-Posttest O1 X O2
Static Group Design Experimental Group X O1 Control Group O2
Three Good Experimental Designs Pretest - Posttest Control Group Design Posttest Only Control Group Solomon Four Group Design
Pretest-Posttest Control Group Design Experimental Group R O1 X O2 Control Group R O3 X O4
Posttest Only Control Group Experimental Group R X O1 Control Group R O2
One-Shot Design Internal Validity Problems History weak Maturation Testing not relevant Instrumentation not relevant Selection weak Mortality
One-Group Pretest-Posttest Internal Validity Problems History weak Maturation Testing Instrumentation weak Selection controlled Mortality
Static-Group Design Internal Validity Problems History controlled Maturation possible source of concern Testing Instrumentation controlled Selection weak Mortality
Pretest-Posttest Control Internal Validity Problems History controlled Maturation Testing Instrumentation controlled Selection Mortality
Solomon Four-Group Design Internal Validity Problems History controlled Maturation Testing Instrumentation controlled Selection Mortality
Posttest-Only Control Internal Validity Problems History controlled Maturation Testing Instrumentation controlled Selection Mortality
Experimental Group 1: R O1 X O2 Experimental Group 2: R X O5 Solomon Four Group Design Experimental Group 1: R O1 X O2 Control Group 1: R O3 O4 Experimental Group 2: R X O5 Control Group 2: R X O6
Advanced Experimental Designs are More Complex Completely randomized Randomized block design Latin square Factorial
Completely Randomized Design An experimental design that uses a random process to assign subjects (test units) and treatments to investigate the effects of only one independent variable.
Completely Randomized Designs Control: no music Experimental treatment: slow music Experimental treatment: fast music Average minutes shopper spends in store 16 18 12
Independent Variable A Level 1 Level 2 Level 3 Group A Group B Group C
Completely Randomized Design With a pretest posttest Group A R O1 X1 O2 Group A R O3 X2 O4 Group A R O5 X3 O6
Completely Randomized Design With a posttest Group A R X1 O1 Group B R X2 O2 Group C R X3 O3
Randomized Block Design An extension of the completely randomized design in which a single extraneous variable that might affect test units’ response to the treatment has been identified and the effects of this variable are isolated by blocking out its effects.
Randomized Block Design Independent Variables Control: no music Experimental treatment slow music Experimental treatment: fast music Mornings and afternoons Evening hours Blocking variable
Factorial Design An experiment that investigates the interaction of two or more variables on a single dependent variable.
Independent Variable 1 No Music Slow Music Fast Music No Music cart signs Independent Variable 2 Grocery cart signs
Factorial Design -- Roller Skates Package Design Price Red Gold $25 Cell 1 Cell 4 $30 Cell 2 Cell 5 $35 Cell 3 Cell 6
Effects Main effect The influence of a single independent variable on a dependent variable. Interaction effect The influence on a dependent variable by combinations of two or more independent variables.
2 x 2 Factorial Design Ad A Ad B Men Women 65 > > 70 60 Main Effects of Gender > 70 60 > Main Effects of Ad
Interaction Between Gender and Advertising Copy 100 90 80 70 60 50 40 30 20 10 Ad A Ad B Women Men Believability
Independent Variable 1 Level 1 Level 2 Level 1 Level 2 Group A Group B Group D Group C
2 x 2 Factorial with a Pretest Posttest Group A R O1 X11 O2 Group B R O3 X21 O4 Group C R O5 X12 O6 Group D R O7 X22 O8
2 x 2 Factorial Design with a Posttest Measure Group A R X11 O1 Group B R X21 O2 Group C R X12 O3 Group D R X22 O4
A Test Market Experiment on Pricing Sales in Units (thousands) Test Market A, B, or C Test Market D, E, or F Test Market G, H, or I Test Market J, K, or L Mean Grand Mean Regular Price $.99 130 118 87 84 X1=104.75 X=119.58 Reduced Price $.89 145 143 120 131 X2=134.75 Cents-Off Coupon Regular Price 153 129 96 99 X1=119.25
Latin Square Design A balanced, two-way classification scheme that attempts to control or block out the effect of two or more extraneous factors by restricting randomization with respect to the row and column effects.
Order of Usage 1 2 3 1 A B C 2 B C A 3 C A B SUBJECT
TEST MARKETING Controlled experimentation Not just trying something out But scientific testing
Test Marketing Controlled experimentation Not just trying something out But scientific testing
Test Marketing An experimental procedure that provides an opportunity to test a new product or a new marketing plan under realistic market conditions to measure sales or profit potential.
Functions of Test Marketing IDENTIFY AND CORRECT WEAKNESSES IN PLANS ESTIMATE OUTCOMES
A Lengthy and Costly Procedure $$$$$ When not to Test? Loss of Secrecy How Long Should a Test Last?
Popular Test Markets Pittsfield, Massachusetts Charlotte, North Carolina Columbus, Ohio Little Rock, Arkansas Evansville, Indiana Cedar Rapids, Iowa Eau Claire,Wisconsin Wichita, Kansas Tulsa, Oklahoma Omaha, Nebraska Grand Junction. Colorado Wichita Falls, Texas Odessa-Midland, Texas
Selecting a Test Market Population size Demographic composition Lifestyle considerations Competitive situation Media Self-contained trading area Overused markets - secrecy
Control Method of Test Marketing Small city Low chance of being detected Distribution is forced (guaranteed)
The Advantages of Using the Control Method of Test Marketing Reduced costs Shorter time period needed for reading test market results Increased secrecy from competitors No distraction of company salespeople from regular product lines
Some Problems Estimating Sales Volume Over-attention Unrealistic store conditions Reading competitive environment incorrectly Incorrect volume forecasts Adjusted data Penetration and repeat purchase rate Time lapse
High Tech Test Markets Electric Test Markets Simulated Test Markets Virtual-reality Simulated Test Markets