USF Libraries
USF Digital Collections

Media influences on response to a natural hazard

MISSING IMAGE

Material Information

Title:
Media influences on response to a natural hazard the Mississippi River salt water intrusion of 1988
Series Title:
Quick response research report ;
Physical Description:
18, 9 p. : ; 28 cm.
Language:
English
Creator:
Spencer, J. William ( James William )
University of Colorado, Boulder -- Natural Hazards Research and Applications Information Center
Publisher:
Natural Hazards Research and Applications Information Center, University of Colorado
Place of Publication:
Boulder, Colo
Publication Date:

Subjects

Subjects / Keywords:
Saltwater encroachment -- Louisiana   ( lcsh )
Disasters -- Press coverage -- Louisiana   ( lcsh )
Mass media and public opinion -- Louisiana   ( lcsh )
Risk perception   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Bibliography:
Includes bibliographical references.
Additional Physical Form:
Also issued online as part of a joint project with the Louis de la Parte Florida Mental Health Institute (FMHI) Research Library’s disaster mental health initiative.
Statement of Responsibility:
by J. William Spencer ... et al..
General Note:
Cover title.
General Note:
"Institute of Behavioral Science #6."
General Note:
"Also published as Working paper #4 of the Environmental Social Science Research Institute, University of New Orleans ..."

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001985161
oclc - 39295898
usfldc doi - F57-00045
usfldc handle - f57.45
System ID:
SFS0001126:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader cam 2200313Ia 4500
controlfield tag 001 001985161
005 20090223163810.0
008 980616s1990 cou b s000 0 eng d
datafield ind1 8 ind2 024
subfield code a F57-00045
035
(OCoLC)39295898
040
DDB
c DDB
d OCL
043
n-us-la
049
FHIM
086
UCB6/57.15/41
2 codocs
0 245
Media influences on response to a natural hazard :
b the Mississippi River salt water intrusion of 1988 /
by J. William Spencer ... [et al.].
260
Boulder, Colo. :
Natural Hazards Research and Applications Information Center, University of Colorado,
1990.
300
18, [9] p. ;
28 cm.
440
Quick response research report ;
v #41
500
Cover title.
"Institute of Behavioral Science #6."
"Also published as Working paper #4 of the Environmental Social Science Research Institute, University of New Orleans ..."
504
Includes bibliographical references.
530
Also issued online as part of a joint project with the Louis de la Parte Florida Mental Health Institute (FMHI) Research Librarys disaster mental health initiative.
650
Saltwater encroachment
z Louisiana.
Disasters
x Press coverage
Louisiana.
Mass media and public opinion
Louisiana.
Risk perception.
1 700
Spencer, J. William
q (James William)
710
University of Colorado, Boulder.
Natural Hazards Research and Applications Information Center.
4 856
u http://digital.lib.usf.edu/?f57.45



PAGE 1

HAZARDHOUSE ME,,,RES-POMSE TO A1... IR l""'TR."""'ONOF"9'8', -

PAGE 2

NaturalHazardsResearchandApplicationsInformationCenterCampusBox482UniversityofColoradoBoulder,Colorado80309-0482MEDIAINFLUENCESONRESPONSETOANATURALHAZARD:THEMISSISSIPPIRIVERSALTWATERINTRUSIONOF1988*ByJ.WilliamSpencerPurdueUniversityShirleyLaskaRuthA.SeydlitzElizabethTricheUniversityofNewOrleansQUICKRESPONSERESEARCHREPORT #41* 1990*AlsopublishedasWorkingPaper#4oftheEnvironmentalSocialScienceResearchInstitute,UniversityofNewOrleans,NewOrleans,Louisiana.This publication is partoftheNatural Hazards & Applications Information Center's ongoing QUick Response Research Report Series. http://www.colorado.edu/hazardsTheviews illl..!li;::'.Lc.tJU.L\.o.........'-..uC authorsandnotnecessarilythoseoftheNaturalHazardsCenterortheUniversityofColorado.InstituteofBehavioralScience#6 (303) 492-6818TELEFAX:(303)492

PAGE 3

MEDIAINFLUENCESONRESPONSESTOANATURALHAZARD:THEMISSISSIPPIRIVERSALTWATERINTRUSIONOF1988AbstractThispaperexamineshowattributesoflocalmediareportsmayeffectlocalresponsestoanaturalhazardevent.ThenaturalhazardofinterestistheintrusionofsaltwateruptheMississippiRivertotheNewOrleansareaduringthesummerof1988.Theindependentvariablesweredrawnfromanintensiveanalysisoflocalnewspaperarticlesandtranscriptsoflocaltelevisionnewscastswhichappearedduring,andreferenced,thehazardevent,whileresponsebehaviorwasoperationalizedasretailsalesofbottledwater.Usingtime-seriesanalysis.wethenexaminetheutilityofamodelofresponsebehaviorwhichstressesthatpeoole'sreactionstoanaturalhazardareafunctionofhowtheyusepersonalexperienceandtherelevanceofthehazardanditsconsequencestoselectivelyattendtocomplexandambiguousmediareporting.Theresultsindicatesomesimilaritiesanddifferencesintheeffectsofthetwotypesofmedia,whileprovidinglimitedsupportforthemodel.Futureresearchanddevelopmentofthetheoreticalmodelaresuggested.

PAGE 4

IntroductionTothesoclalsClentist.anaturalhazardismorethanJustaqeophysicalprocess.Anearthquakeinaremote,uninhabitedpartoftheworldisveryrealandltSphYS1Calattrlbutesandconsequencescanbemeasured.However,tothesocialscientistthesameeventwouldnotbeconslderedadisasterorhazardunlesslt"1nteractswithandaffectshumansystems"(SorensonandMileti1987.P.209).SocialSClenceresearchaboundsonmanyaspectsofdisastersincludingtheireffectsonsocialorganization.socialpsycholoqicalresponsesofvictims.mltlgationbehavior,andthelike.Animportantlinkbetweennaturalhazardsandthehumansystemswhichtheyaffectisthemassmedia.Traditionally,studiesofmassmediaanddisastershavetendedtofocusonorganizationalchangesinthemediaduringdisasters,inaccurateand"mythical"reportingpractices(NeedhamandNelson1977;Quarentelli1987),differencesininformationpresentedbydifferentsources(NeedhamandNelson1977;Alexander1980),andthepowerofthemediato"createdisasters"(KepplingerandRoth1980).Therehavebeenfewstudieswhichempiricallyexaminehumanresponsebehaviorsasafunctionofchangesincharacteristicsofmassmedlareportsduringthecourseofahazardevent.Thispaperisintendedasacontributiontotheliteratureinthisarea.Ourresearchquestionis:howdoesthepublicrespondduringthecourseofalocalhazardeventandtowhatspecificaspectsofthelocalmediacoverageoftheeventcantheseresponsesbeattributed?Ourstudyisbasedondatafromthedroughtofthesummerof1988.duringwhichsaltwaterfromtheGulfofMexicomoveduptheMississippiRiverthreateningthewatersUPPliesofcommunitiesandlndustryalonqtheriver.MEDIA. HAZARDS.ANDBEHAVIORThemassmedlahavebeenacentralfocusofmuchsoclalscienceresearchandtheoryonhazards.Themediacanreachlargeaudiences,providebasicinformationandcreateawareness(Rogers1987).Forexample,thetelevisionnewshasbeencitedasthemostimportantfirstsourceofinformationformanynewsevents(Larson1980).Inmostdisastersituations,fewpeoplehavehaddirectexperiencewiththedisaster-at-handandforthemajorityofpersonsthemediaaretheironlysourceofinformation(Carter1980;Wengeretal1975).Thus,itseemsreasonabletoconceptualizethemassmediaasanimportantfactorintheresponsesofhumansystemstonaturalhazards.Somearquethatthemedladirectlycreatessoclaldeflnltlonsofhazardsituatl0ns.BycontlnuallYpresentlnglnformatl0n

