Skip to main content
Search form
  • 00:00


  • 00:15

    ELIZABETH MILLER: I'm Elizabeth Miller.I'm an Associate Professor of Marketingat the University of Massachusetts in Amherst,and I study consumer behavior.Particularly I've been interested in how emotionsand the way we present information influences consumerchoices.And I've also been interested in how we can encourage consumersto make healthier choices.

  • 00:35

    ELIZABETH MILLER [continued]: Today I'm going to be talking about experiments.Experiments are a systematic studydesigned to examine the consequences of deliberatelymanipulating potential causal agents.So it's a way to study cause and effect.And it has a couple of different pieces.So you need to have a dependent variable, which is that thingthat you want to study.That is the effect.You need to have an independent variable, which isthe thing that you manipulate.

  • 00:59

    ELIZABETH MILLER [continued]: And then, of course, you need to have at least one observation.And you need to have a way for inferring the cause.So experiments are pretty common in psychology.So my undergraduate degree is in psychology.

  • 01:19

    ELIZABETH MILLER [continued]: And consumer behavior is really the studyof the psychology of consumers.And so that's a common method that's used.The only method that can establish causality,because you have the control over and youcan set things up to see that this is justthe only thing that has changed.And so it's really the only methodthat allows you to infer cause.

  • 01:49

    ELIZABETH MILLER [continued]: This method can be used to study a wide variety of questions.In my own research, I study thingslike how does shelf location influenceperceptions of the product, the waya color is named influence product perceptions.Do children pay attention to health claims on packages?Really, there's a very broad set of questionsthat can be varied.

  • 02:09

    ELIZABETH MILLER [continued]: You have to be able to manipulate the causeand you have to be able to measure the effects.And as long as you can do those two things,there's probably no end to the types of questionsthat can be studied.There's a wide variety of data that can be collectedand studied with experiments.

  • 02:32

    ELIZABETH MILLER [continued]: You can look at attitudes.You can look at behavior.You can look at behavioral intentions.You can look at actual behavior, stress levels.There are some studies that you cantrace where people are looking in termsof their like eye-tracking kinds of studies,really a wide number of dependent variable sidequestions that can be studied.

  • 02:58

    ELIZABETH MILLER [continued]: We looked at whether children use nutritional informationon packages.And so we actually made sort of two-dimensional cereal boxesthat children could look at.And it had the name of the cereal,a picture of the cereal, and then it either had a claimon it or it didn't.And we looked at different types of claims--a nutrient content claim, which was they had fiber.We looked at a health claim that'sconnected the fiber to reductions in heart disease.

  • 03:22

    ELIZABETH MILLER [continued]: And we looked at a general claim that was just saidthe cereal is good for you.And then based on what box they saw,we asked them how likely would you buy this cereal.And I think we also asked them some perceptions.Based on that, we were able to assess that they did.I mean, they did see the claims and they--as you, maybe not surprisingly-- they had a negative impact.

  • 03:42

    ELIZABETH MILLER [continued]: You know, experiments are always a little bitof trial and error, where you--because you might ask a question and you think it's clear to youand then it turns out it's not.Or try to manipulate the variable of interest,the independent variable, and the manipulation justdoesn't do what you thought it would do.So I study stuff with emotion, too.And we're studying the effects of fearon perceptions of creativity.

  • 04:06

    ELIZABETH MILLER [continued]: We wanted to compare fear to other negative emotions.And one of the ways we did it was with print ads originally.In psychology, they have a method calledIAPS, which is International Affective Picture System,I think is what it stands for.And they've developed images that they'vetested and are known to create particular emotions.But it turns out that anger is actually really hardto create with a static image.

  • 04:30

    ELIZABETH MILLER [continued]: And so we pretested an ad that had an anger image.But we found, as other researchers found,that we just couldn't create anger with the static image.And so in later studies, we moved to using videosto create the emotions.

