Experimental Design: Part I MAR 6648: Marketing Research February 3, 2010.

Slides:



Advertisements
Similar presentations
Making Inferences about Causality In general, children who watch violent television programs tend to behave more aggressively toward their peers and siblings.
Advertisements

Lecture 28 Categorical variables: –Review of slides from lecture 27 (reprint of lecture 27 categorical variables slides with typos corrected) –Practice.
GROUP-LEVEL DESIGNS Chapter 9.
Chapter 14 Comparing two groups Dr Richard Bußmann.
Lecture 4 Econ 488. Ordinary Least Squares (OLS) Objective of OLS  Minimize the sum of squared residuals: where Remember that OLS is not the only possible.
EXPERIMENTAL DESIGNS Criteria for Experiments
Validity, Sampling & Experimental Control Psych 231: Research Methods in Psychology.
Who are the participants? Creating a Quality Sample 47:269: Research Methods I Dr. Leonard March 22, 2010.
Validity, Sampling & Experimental Control Psych 231: Research Methods in Psychology.
Using Statistics in Research Psych 231: Research Methods in Psychology.
Sampling & Experimental Control Psych 231: Research Methods in Psychology.
More Simple Linear Regression 1. Variation 2 Remember to calculate the standard deviation of a variable we take each value and subtract off the mean and.
Experimental Control Psych 231: Research Methods in Psychology.
Sampling & Experimental Control Psych 231: Research Methods in Psychology.
Independent Sample T-test Often used with experimental designs N subjects are randomly assigned to two groups (Control * Treatment). After treatment, the.
Today Concepts underlying inferential statistics
Using Statistics in Research Psych 231: Research Methods in Psychology.
Experimental Research
1 Psych 5500/6500 Confounding Variables Fall, 2008.
+ Controlled User studies HCI /6610 Winter 2013.
Learning Objectives 1 Copyright © 2002 South-Western/Thomson Learning Primary Data Collection: Experimentation CHAPTER eight.
Research Methods in Psychology
Ms. Carmelitano RESEARCH METHODS EXPERIMENTAL STUDIES.
Copyright © 2010, 2007, 2004 Pearson Education, Inc. Chapter 26 Comparing Counts.
Experiment Basics: Variables Psych 231: Research Methods in Psychology.
Experimental Research. What is experimental research?  Research investigation in which conditions are controlled so that hypotheses can be tested and.
Introduction ANOVA Mike Tucker School of Psychology B209 Portland Square University of Plymouth Drake Circus Plymouth, PL4 8AA Tel: +44 (0)
Analysis of Variance ( ANOVA )
Statistics and Research methods Wiskunde voor HMI Bijeenkomst 3 Relating statistics and experimental design.
Analysis of Variance ST 511 Introduction n Analysis of variance compares two or more populations of quantitative data. n Specifically, we are interested.
Observation & Analysis. Observation Field Research In the fields of social science, psychology and medicine, amongst others, observational study is an.
Chapter Seven Causal Research Design: Experimentation.
Chapter Eight. Figure 8.1 Relationship of Experimentation to the Previous Chapters and the Marketing Research Process Focus of This Chapter Relationship.
1 Psych 5500/6500 t Test for Two Independent Means Fall, 2008.
Statistics (cont.) Psych 231: Research Methods in Psychology.
Consumer Behavior Research Methods MAR 3503 January 12, 2012.
Primary Data Collection: Experimentation Chapter 7.
DIRECTIONAL HYPOTHESIS The 1-tailed test: –Instead of dividing alpha by 2, you are looking for unlikely outcomes on only 1 side of the distribution –No.
Educational Research Chapter 13 Inferential Statistics Gay, Mills, and Airasian 10 th Edition.
Copyright © 2010 Pearson Education, Inc. Slide
1 Overview of Experimental Design. 2 3 Examples of Experimental Designs.
1.) *Experiment* 2.) Quasi-Experiment 3.) Correlation 4.) Naturalistic Observation 5.) Case Study 6.) Survey Research.
 Descriptive Methods ◦ Observation ◦ Survey Research  Experimental Methods ◦ Independent Groups Designs ◦ Repeated Measures Designs ◦ Complex Designs.
Hypotheses. What is a hypothesis? A precise and testable statement A prediction about what the outcome of an experiment will be Usually derived from a.
Experimental Design Econ 176, Fall Some Terminology Session: A single meeting at which observations are made on a group of subjects. Experiment:
1 Psych 5510/6510 Chapter 13 ANCOVA: Models with Continuous and Categorical Predictors Part 2: Controlling for Confounding Variables Spring, 2009.
Chapter 13 Repeated-Measures and Two-Factor Analysis of Variance
Finishing up: Statistics & Developmental designs Psych 231: Research Methods in Psychology.
Introduction to Statistics for the Social Sciences SBS200, COMM200, GEOG200, PA200, POL200, or SOC200 Lecture Section 001, Fall 2015 Room 150 Harvill.
Chapter Nine Primary Data Collection: Experimentation and
Sampling Sampling – the process of obtaining a sample from a population Simple Random Sampling – sample selected at random from a population in which every.
Essentials of Marketing Research William G. Zikmund Chapter 9: Experimental Research.
11 Chapter 9 Experimental Designs © 2009 John Wiley & Sons Ltd.
INTRODUCTION TO METHODS Higher Psychology. What do Psychologists do?  Discuss in groups  5MINS.
Statistics (cont.) Psych 231: Research Methods in Psychology.
Chapter 13 Understanding research results: statistical inference.
Research in Psychology Chapter Two 8-10% of Exam AP Psychology.
 Allows researchers to detect cause and effect relationships  Researchers manipulate a variable and observe whether any changes occur in a second variable.
Fundamentals of Data Analysis Lecture 4 Testing of statistical hypotheses pt.1.
Review Statistical inference and test of significance.
Lecturer’s desk INTEGRATED LEARNING CENTER ILC 120 Screen Row A Row B Row C Row D Row E Row F Row G Row.
Psychological Experimentation The Experimental Method: Discovering the Causes of Behavior Experiment: A controlled situation in which the researcher.
Essentials of Marketing Research William G. Zikmund
Inference and Tests of Hypotheses
Chapter 8 Experimental Design The nature of an experimental design
2 independent Groups Graziano & Raulin (1997).
Two-Sample Between-Subjects Experiments and Independent-Samples t-Tests So far, we’ve talked about experiments in which we needed to take only one sample.
Introduction to Experimental Design
Psych 231: Research Methods in Psychology
Presentation transcript:

Experimental Design: Part I MAR 6648: Marketing Research February 3, 2010

Overview What are the basic features of an experiment? How do those features get implemented in a real experiment? How do we adapt experiments to meet our goals and resources?

1.Experimentation is the conscious manipulation of one or more variables by the experimenter in such a way that its effect on one or more variables can be measured. 2.The variable being manipulated is called the independent variable (a.k.a. cause). 3.The variable being measured is called the dependent variable (a.k.a. effect). 4.Elimination of other possible causal factors: i.e., the research design should rule out the other factors (exogenous variables) as potentially causal ones. 5.This is typically done through random assignment to condition

An example of an experiment Suppose you want to know whether commercials make people enjoy TV shows less This means you’ll want to have some shows without commercials, and some shows with them – Therefore, commercials (or not) is the independent variable And you’ll want to measure enjoyment of the TV shows they watch – Therefore, enjoyment is the dependent variable

Conditions Not in terms of what you can and can’t do… Each independent variable (or combination of IVs) is called a condition Condition 1Condition 2

Hypotheses Experimentation is essentially the process of trying to determine which of two hypotheses is not false The null hypothesis: – H 0 : Usually that there are no differences between conditions The alternative hypothesis: – H 1 : usually that there is a difference between conditions P-values in stats essentially represent the likelihood that we found evidence for H 1 by chance alone

Confirmation Bias We are inclined to confirm our beliefs but less inclined (or able) to disconfirm them A real world example: – Business managers don’t keep track of those they don’t hire Why? – Theories lead to unwarranted confidence – Inability to search out disconfirmation – Fixation or mental set

Control condition Control conditions allow us to see that our manipulations caused (or didn’t cause) a change in the dependent variable Usually a control condition is just no manipulation – This is sometimes done by adjusting when you run your manipulation Sometimes, though, you want to compare your new manipulation to what’s typically done now – The control condition may be the standard or default

Random Assignment This essentially means that any one participant is equally likely to be in any condition – Usually you put your conditions in random order, and assign participants in the order that they “arrive” – Computers now allow you to assign people on the spot Randomizer.org or random.org are good sources

An example of an experiment The hallmark of an experiment is random assignment to conditions – Let’s say the groups (the commercial watchers and the people who watch it straight through) now look different! – Random assignment means that the two groups should not have differed systematically at the start – It also means that only your independent variable was different between groups Random assignment and manipulation of the IV mean that you can infer that the IV causes a change in the DV

An example of an experiment Question: do commercials make you enjoy a TV show less? Do people correctly predict this? Randomly assign your participants to groups – Half will predict how they enjoy a TV show with or without them, half will actually experience it and report how they feel – Half will watch a TV show with commercials, half will watch the same show without them Measure enjoyment or predicted enjoyment

An example of an experiment Nelson, Meyvis, & Galak, 2009

Example Objective: GAP wishes to gauge whether new more aggressive sales techniques employed by store assistants increase sales What is the best experimental design?