PAGE 5

aboutahazards1tuat10n."thepubl1Cbeqinstoperce1veasexperienC1nqadisaster"(RogersandSood1980.P.140):eventhoughmediaattentiongiventoaneventisoftenunrelatedtoits obJective severity(Adams1986).suchasnumberofdeaths(SingerandEndreny1987).AsRobinsonandLevy(1986,p.173)state:"publ1Cawarenessandcomprehensionofthenews...oftenbearsastrikingsimilaritytojournalisticallyencodedreality".Ifpeoplerespondtomediareportsofahazard.andtheattributesofmediacoveragedonotnecessarilyrelatetotheeventitself.itseemsreasonabletoconcludethatpeopleoftenrespondmoretothesocialrealityconstructedbythemediathantothereal1tyofthehazarditself.ThisconclusionissupportedbYKepplingerandRoth(1979),whoreportontheanalysisofdataconcerningtheeffectsofthemassmediaonpeople'sperceptionsoftheoils1tuationof1973intheFederalRepublicofGermany.Ininterpretingtheirfindings.KepplingerandRoth(1979)suggestthatthemassmediadirectlyinfluencedthedevelopmentofsocialreality.Mediareportscreatedthe1mpressionofanoilshortage.towhichthepublicrespondedbYincreaseddemand.creatingashorttermshortaqe.Ifthemassmed1acreatean1maqeofahazardcongruentwiththereal1tyofthehazard1tself.individualsreacttoreportingastheywouldtoreal1ty.However.1fafalseor1nadeQuateimageiscreated.individualsreactdifferentlythantheywouldtoreality.Themoretheindividual'sconceptionofrealitydependsonthereportingofthemassmedia,themorelikelYthisistohappen.It1Sprobablyunrealistic.however.toassumethatthemassmediawouldbeabletosinglehandedlypersuadeviewerstoadoptaparticularviewpointoraparticularbehavior.Themassmediamorelikelyinfluenceperceptionandbehaviorindirectly(Rogers1987)oritseffectsinteractwithothervariables.Specifically,attributesofmassmediahazardreportingwouldseemtointeractwithothervariablesinproducingdefinitionsofthehazardeventand,inturn,publicresponses.Theseothervariablesinclude.butarenotlimitedto,interpersonalcommunicationnetworks,individualattributes(e.g.,previousexperienceandpriorattitudes).andhumancognitiveandmemoryprocesses.Thelackofasimpleanddirecteffectofthemediaonbehaviormaybeduetohowtheaudiencerece1vesandinterpretsthemediareports.Humansareprocessorsoflimitedinformationandmemoryresearchshowsusthatwhenpresentedwithinformation.peoplestorecognitiveschemawhichcomprisealimitednumberofdetailsinaprocesstermed"leveling"or"flattening"(Staufferetal.1983).Thistendencytowardselectivityisstrengthenedbythenatureofsomemedia."StudY.repetition.connectiontopriorknowledgeandelaborationwithadditionalredundantinformationallfacilitaterecall"whilethe"rapidlymovingandnon-repetitiveformatoftelevisionnewsisnot2

PAGE 6

conducive"totheway weprocess1nformation(Staufferetal.1983.P.35).Sincerecallisrelatedto,amongotherfactors.personalrelevance,wemightreasonablyassumethatthefewdetailsthatarereta1nedarethosethatrelatetopeoples'specificneedsandrisks.Forexample,theamountofspaceortimeallocatedtoastorYisoftennotproportionatelyrelatedtotheimportanceplacedonthestorybyviewers.Rather.respondentsaremorelikelYtoplace1mportanceonandrecallmorefreauentlythosestoriesthatrelatedtotheirpersonall1vessuchasthoseofrealorpotentialdanqerorhuman1mpact(Larson1980:RobinsonandLevy1986:Staufferetal.1983).Turner(1980)hasalsofoundthatindividualshaveonlya"rathervagueandlimitedawarenessofthemedia'spredictionsandannouncements..."andthatthepublicformsanoverallimpressionofasituationoreventratherthantoretainanydetailsofannouncementsorpredictionsthatledthemtothatimpression.(Turner1980,p.291).Theliteratureonriskperceptionalsoprovidescluesontherelationshipbetweenmediahazardreportingandpublicresponsesbyfocusingonthedecision-makingprocesseswhichpromptresponsebehaviors.Slovicetal.(1987)suggestthathumansare"boundedlyrational"anduseimperfectmethodstomakesenseoftheirenvironment.Asimpleoridealriskdecision-makingmodelcanbedescribedasal1nearprocesscomposedofassessingtheprobabilityorsusceptibilityofthehazard,'reviewingavailablemitigatingbehaviorsandtheirefficaciousness.evaluat1ngcostsorbarrierstoprotectiveactions,receivingcuestoaction(suchassymptomsofthehazardormasscommunicationinformation),andfinallymakingadecisiontoact(Cleary1987:MiletiandSorenson 1987), However.it1Sgenerallyagreedthatind1v1dualsarenotthisrational.Inadequateorincompleteinformationandpreconceivednotionsaboutr1sksandprecaut10nscanaltertheoutcomeofthedecision-makingprocess(MiletiandSorenson1987).Thesepreconceivednotionscand1storttheperceptionofinformationalmaterial.Sincenewinformationisinterpretedbyarticulatingitwithpreviouslyacquiredknowledge,Wh1Chexistascognitiveschema,Ifanindividualhasformedstronginitialimpressionsaboutahazard,resultsfromcognitivesocialpsychologysuggestthatthosebeliefsmaystructurethewaysubsequentevidenceisinterpreted.Newevidencewillappeartobereliableandinformativeifitisconsistentwithone'sinitialbeliefs;contraryevidencemaybedismissedasunreliable,erroneous,orunrepresentative(Slovicetal.1987,p.36).3