  • 04:53

    ELIZABETH MILLER [continued]: Shadish, Cook, and Campbell, 2002have a book on experimental design that's really good.That's probably the main standardin terms of experimental design.But the other thing that's really useful in termsof designing experiments is just reallyhaving a good knowledge of the literatureand writing a good hypothesis.

  • 05:13

    ELIZABETH MILLER [continued]: If you have a really strong hypothesis,it really tells you how to run an experiment.And so a strong hypothesis shouldclear the independent variable and the dependent variable.It should be based on the prior literature or some sortof logical arguments.It should make a clear prediction.And those predictions need to be testable.And so having a strong hypothesisis a really good place to start in termsof designing your experiments.

  • 05:38

    ELIZABETH MILLER [continued]: And then also, knowing the literatureis useful not only for creating those hypotheses,but also in terms of knowing what kinds of manipulationshave been done in the past.And one of the most difficult aspects of an experimentis being able to rule out what are known as confounds.So a confound is when the thing that you manipulated alsochanges with something unintended.

  • 05:59

    ELIZABETH MILLER [continued]: So for example, the emotions, you want to make sure that--if the image you use creates fear,but it also creates anger, then youdon't know if you have effect.You don't know was it fear that caused that effect or angerthat caused the effect.And so probably the trickiest thing about doing an experimentis ruling out these confounds.

  • 06:20

    ELIZABETH MILLER [continued]: That ability to sort of ask why and to understanddifferent potential alternative explanations is really useful.And the other thing I think is really importantis that when you do design an experiment,before you launch it, you should take it as a test subject,because there are many times when you go-- you know,you've designed it, you wrote all the questions.But when you actually go through it, as in put yourselfin the place of the subject, you oftensee things that you didn't see before.

  • 06:45

    ELIZABETH MILLER [continued]: Either the flow of the question seems weird or you're like,oh, I don't know how I want to answer that question.And so that-- and you don't you justdon't see those things until you actually put yourselfin the place of the subject.So I highly recommend doing that.Sometimes with open-ended questions,you don't always know how people are going to address them.And sometimes you don't know how to interpret what they wrote.

  • 07:07

    ELIZABETH MILLER [continued]: I mean, there are many times whereopen-ended is what you want.We just ran a study where we were looking reallyat the impact service-- how peopleinteract with service robots.In looking at that, one of the questions we wondered was wewanted to have a measure of engagement.And so the context we were using was a health advisor, I guess.We had them enter in, like, so what questions wouldyou ask the advisor.

  • 07:30

    ELIZABETH MILLER [continued]: We then counted the number of questions that were asked.So there, the content wasn't so important.Although, we were finding it in the content that kindof interesting, too, because sometimes the questionswere about the quality.So they weren't necessarily like thingslike, well, what would you do next with me, you know,or how would my nutrition plan work,or what exercises would you recommend,which I think is what we sort of expected.

  • 07:53

    ELIZABETH MILLER [continued]: But they were things like, what are your credentials,why do you know this.So there were questions asking about that expertise.And so now we're going to also analyze that aspect of it,too, in terms of what kinds of questions they're asking.I mean, I think experiments are a really useful method.They sometimes get a bad rap in that, because a lot of timesexperiments are moved into the laband you need to set up situationsto be able to control things.

  • 08:15

    ELIZABETH MILLER [continued]: And so a lot of times, people criticize themfor sort of lacking reality.But there's also, with a lot of creativity,you can do interesting field studies.And if you combine field studies with studiesin the lab and even--now people can scape data and other types of methods--you can get pretty strong evidencefor the questions you're studying.[MUSIC PLAYING]


Elizabeth Miller, PhD, Associate Professor of Marketing at the University of Massachusetts, discusses the use of experiments to research consumer behavior.

Looks like you do not have access to this content.

An Introduction to Experiments in Consumer Behavior

Elizabeth Miller, PhD, Associate Professor of Marketing at the University of Massachusetts, discusses the use of experiments to research consumer behavior.

Copy and paste the following HTML into your website