Design Issues for Controls

• Author: Mike Rappeport
• Pages: 217-239

217

10

Design Issues for

Controls

By Mike Rappeport

INTRODUCTION

Surveys areused in Lanham Actlitigation toprovide evidence of

whatwould happen if apotential buyerof some productorservice

encountered the trademark/advertising stimuli of interest in the real

world. Thischapterconcentrateson the role and choice of “controls” in

evidentiary surveys.The approach taken hereispredicated on the idea

thatasurveycan produce “evidence”(e.g., alikelihood of confusion),

but itcannotand should notbe seen as“proof”(e.g., level of actual

confusion). In thislight,the goal of anysurveyistoproduce evidence

thatisascredible aspossible.1Controlsareused becausetheymake

surveyevidence significantlymorereflectiveofthe real world, and

thereforesignificantlymorecredible.

The chapterbeginsbydiscussing acommon problem thathasan effect

on almost anyexperimentinvolving humansassubjects,including not

just surveys,but such other“experiments” asatest of medical products or

acriminal line-up. Thisproblem isthatthe respondents’ knowledge that

theyaretaking part in asurveymeansthatbydefinition the surveydoes

notactuallyreplicatemarketconditions.Sosome procedure(s)must be

used toallowustotranslatethe surveyresults intowhatwould happen in

the real world. The usual procedureisthe useofwhatarecalled controls.

1. Forinstance, see HarveyS. Perlman, The Restatement of the Law of Unfair

Competition: AWork in Progress, 80TR. 461, 473 (1990)(“The issue

becomeshowstrong an inference of confusion can be drawnfromthe particularsurvey”).

Section IV

218

Thischapterwill explorehow“controls” areused toincreasethe reliability of

evidentiary surveys—a particularformofexperimenton human subjects.Using

specific examples,thischapterwill presentand discuss the prosand consof a

variety of control designsthatareregularlyused in experiments involving human

beings(including evidentiary surveys). The discussion of each design will include

the specific categoriesof legal problemsforwhich the specific design isnormally

applicable (e.g., likelihood of confusion between trademarks,falseadvertising, etc.).

THE MYTH OF REPLICATION

Surveys areused in Lanham Actlitigation toprovide evidence of whatwould happen

in the real world. Thus,intheory an optimummethodologywould simplyprovide

acompletereplication of the real-world reactionsand the behaviorof aprojectable

sample of the population of interest.

Thatbeing the case, whatdoesitmean toreplicatethe real world?Itisusefulto

considerthe intentof replication. The intentof replication isnotsimplytoduplicate

real-world conditions,but rathertoproduce datathatallows reliable prediction of one

ormorespecific real-world facts.Some (including anumberof judges)see thisasa

distinction without adifference. Such people areessentiallyclaiming (mistakenly,in

thisauthor’s opinion) thataccurateprediction of real-world behavior(which iscertainly

the goal) can come onlyfrom closeduplication of whathappensin the real world.

The underlying problem isthatbecauseinthe greatmajority of survey

methodologiesrespondents arefullyawarethattheyaretaking part in asurvey,in

most surveysituations,full replication of the real world isimpossible. Obviously,

responding toaninterviewer’s questionsisnothowconsumers ordinarilyfunction

when theyareconsidering buying something in the real world. Thus,while ideally,a

surveyaimsatbeing adirectreplication of potential buyers’ real-world behavior,iffor

no otherreason than thiscritical respondentawareness of being in asurvey,the great

majority of surveys areinherentlylikelytodifferfrom the real world.

Nonetheless,byusing controls,reasonable prediction of real-world behaviorcan

still be achieved. The goal of “controls” istoprovide an estimateofthe real-world

behaviorbyseparating the results of the experimentintotwocomponents:

1. whatwould happen in the real world, and

2.the deviationsfrom real-world behaviorthatarisewhen humansarethe

subjects of an experiment.

Therefore, judges—awarethatgood design cannoteliminate, but aswewill see, can

accountfor,the inevitable “surveyartifact” differencesbetween the “surveyworld”and

the real world—can (and should) concentrateonasking howclose the surveymethodology

including the useofcontrolscomesto“predicting”whatwould happen in the real world.2

2.See forinstance, MC T UC, Article 32:163 (4thed.

2005).