PAGE 7

Likeanyotherformofsoclalperceptl0n,riskassessment1nvo1vesbothob.lectiveaswei1assub.lect1vee'ements(JohnsonandCovello1987).As aresult,peopleusuallydonotcorrectlyestimaterisksandshowadefinitepropensltytounderestlmateorevendenYrisk(MiletiandSorensen1987:Sandmanetal1987).Thisresults.inpart.fromthefactthatpeoplerarelyreceive"seriousinstructionintheassessmentofrisks"(Slovicetal1987.p.19)andmustthereforerelyonmediaaccountsandpersonalexperience.Previouspersonalexperiencewithahazardandattributesofmediareports(e.g.,expertnessofsource,frequencyofwarnings.detailofinformation)interactineffectingriskassessmentand.inturn,responsebehavior.AccordingtoRuchandChristensen(1981),personswlthnoprlorexperlencewlthhurrlcaneswerebestmotivatedbyawellrespectedauthorityfiqure,whilepersonswithpriorexperiencewerebestmotivatedbYindlcatorsWhlCh.throuqhtheirpriorexperience.theyassociatedwiththeserl0usnessofthestormthreat.Leiketal.(1981)foundthatlnexperlencedpersonstakedefensiveactionsoonerinthefaceoffrequentanddetailedWarnlnqs.whileformoreexperlencedpersons.detalledandfrequentmessagesproducedlonqerdelaysthandidincompleteorinfrequentmessaqes.Attrlbutesofmediareportsalsolnteractwithinterpersonalcommunlcationtoeffecttheriskassessmentprocess.Theattributesof,anddangersposedby,most razard situationsareambiguousandthemediaoftenexacerbatesthisambiguitybyitsreportingpractices.Inanattempttoavoidmasspanic,themedia"almostinvariably...combinewarningswithreassurancesthattendtounderminetheseriousnessofthethreat"(Turner1980,p.284),resultinginamixedmessage.Peopleoftenlooktotheirpeersforcuesandinformationtohelpreducetheambiguity(Rogers1987;Weinstein1987;Nigg1982)andmoreclearlydefinethehazardsituationandchooseamongreponsealternatives.Unavailableinformation,questionsaboutrelevancetopersonalsituationsandunknowneffectivenessofproposedprecautionsoftenleadpeopletofollowneighbors'actions,"assuming(oftenincorrectly)thattheneighborshavemadeacarefulstudyofthesituation"(Weinstein1987,p.330).Furthermore,themedia'sattempttoaddentertainmentasafeatureofthenewsmayleadtoincreasedbeliefininformationthemediaactuallyintendstodownplay(Turner1980)becauseoftheperceivedambiguity.Weoffertwogeneralobservationsontheliteraturereviewedabove.First,therehasbeenlittleempiricalresearchWhlChdirectlyexaminestherelatioshipbetweenpublicresponsesandthemedia'sreportingofdisaster-relevantinformationduringtheactualcourseofthehazardevent.Someobservershavenotedthisrelativelackofempiricalresearch(seeforexample,Kreps1980:LangandLang1980;Larson1980)andhave,inturn,noted4

PAGE 8

theneedforsuchresearch.Sincebothmediareportingandpublicresponsesundergochangesduringthecourseofahazardevent.ltwouldseemthatareasonableresearchdesignshouldincludeallphasesofahazardeventlnordertoexamlnetheconcommltantchangesinbothmediareportingandpublicresponses.Second,therehavebeenfewattemptsatbuildinga ofpublicresponsestomediareportsduringhazardsituations.Nigg's(1982)workrepresentsonesuchattempttodevelopageneralmodel,focusingonmanycharacteristicsofmediareportingandpublicresponsebehaviorwhichhavebeennotedabove.Accordingtothemodel,therearespecificcharacteristicsofmediareporting(e.g..ambiguityordiscrepanciesininformation)whichleadtouncertaintyonthepartofthepublic.As aresultofthisuncertainty.thepubllCattemptstoselectlvelyinterpretmediareportsofthehazardeventvis-a-visitspersonalrelevance.Whenahazardeventisdefinedasposingrelevantconsequences,thepublicisseenasagainlookingtothemedia'spresentationofpossibleresponsealternatives.ThisstudyisdesignedtocontributetotheliteratureintheareabyaddressingthesetwopOlntsofresearchandtheoryonmediaandhazards.Utilizingtheextantliterature,wefirstdevelopageneralmodelofpublicresponsestomediahazardreports.Weusethismodeltogeneratespecifichypothesesconcerningtherelationshipsbetweenattributesofmedlareportsandpublicresponses,andhowtheserelationshipschangeduringthecourseofthehazardevent.FinallY.usingatimeseriesanalysis,weexaminethesehypothesesvis-a-visdataonasinglehazardevent.AMODELOFPUBLIC RESPONSESTOHAZARDREPORTINGOnthebasisoftheextantliteratureonmediaandhazards.weconstructthefollowingmodelofresponsebehaviors.First,mediareportsofmosthazardeventsarelikelytobeambiguous,oftenpresentingconflictinginformation.Sincehazardeventsoftenevokeanxiety,theambiguityofmediareportswouldbeexpectedtoheiqhtenthatanxiety.Second.howpeoplerespondtoahazardeventlSafunctionofhowtheydetlneIt.Thesedetlnltlonsare.lnturn.toaqreatextentafunctl0nofmedlareportsratherthantheattrlbutesoftheeventltself.Whilemediareportsprovldea maJor. lfnotthesole,sourceofinformationaboutahazardevent,theambigultythatcharacterizesmediareportingmakesthisdefinitionalworkacomplicatedendeavor.Thisambiguityturnsonseveraldimensions.First.mediareportsoftencombineinformationwhichstressesthelikelihoodofthehazardeventand/oritsnegativeconsequences.Second,mediareportsalsocontaininformationaboutvariousconsequencesofthehazard,somedirectlyrelevanttopeoples'everydaylives,otherslessdirectlyrelevant.5

PAGE 9

Sincepeoplecannotanddonotprocessalltheinformationcontainedinmediareports,theyselectivelyattendtothem.Thisselectivityistwo-fold.First.peoplemayselectivelyattendtospecificpartsofthecontentofreports;theymayattendtosomebutnotother pieces ofinformation.Second,theymaylistentoorreadreportstogleananoverallimageorgestaltofthehazardevent.Thespecifickindofselectivitypeopleuseismostlikelyafunctionofpreviouslyacquiredknowledgeaboutthehazardevent,whichpeopleusetointerpretmediainformation.Aspectsofmediareportswillinteractwiththeseothervariablestoinfluenceresponses;givendifferentconfigurationsofbackgroundknowledge,differentaspectsofmediareportswillhavedifferenteffects.Theinteractionsbetweenspecifictypesofbackgroundknowledgeandspecificaspectsofmediainformationisprobablycomplex,butwecanproposethefollowing.First,intheabsenseofpriorexperiencewithahazardevent.peoplewi11nothighlydiscrlminatebetweenpersonalIvrelevantconsequencesofahazardandother.lessrelevantconsequences.Whenpeoplehavepersonalexperiencewithahazard,theywll1bemoredlscriminatlng:themorerelevanttheconsequence.themoreactlvetheresponseswl1Ibe.Undertheconditionofalackofpersonalexperlence.amblQUltvconCernlnQtheseverltyorl1kelihoodofthehazardwillhavethesameeffectsasreportsWhlChmoresimplystresstheseverltyor11kellhoodoftheevent.Inshort.peoplewl11attendlesstothespecificcontentofthereportwhentheylackpersonalexperience.Underthecondltlonofpersonalexperience.peoplewi11bemoredlscrlmlnat,nQ,ntheirattentlontothedetailsofthecontent.Intheabsenseofpersonalexperienceandinthefaceofambiguousmediainformation,twofactorswillinfluencepeoples'responses.First.peoplewillutilizepersonalexperiencewithotherhazardswhicharesimilartothepresentone.Responseswhichhaveprovenefficaciousindealingwithaprevious,relatedhazardwillprovidepeoplewithalternatives.Inadditionto,orinlieuofexperiencewithrelatedhazards,peoplewilllooktootherstohelpdefinethehazardevent.Thismostlikelytakesplaceinthecontextofinterpersonalcommunicationnetworks.Inadditionto,orinlieuof,thesecommunicationnetworks,mediareportsofothers'responseswillservethefunctionofdefiningpossibleorappropriateresponses.Others'responseswillhavetheeffectofprovidingamodelforresponsesand/ordefiningthehazardevent.Undertheconditionofpersonalexperience,others'responseswillhavelessofaneffect.THEHAZARDEVENTThehazardeventofinterestlnthisstudyoccurredduringthesummerof1988.ThemidwestUnitedStateswasexperiencinga 6