Experiment 1 Design: – 50 stores are sampled at random and assistants are trained in the new approach Metric = MINUS Average sales for the 50 stores in the next six months Average sales for the 50 stores in the next six months Average sales for the 50 stores in the prior six months Average sales for the 50 stores in the prior six months

Notation X = Exposure of a sample to the independent variable (i.e., what we manipulate – “treatment”) O = Observation of measurement of the dependent variable (i.e., what we measure / want to affect) Movement through time is represented by the horizontal arrangement of Xs and Os from left to right.

Experiment 1: One group – before after Causal Effect of X = O 2 -O 1

Problems with this design? History or maturation Defensiveness Mortality Instrumentation

Experiment 2 Design: – 50 stores are sampled at random and the assistants are trained in the new approach – Another 50 stores are sampled at random as control Metric = MINUS Average sales for the 50 test stores in the next six months Average sales for the 50 test stores in the next six months Average sales for the 50 control stores in the next six months Average sales for the 50 control stores in the next six months

Experiment 2: Two group – only after Causal Effect of X = O 2 -O 1

Problems with this design?

Experiment 3 Design: – 50 stores are sampled at random and assistants are trained in the new approach – Another 50 stores are sampled at random as control Metric = MINU S Average sales for the 50 test stores in the next six months Average sales for the 50 test stores in the prior six months Average sales for the 50 control stores in the next six months Average sales for the 50 control stores in the prior six months

Experiment 3: Two group – before after Causal Effect of X = O 4 -O 3 – (O 2 -O 1 )

More Advanced Experiments We have so far mainly looked at simple experiments But often we need to test several variables When deciding on a marketing plan for a new product there are many factors involved

Factorial Design Suppose we wish to test both product price and web-design for an e-business $9.99$14.99$19.99 Design 1 Design 2 Price Design Full Factorial Design! Full Factorial Design!

Interactions and main effects

Factorial Design What do we do if we have many factors and levels? Example: – 5 prices, 4 product designs, 3 ad-copies  5*4*3 = 60 experimental cells! Solution: Use a fractional factorial design – Only use a subset of all 60 cells in experiment – Rely on regression analysis to extrapolate

Latin Squares 1 st ad2 nd ad3 rd ad4 th ad Group #1 Positive ad, Male speaker Positive ad, Female speaker Negative ad, Female speaker Negative ad, Male speaker Group #2 Positive ad, Female speaker Negative ad, Male speaker Positive ad, Male speaker Negative ad, Female speaker Group #3 Negative ad, Male speaker Negative ad, Female speaker Positive ad, Female speaker Positive ad, Male speaker Group #4 Negative ad, Female speaker Positive ad, Male speaker Negative ad, Male speaker Positive ad, Female speaker

Latin Squares 1 st ad2 nd ad3 rd ad4 th ad Group #1 αβδγ Group #2 βγαδ Group #3 γδβα Group #4 δ αγβ

An experiment? Steve was interested to see how much labels on wine bottles affect how much people enjoy the wine inside them. At a party, he served the wines like normal, leaving the bottles out for people to pour from, labels still on. He asked everyone to indicate which wine they liked the best. At the next party he threw, he poured the wine into decanters, so that his guests couldn’t see the labels when they poured the wine. They again indicate which wine they liked best, and they had different preferences from the last party.

An experiment? The owner of two McDonalds franchises here in Gainesville wants to see if transactions run more quickly if he uses both drive-thru windows or only one. He picks one restaurant to use both windows at all times for a month, and the other he has closed at all times for a month. He finds that the drive-thru that uses both windows has notably faster service times.

An experiment?

Summary Experiments are very useful for determining causality – The main hallmarks of experiments are random assignment to condition, manipulation of the independent variable, and a control group – There are many different types of experiments, which vary largely on whether they are run within or between subjects (or both), when the manipulation is run, and how many conditions are used