PAGE 10

lengthyandseveredroughtwhich,amongotherthings,wasproducinglowwaterlevelsanddecreasedflow1ntheMississippiR1veranditstributaries.Thisdecreaseinthewaterflow,inturn.allowedsaltwaterfromtheGulfofMexicotomove uptheMississippiRivermuchfartherthanitusuallydoesdur.inglowwater.TheresultingsaltwaterconcentrationsposedathreattocommunitiesandindustrieswhodependedontheMississippiforwatersuppl1es.ThesaltwaterintrusionisanannualeventduringlateAugustandSeptemberwhentheriverflowistypicallylow.However.theintrusionistypicallylimitedtocommunitiesadjacenttothemouthoftheriverandtheconcentrationsofsaltaregeneralllylowenoughthatstandardtreatmentprocedurescanremoveit.Inshort.itgenerallydoesnotposeahazardtocommunitiesalongtheriver.Thisoarticularsummer,however,theseveredroughtcoupledwitharecentdredgingoftheriverbytheCorpsofEngineers.allowedthesalttomovesignificantlyfartheruptheriverandinincreasedconcentrationsthanwastypicallythecase.Anintrusionofthismagnitudehadnotoccurredsince1937,whennewspapersreportedpeoplecatchingsaltwaterfishoffthedocksindowntownNewOrleans.Thecurrenteventcreatedahazardsituationthatthreatenedmunicipalwatersupplies,industrialequipmentandmachineryasfarupriverasNewOrleans,ametropolitanareaofapproximatelyonemillionpersons,onehundredmilesfromthemouthoftheriver.Thereareseveralattributesofthiseventthatmakeitparticularlyinterestingtostudyinthecontextofthemediaandhazardevents.Althoughanannualevent,sincethesaltrarelyintrudedasfarupriverasNewOrleans,fewpeopleinthemetropolitanareahaddirectexperiencewiththehazard.Accordingtotheliterature,thiswouldmeanthatmostpeopleinthemetropolitanareawouldrelyalmostexclusivelyonthemediaforinformation.Inaddition,theeventlastedapproximatelytwomonthsduringwhichtimeitoccupiedarelativelyprominentplaceinboththeprintandelectronicmedia.Further,theintrusionallowedforlittlepredictiveability;officialsseemedunabletounequivocallYstatehowseverelythehazardwouldimoacttheNewOrleansarea:howhightheconcentrationsofsaltwouldreach.whetherornotitwouldposeahealthhazard.whetherornotpeoplewouldbeabletotastethesaltintheirtapwater.Infactthemediareportswerefilledwiththekindsofambiguitiesdiscussedaboveintheliteraturereview.Theeventwasbignewsinthelocalmediaandreceivedmuchtimeandspace;ontheotherhandthereportswerefilledwithstatementswhichalsodownplayedtheseverityandsignificanceoftheevent.Forexample,whilesmallercommunitiesdownriverwereabletobargeindrinkingwater,theNewOrleansarea,beingsolargeanddenselypopulated,hadnosuchoption.Ineffect,NewOrleansfaceda7

PAGE 11

threattodrinkingwatertoWhlChtherewerefewviablemltiqationresponses.ThlSfactreceivedmuchplayinthemedlareportswhllethesamereportsalsoincludedreportswhichstatedthattheeventwouldmostlikelynotlmoacttheareainanyslgnlflcantway.Inadditiontoallofthis,theNewOrleansarea(andinfact,mostofsoutheasternLouisiana)hasexhibitedagrowingconcernwiththequalityofdrinkingwater.SincethemetropolitanareadrawsitsdrinkingwaterfromtheMississippiRiver,concernsoverthepresenseofindustrialandtoxicwasteinthewaterhavetraditionallygeneratedhighlevelsofbottledwatersales.ThisconcernisheightenedbythefactthatalongtheRiverbetweenBatonRougeandNewOrleansthereisaseriesofchemicalplants.Withintheprevioustwelvemonths,GreenpeacehaddemonstratedatseveralplacesalongtheriverbetweenBatonRougeandNewOrleans.Thelocalandnationalmediahasrecentlybeenfocusedonthecancerratesinthearea,reportedlyseveraltimesthenationalaverage.Theslowonsetofthehazard,itsrelativeunpredictability,theambiguitywhichcharacterizedthelocalmediareports,andthelackofresponseoptionsbeyondpurchasingbottledwater,coupledwiththelackofdirectexperienceonthepartofthepublicallowsforauniqueopportunitytoexaminechangesinthecontentofmediareportingandchangesinresponsebehavior.DATAThedatacamefromthreesources.WeusedrecordsofretalIsalesofbottledwaterbetweenJune17thandAugust14thfromthewatercompanywiththelarqestmarketsharelnNewOrleansatthetlmeofthesaltwaterlntrusl0n.Also.weexaminednewsstoriesfromtheonlyNewOrleansnewspaper.TheTimes-PicaVune,andthetelevisionstationwiththelargestmarketsharelnthesummerof1988.OnlystoriesconcerningthelowleveloftheMississippiRiverwereanalyzed.Thedependentvariablewastheretailsalesofbottledwaterinnumberofgallonsperday.Thecompanydoesnotdeliverwatertostoressevendaysaweek;thusthereweredayswhensaleswerezero.Linearinterpolationwasusedtocalculateestimatesofsalesforthesedays.Anaturallogarithmictransformationofthesaleswasnecessary.Thevariabilityofthesalesfigureswasgreatestwhenthesaleswerehighestandtheserieswaspositivelyskewed.1 1Thesalesfiguresusedinthisstudyareconfidential.Weareunabletoreportdescriptivestatisticsforthisvariableortoincludeplotsoftheseries.8

PAGE 12

SiXindependentvariableswereused.Proximitywasa dummyvariablecoded0beforethesaltwaterwedgearrivedinNewOrleansand1afterthewedgearrived.Mediamentionsoffivetopicsrelatedtotheincreasedsaltcontentintheriverwerecountedforeachday.Toxinswasthenumberofmentionsinthestoriesconcerningthepotentialhazardofincreasedlevelsoftoxinsinthetapwater.Saltwasthenumberofmentionsofthesaltcontentinthetapwater.Healthwasthenumberofmentionsofhealthconsequencesduetodrinkingthetapwater.Tastewasthenumberofmentionsconcerningthetasteofthetapwater.Buyingwaterwasthenumberofmentionsofpeoplebuyingbottledwater.Mentionsofthesetopicswerecalculatedseparatelyforthetwomedia,thenewspaperandthetelevision.METHODSANDRESULTSBeforeexaminingtheeffectoftheindependentvariablesonsales,wedidaninterruptedtimeseriesanalysisofthesalesusingtheARIMAprocedureinSPSS-PC+.Weusedthefirstdifferencesofthenaturallogarithmsbecausetheserieswasnotstationary,butnofurthermodelingwasnecessary.Wefoundninesignificantchangesinthesales.FoursignificantincreasesoccurredbeforethesaltwaterwedgearrivedinNewOrleansandtwowereduringtheweekimmediatelypriortothearrivalofthewedge.Fivesignificantchangesoccurredafterthewedgearrived.Fourweredecreasesandtwowerewithinninedaysafterthearrivalofthewedge.Theincreaseoccurredtwoandahalfweeksafterthewedgearrived.Next.weusedmultipleregressiontoanalyzetheeffectoftheindependentvariablesonsalesofbottledwater.Sevenregressionequationswereusedforeachindependentseries.First,thenaturallogarithmoftheinterpolatedsalesserieswasregressedontheindependentseries.Forthenewspaper,theindependentserieswasthesamedaymentionsofhealth,taste,toxins,saltandbuyingwater.Forthetelevision,theindependentserieswasthementionsthepreviousday.Second,thesalesserieswasregressedontheindependentseriesandproximity.Third,thesalesserieswasregressedontheindependentseries,proximityandtheirinteraction.Fourth.thesalesserieswasregressedontheindependentseriesandthelagoftheindependentseries.Forthenewspaperdata.thelagoftheindependentserieswasthepreviousday'smentionsofthemediatopics.Forthetelevisiondata.thelagoftheindependentserieswasmentions2dayspreviously.Fifth,thesalesserieswasregressedontheindependentseries.thelagoftheindependentseriesandproximity.Sixth.theinteractionofproximityandtheindependentserieswasadded.Seventh.theinteractionofproximityandthelagoftheindependentserieswasaddedtotheequationcontainingtheindependentseries,thelagoftheindependentseries,proximity,andtheinteractionofproximityandtheindependentseries.9

PAGE 13

TheARIMAprocedureinSASwasusedtoanalyzetheresiduals.Theseanalysesshowedthatnoneofthemodelsabovefitthedatabecauseanautoregressivetermwasomitted.Thus,themodelspresentedabovewererunasecondtlmewlththelagofthedependentvariableincludedtoaccountfortheautoregressiveterm.2Theanalysesoftheresidualsshowedthattheerrorswererandomandnospikesoccurredineithertheautocorrelationsorthepartialautocorrelations.Thebestmode]wasdeterminedbYcalculatlnganFtestfortheincrementto R' fortheadditionofvariablesintothemodel,whenposslble,andbycomparingtheadJustedR 2 ,themeansquareerrorandthepressstatlsticacrossequatlons.ThefinalmodelsshowninTable1aresignificantlydifferentfrommoreparsimoniousmodelsandtheadjustedR 2 sarehigherwhilethemeansquareerrorsandpressstatisticsarelower.Thefinalmodelsforthenewspaper(seeTable1,panelA)indicatedthatmentionsofhealthconsequencesaresigniflcantlyrelatedtoincreasesinsalesofbottledwater.Also,mentionsofpeoplebuyingbottledwatersignificantlyincreasesales.Further,therewasasignificantinteractionbetweenmentionsconcerningthetasteoftapwaterandproximityofthesaltwaterwedge.Thisinteractionindicatedthatmentionsconcerningtastesignificantlyincreasedsalesofbottledwateronlybeforethewedgearrived.Mentionsinthenewspaperofsaltandtoxinlevelslnthetapwaterhadnoeffectonsales.Whenthemodelsforhealth,buyingwaterandtastewerecombinedinoneequation,themodelwassignlflcantbutnoneofthevariableswere.Thisresultwasprobablyduetomulticollinearitysincethevariablesweresignificantlyintercorrelated.Also,thesetopicstendedtobementionedonthesameday.ThevarianceexplainedbYcombiningthemodelswasnotsignificantlygreaterthanthevarianceexplainedbytheseparatemodels.Further,thestandardizedregressioncoefficientsindicatedthatnoindividualmediatermwasmoreimportantthantheotherterms.Theresultsweredifferentforthetelevision(seeTable1,panelB).Mentionsconcerningthetasteoftapwateronthepreviousdaysignificantlyincreasedsalesandsalesweresignificantly(.10level)greaterbeforethewedgearrived.Mentionsofpeoplebuyingwateronthepreviousday'snewscastsignificantlyincreasedsalesandsalesweresignificantly(.10level)greaterbeforethewedgearrived.Therewasasignificantinteractionbetweenproximityandmentionsofhealthconsequenc-2ThisuseofthelagofthedependentvariableisdemonstratedintheSASapplicationsmanual,SASSystemforForecastingTimeSeries.10

PAGE 14

es.Mentionsofhealthconsequencessignlflcantlyincreasedsalesonlybeforethewedgearrived.Themodelformentionsofsaltindicatedthatmentionsofsaltinthetaowateronthepreviousdaynonsignificantlyincreasedsaleswhilesuchmentionstwodayspreviouslysignificantlydecreasedsales.Thismodelsuggeststhatmorewaterthanexpectedbythetrendwaspurchasedthedayafterthestory;thussalesfelltheseconddayafterthenewcast.Whenmostofthestoriesconcerningsaltcontaminationofthedrinkingwateroccurred.thesalesofbottledwaterwereincreasing.Thus,thenewscastswouldhavehadtoincreasesalesdramaticallytohaveasignificantposltiveeffect.butthedecreaselnsalestwodaysafterthenewscastwouldnothavetobeaslargefortheeffecttobesignificant.Thus.thesignificantnegativeeffecttwodaysafterthenewscastsprobablyrepresentsarecoveryinthesalesseriesfollowingadaywhenthelevelwashigherthanexpectedbythetrendinsales.Mentionsoftoxinswereunrelatedtosales.Whenthemodelsfortaste.buyingwater.healthandsaltwerecombinedinoneequation.themodelwassignificantbutnoneofthevariableswere.Thisresultwasprobablyduetomulticollinearitysincethevariablesweresignificantlyintercorrelated.Thetopicswereusuallymentionedonthesameday.Thevarianceexplainedbycombiningthemodelswasnotsignificantlygreaterthanthevarianceexplainedbyeachtopicseparately.However,thestandardizedcoefficientsindicatedthatthetermsforthehealthmodelweremoreimportantthantheothermediaterms.Toexaminefurtertheeffectsoftheindependentvariablesonsales,weanalyzedthetoneofthementionsoftoxins,taste,healthandsalt.Thesementionswerecodedascertainthattheconsequencewouldoccur,uncertainthattheconsequencewouldoccur,andnotlikelythattheconsequencewouldoccur.Thelattermentionswerethosethatdiminishedtheprobabilityoftheconsequence.Forthenewspaperdata,themostfrequenttoneforthevariablesthataffectedsaleswasdiminishment.Forthetelevisiondata,themostfrequenttoneforthevariablesthataffectedsaleswascertainty.Totestthehypothesisthatthetoneofthementionsisimportant,threenewserieswerecreated:certainty,uncertainty,anddiminishment.Mentionsthatwerecertainabouttheconsequenceweresummedforeachdayforeachmedium.Similarly,mentionsthatwereuncertainweresummedaswerementionsthatdiminishedtheprobabilityoftheconsequence.Weregressedsalesonthethreevariablesforeachmediumusingthesevenequationsthatwereusedpreviously.Theregressionresultssupportedtheideathatthetoneof11

PAGE 15

thementionsisimportant.Newspapermentionsthatdiminishtheconsequencesignificantlyincreasesalesofbottledwater(seeTable2,panelA).Newspapermentionsthatarecertainoruncertainabouttheconsequencedidnotaffectsales.Televisionmentionsthatarecertainthattheconsequencewilloccurwererelatedtosalesthenextday(seeTable2,panelB).Theeffectofthesementionsinteractedwithproximityindicatingthatthesementionssignificantlyincreasedsalesbeforethewedgearrived.Televisionmentionsofthehazardsthatwereuncertainorthatdiminishedtheprobabilityoftheconsequencedidnotaffectsalesofbottledwater.Totestthehypothesisthattheeffectofmentionsofsalt,health,taste,toxinsandbuyingwaterdidnotvarybytheplaceofthemention,threeserieswerecreatedforeachmedium:NewOrleansmentions,Plaqueminesparishmentions,andmentionselsewhere(SoutheasternandNorthernLouisianaandplacesoutsidethestate).Foreachday,thementionsconcerningNewOrleansweresummed.Similarly,thementionsconcerningPlaqueminesparishweresummedaswerethementionsconcerningotherplaces.Thesalesserieswasregressedonthesethreeindependentseriesforeachmediumusingthesamesevenequations.Theresultssupportedthehypothesisthattheeffectofthementionsdidnotdependontheplaceofthemention.Forthenewspaperdata,allthreeseriesaffectedsales(seeTable3,panelA).TherewasasignificantinteractionbetweenproximityandmentionsconcerningNewOrleans.MentionsconcerningtheNewOrleansareasignificantlyincreasedsalesofbottledwateronlybeforethewedgearrived.TheresultsweresimilarformentionsconcerningPlaqueminesparish.MentionsconcerningplacesoutsideSoutheasternLouisianasignificantlyincreasedsalesbothbeforeandafterthewedgearrived.Forthetelevisiondata.theeffectswerestatisticallysignificantatthe.10level(seeTable6,panelB).MentionsconcernlngtheNewOrleansareaincreasedsalesofbottledwaterthenextday.Also,inthemodelforNewOrleansmentions,proximityindicatedthatsalesweregreaterbeforethewedgearrived.Further,mentionsconcerningPlaqueminesparishincreasedsalesthenextday.Wehypothesizedthatpeoplemorereadilyrespondtoreportsthatcitepersonallyrelevantconsequences.Thenewsstoriesincludeddiscussionsofconsequencesthataffectedbusiness,suchasequipmentdamage,andcollectiveresponsestothesaltwaterwedge,suchasbargingwater,dredgingtheriver,buildingoraddingtoalevee.Threenewserieswerecreated:mentionsofbargingwater,mentionsofbusinessconsequences,andmentionsofothercollectiveresponses.Contrarytothehypothesis,mentionsofthesetopicsalsoaffectedsales(seeTable4).Forthenewspaperdata,mentionsofbargingwatersignificantlyinteract-12

PAGE 16

edwithproximityandsalesofbottledwater(seeTable4.panelA).Mentionsofbarginqwatersigniflcantlyincreasedsalesbeforethewedgearrived.Theinteractionbetweenmentionsofotherresponsesandproximitywassignificantatthe.10level:thusthesementionshadastrongerpositiveeffectonsalesbeforethewedgearrived.Whenthemodelsforbargingwaterandotherresponseswerecombinedintooneequation,noindividualmediatermwassignificantlyrelatedtosales,althoughthemodelwassignificant.Thisresultwasprobablyduetomulticollinearity.Thevarianceexplainedbythecombinedequationwasnotsignificantlygreaterthanthevarianceexplainedforeachtopicseparately.However,thestandardizedcoefficientsindicatedthatbargingwaterwasmoreimportantthanotherresponses.Forthetelevisiondata,mentionsofbargingwater,businessconsequencesandotherresponsestothesaltwaterwedgeaffectedsalesofbottledwater(seeTable4,panelB).Therewasasignificantinteractionbetweenproximityandmentionsofbargingwater.Mentionsofbargingwateronthepreviousday'snewscastsignificantlyincreasedsalesbeforethewedgearrived.Mentionsofbusinessconsequencesthepreviousdaynonsignificantlyincreasedsaleswhilementionstwodayspreviouslysignificantlydecreasedsales.Mentionsofotherresponsesthepreviousdaynonsignificantlyincreasedsaleswhilementionstwodayspreviouslysignificantlydecreasedsales.Thesetwomodelsaresimilartothemodelfortelevisionmentionsconcerningsaltandsuggestthatmorewaterthanexpectedbythetrendinsaleswaspurchasedthedayafterthestory;thussalesfelltheseconddayafterthenewscast.Again,mostofthesestoriesoccurredwhensaleswereincreasing;thustheexplanationforthenonsignificantincreasethedayafterthebroadcastandthesignificantdecreasetwodaysafterthebroadcastisthesameasthatpresentedearlier.Whentheseparatemodelsforbargingwater,businessconsequencesandotherresponseswerecombinedintooneequation,bargingwatersignificantlyaffectedsaleswhilebusinessconsequencesandotherresponsesdidnot(seeTable5).Mentionsofbargingwateronthepreviousday'stelevisionnewscastssignificantlyincreasedsalesofbottledwaterbeforethewedgearrivedinNewOrleans.Thevarianceexplainedbythecombinedequationwasnotsignificantlygreaterthanthevarianceexplainedbyeachseparatemodel.Itispossiblethatbusinessconsequences,bargingwater,andotherresponsestothesaltwaterwedgeweresignificantlyrelatedtosalesbecausetheyaresignificantlycorrelatedwithpersonalconsequencesandresponses.Mentionsoftheseeighttopicswerehighlyintercorrelatedandtendedtooccuronthesamedays.Furtherthepresenceorabsenceofmentionsofthese13

PAGE 17

topicswerehighlyinterrelated.Todeterminetheeffectofeachtopicnetoftheothertopics,thesignificantmodelswerecombinedforeachmedlumseparately.Forthenewspaperdata.thesalesserieswasregressedonthemodelsforhealth,taste.bUYlngwater.bargingwaterandotherresponses.Noindlvidualtermformentionswassignificant.Thiswasprobablyduetotheintercorrelationamongthesetopics.WhlChwasdiscussedearlier.ThlScombinedmodelexplainedasignificantlYgreaterproportionofthevarianceinsalesthandidthemodelsforbuyingwater.bargingwaterorotherresponses.Thevarlanceexplainedbythecombinedmodelwasnotsignificantlygreaterthanthevarianceexplalnedbytheseparatemodelsforhealthandtaste.ThisresultsuggeststhathealthandtastearethemostimportanttOP1CSforaffectlnQsalesofbottledwaterforthenewspaper.Theanalogousanalysiscouldnotbecompletedforthetelevisiondata.Whentheseparatemodelsfortaste,buyingwater,health,salt,bargingwater,businessconsequencesandotherresponseswerecombinedintooneequation,theinteractionofproximityandbargingwaterwasalinearcombinationofhealth,bargingwater,andtheinteractionofproximityandhealth.Forthetelevisiondata,healthwasthemostimportantofthemorepersonallyrelevanttopicsandbargingwaterwasthemostimportantofthelesspersonallyrelevanttopics.Bargingwaterismorepersonallyrelevantthanarebusinessconsequencesandotherresponses;thusitappearsthatpersonalhazardsandconsequenceshaveagreateraffectonsalesofbottledwaterashypothesized.DISCUSSIONTheresultsofouranalysisareadmlttedlycomplex.sothisdiscussionwillcenteraroundtheextenttoWhlChtheresultssupportorfailtosupporttheconceptualmodelofferedabove.Inourmodelofresponsebehavlor.weassumedthatmediareportlngwouldoftenbecomplexandambiguous.Whenpeoplelackedpersonalexperlencewlththehazardevent.thisamblqultywouldleadtopeopletoreactqUlckerandmoreactlvelyascomparedtowhentheydldhavesuchpersonalexperlence.Inthepresentcase.thiswouldmeanthatretailsalesofbottledwaterwouldbehlgherbeforethesaltwaterwedgearrivedatNewOrleans(July6th)thanafteritarrived.Severalfindingssupportthisexpectation.First.withonlyoneexception.thesignificantbreaksinretallssalesbeforearrivalwerealllncreases.whilethoseoccurring after werealldecreases.Inaddition.retailsalespeakedonJuly6threpresentingalarge,butnonsignificantincreasefromJuly5thtoJuly6th.Sincetherewerethreesignificantincreasesinretailsalesinthe14

PAGE 18

tourdaysprlortothislastincrease.ltwouldhadtohavebeenahuoelncreasetobeS1Qnlficant.TheexceptlonwasanlncreasebetweenJuly25th-July26th.Thedatawehavedonotallowustoassessthereasonforthisanomaly.Further.lnfourofthespecificmodels.proxlmityindicatedthatsalesofbottledwaterwereqreaterbeforethewedgearrived.Thisprovidesindirectsupportforthemodel.Beforethearrivalofthewedge,mostresidentsdldnothavedirect,personalexperiencewithasaltwaterintrusionandthedataindicatethattheyrespondedbybuyingbottledwater.However.afteritsarrival.thedecreasesinsalessuggestthat,onthebasisoftheirdirectexperience,peopledidnotassessthehazardtobesufficientlyseriousorthreateningtowarrantbuyingwater.Ofcourseourdatadonotallowustodirectlyexaminewhetherthiswasduetochangesinthelevelofanxietycreatedbytheconditionofpersonalexperience.Themodelalsoconceptualizedresponsestoahazardeventtobeafunctionofhowpeopleselectivelyattendedtomediareportsofthatevent.Thisselectivitywashypothesizedtooperateinoneormoreofseveraldirections.First,themoreseriousorthreateningthehazardanditsconsequences,themorelikelypeoplewouldrespond.Thishypothesiswaspartiallysupported.Theorderofseriousnessofthehazardsandconsequenceswas:toxins,health,salt,tasteandbusiness.Toxinswereexpectedtobethemostimportanttopic,butitwasneverrelatedtosales.ToxinsinthewaterhasbeenalongstandingsalientissueinNewOrleans.Itispossiblethatpeoplehavebecomedesensitizedtotoxinsasahazard;therefore,theyfailedtorespondtothisthreatwhiletheyrespondedtolessseriousthreats.Thehealth,salt,tasteandbusinessconsequenceswerenewthreatsandtheirnewnessmayhaveprovokedtheresponse.Further,theresultsshowthathealthwasoneofthetwomostimportanttopicsforbothmediaandbusinessconsequenceswereamongtheleastimportant.Thus,theresultssuggestthatpeoplerespondmorereadilytomoreseriousthreatsifthesethreatsarenew.Also,itisworthnotingthatpurchasingbottledwaterhasbecomea commonresponsetothelongstandingproblemoftoxins.Theresultssuggestthatpeopleusedthesameresponseinthefaceofthenewhazard.Second,peoplewerehypothesizedtorespondmorereadilytoreportsthatcitedpersonallyrelevantconsequencesofthehazardthanlessrelevantconsequences.Theresultsprovidesupportforthispartofthemodel.Althoughbusinessconsequencesandcollectiveresponseswererelatedtoretailsales.thetopicmoststronglyandconsistentlyrelatedwashealthconsequences.Thiswasoneofthetwomostimportanttopicsforbothmedia.Forthenewspaper,thesecondtopicwasalsoapersonalconsequence,taste.Forthetelevision,thesecondtopicwasacollectiveresponse,bargingwater.Althoughbargingwaterislesspersonal15

PAGE 19

thanhealth.taste,toxins,andsalt,it is morepersonallyrelevantthanbusinessconsequencesandtheothercollectiveresponses.Further,themodelsuggestedthatintheabsenceofpriorexperiencewiththehazard,peoplewillnotdiscriminateasmuchbetweenpersonallyrelevantconsequencesandlessrelevantconsequences.SincetheNewOrleansareahadnothadasaltwaterintrusionofthismagnitudesince1937,priorexperiencewiththishazardwaslimited.Thus,wewouldexpectpeopletobelessdiscriminatingconcerningthepersonalrelevanceoftheconsequence.Theresultsthatbusinessconsequencesandmentionsofcollectiveresponsesaffectedsalesofbottledwatersupportstheideathat.whenpriorexperienceislimited,peoplearelessdiscriminatingaboutthepersonalrelevanceoftheconsequence.especiallysincethesetopicstendedtoaffectsalesonlybeforethewedgearrived.Thlrd.themodelpredictedthatresponsestomediareportlngwouldbeafunctionoftherelevanceofresponsesmentionedinthereports.Ifsocialactorsexamineothers'responsesinordertohelpdefineambiguoussituations,thenmentionsofpeoplebuyingbottledwaterwouldeffectretailsales.Theresultssupportthisprediction;inbothmedia,mentionsofpeoplebuyingbottledwaterwasasignificantpredictorofretailsales.However,mentionsofbargingwater(categorizedasacollectiveresponse)wasalsoasignificantpredictor.Ifbuyingbottledwaterisamorepersonallyrelevantresponsethanthebargingofwater,theresultsonlyprovidepartialsupportforthisaspectofthemodel.Ontheotherhand,sincethewaterwasbargedtohelpfreshencommunity,aswellasbusiness,watersupplies,peoplemayhaveinterpretedthesetwotypesofresponsesinmuchthesameway:asinformationrelevanttotheissueofpersonalwatersupplies.Whileourdatadonotallowustodirectlytestthishypothesis,theresultsdoshowthatbargingwaterhadbyfarthegreatesteffectamongallthelesspersonalresponses.Fourth,peoplewerehypothesizedtomorereadilyrespondtoreportswhichstressedthelikelihoodoftheeventoritsconsequences,especiallyiftheyhadpriorexperiencewiththehazard.Ourresultsconcerningthetoneofthestoriesproducemixedsupportforthemodel.Forthetelevislonreports.mentionsofpersonallyrelevantconseauencesandresponsesthatstressedthe certainty ofthesaltwedgeanditsconseauenceshadaslgnlflcanteffectonsales.Ontheotherhand,forthenewspaperstories,mentionsofsuchtopicsthatdiminishedthehazardwasasignificantpredictorofsales.Finally,themodelhypothesizedthattheseaspectsofselectivityinattendingandrespondingtomediareportswouldvaryintheireffectsdependentuponwhetherpeoplehaddirect,personalexperiencewiththehazardevent.Theresultsprovide16

PAGE 20

mixed.thoughstrong.supportforthemodel.Forsometopics.thereweresiqnificantlnteractlonsbetweenthementlonsandproxlmltylndlcatinqthatmentionsofthesetOP1CSincreasedsalesonlybeforepeoplehaddlrectexperlencewlththehazard.Theseinteractlonsoccurredfortaste(newspaper).health(televlsion).barglngwater(bothmedla) and otherresponses(newspaper).Further.proxlmitywaslmportantinthreeothermodels:taste.buyingwater.andbuslnessconsequences(alltelevlslon).Theresultsthatmostnon-personalconsequencesandresponsessignificantlYincreasesalesonlybeforethewedgearrivedsupportsafinalaspectofthemodel:inthefaceofuncertainty.peoplewillnotdiscriminatebetweenpersonalandnon-personalconsequencesandresponses;sincebothnecessarilymentionthehazardand/oritseffects.eitherwouldserveasanimpetusforbuyingwater.However,afterthewedgearrived,frompersonalexperiencemostpeoplewouldhavebeenabletoassessthelackofapersonaleffect.andwouldtherebyhavebeenmoreselectiveintheirattentionandresponses.Thisisespeciallylikelyinthecaseofbargingwater,sincewhilewecategorizedthisasanon-personal.andthereforelessrelevant.response.peoplemayhaveattendedandreactedtothisaspectofmediareportsmorereadilybeforethewedgearrived.Infact.thatbargingwateristhesinglemostimportantpredictorofsalesamongthenonpersonalvariablesprovidesadditional.thoughindirect.supportforthemodel.However,mentionsofbusinessconsequencesinthetelevisionreportscontinuedtoaffectsalesafterthewedgearrived.Asmentionedabove,thismaybeduetothesenonpersonalconsequencesoftenoccurringinthesamestoriesaspersonalconsequences.However.wewerenotabletotestthishypothesissincewecannotseparatetheirindividualeffectsonsales.CONCLUSIONWehaveintendedthispaperasabothatheoreticalandmethodologicalcontributiontotheemPlrlcalstudyoftheeffectsofmedlareportsofnaturalhazardsonhumanresponses.Wetookasabasicpostulatethatpeoplerespondtoahazardprimarllyonthebasisofinformationgleanedfromthemedia.Further.wealsopostulatethatmediareportingundergoeschangesdurlngthecourseofthehazardevent.Onthebasisofthesetwoassumptions.themostadequatewaytoassesstheeffectsofmediareportingonresponsebehaviorswouldbearesearchdesignthatfollowedthisrelationshipoverthecourseofawholehazardevent.Weincorporatedthisldeaintoourresearchbycollectingdataonthemediareportsaswellasresponsebehaviorsoverthecourseofthesaltwaterintrusion.Further,weusedatimeseriesanalYsistoexaminetheeffectsoftheformerovertime.Inaddition.weusedanintenslveanalysisofthemediareportstoqenerateourindependentvariables.A commontradltion17

PAGE 21

inthestudyofmediaeffectsisacontentanalysisinwhichasmallnumberofattributesofstoriesareidentifiedandthensummed.Ouruseofadetailedqualitativeanalysisallowedustogenerateamuchricherandelaboratepictureofthecontoursofthemediastories.Inturn,thisallowedustoexamineafargreaternumberofattributes'ofthesestoriesthanwouldhaveotherwisebeenthecase.Despitethemethodologicaladvantagesofourresearch,severalimprovementsaresuggested.Whileourconceptualmodelstressestheimportanceofambiguityanduncertaintyaswellasselectivity,ourdatadidnotallowustodirectlytestthisaspectofourmodel.Futureresearchwhichutllizedsurveyorquestionnairedataduringthecourseofthehazardeventwouldbeabletodevelopmeasuresofanddirectlyassesstheroleofthesesocialpsychologicalprocessesinresponses.Obviouslythisisnosmalltask.asitrequiressomedegreeofplanningandacertainrapidityinordertocollectthequestionnairedataduringthecourseofthehazardevent.ItwouldbebeneflclalItself-reportsofbottledwaterpurchases.aswelIasotherresponses.werelncludedlnthedata.Ouruseofbottledwatersalesasanaggregatemeasureofresponseswasappropriate.butitseemsreasonabletoarguethatpeopleintheNewOrleansarearespondedinotherways.evenwhentheyinterpretedthemediareportsinthesamewayasthepeoplewhoboughtbottledwater.Thereseemsarelativelackofattentionintheliteratureonnaturalhazardstodevelopinggeneralmodelsofresponsebehaviors,althoughNigg's(1982)researchprovidesasignificantexceptiontothistrend.Notonlyshouldattentionbepaidtodevelopingmodelsthataregeneralizableenoughtoapplytovarioustypesofhazardsundervariousconditions,butthesemodelsshouldmayrelevanttothisarea.Forexample,somerecentconceptualschemesarebeginningtomakeuseofworkincognitivesocialpsychology.Further.researchandtheoryintheareasofurbanecology,politicalsystems,communicationandsociolinguisticscancontributetoresearchandconceptualdevelopmentintheareaofnaturalhazardsandhumanresponses.Responsestohazardscanbeexaminedbothatasocialpsychologicalandsociallevel,furtherconceptualdevelopmentinthisfieldwouldseemtodependongreateruseofextantknowledgelnsociologicalresearchandtheoryatbothoftheselevelsofanalYsis.18

PAGE 22

Table 1. Regression AnalysesaoftheEffectoftheTypeofMedia MentionsonSalesofBottledWater. DependentVariable:Natural LogarithmoftheGallonsSold Daily .060 Bbeta.840* .838* .052*.151*1.479.788* .781* 1.870Variable:Buying Water IndependentLagsalesBuying Water ConstantR 2Adjusted R2Durbin-WatsonFirst-orderautocorrelation.095A.Newspaper Independent Vari ab 1thBbeta.800* .798* .130*.211* 1.829 .806* .798* 1.799LagsalesHealthConstantR 2Adjusted R2Durbin-WatsonFirst-orderautocorrelation LagsalesProximityTasteProximity *TasteConstant2RAdjusted R2Durbin-WatsonFirst-orderautocorrelationBbeta.817*.814*-.080-.064.367*.505*-.308*-.404*1.742 .810* .795* 1.879 .051B.TelevisionIndependentVariable:TasteVariable:Buying WaterLagsalesLagtasteProximityConstantR 2Adjusted R2Durbin-WatsonFirst-orderautocorrelationIndependent Bbeta.805*.803* .071*.160*-.139b-.lllb1.883 .796* .785* 1.731 .123LagsalesLagbuyingwaterProximityConstantR 2Adjusted R2Durbin-WatsonFirst-orderautocorrelationBbeta.808* .806* .169* .187*-.131e-.105e1.864 .806* .795*1.778.098 Note:Laqsalesisthelaqofthedependentvariable.proximityisthedummyvar1ablecoded 0beforethesaltwaterwedgearrivedand 1afterthewedqearrived.laqtasteisthepreviousdaY's mentionsoftasteand laqbuyingwateristhepreviousday'smentionsofpeople buyinqbottledwater.*p<.05.a.TheanalYses werecorrectedforfirst-orderautocorrelation(AR1). b. 0=.0805.TheFtestfortheincrement duetoproximitywas3.98whileanFof4.0815neededforsiqnificanceatthe.05level.c.p=.0896.TheFtestforthelncrement duetoproximitywas2.98 whileanFof4.08isneededforsiqnificanceatthe.05level.

PAGE 23

Table 1continued.Regression Analyses aoftheEffectoftheTypeofMedia MentionsonSalesofBottledWatbeta.904*.178d-.240*.078 .928 .797* .786* 1.840 Y:ari B.903* .030d-.041*LagsalesLaqsaltLaq2saltConstantR 2Adjusted R2Durbin-WatsonFirst-orderautocorrelation.030 DependentVariable:Natural LogarithmoftheGallons Sold DailyB.Television Va [i abJe:_ Hectlt:h Bbeta.856*.854*-.084-.067 .637* .560*-.629*-.541*1.397 .795* .779* 1.922LagsalesProximity LaqhealthProximity*la9healthConstantR 2Adjusted R2Durbin-WatsonFirst-orderautocorrelationNote:Lagsalesisthelagofthedependentvariable,proximityisthedummyvaria coded 0beforethesaltwaterwedqearrivedand 1afterthewedgearrived.laghealthispreviousday'smentionsofhealthconsequences,lagsaltisthepreviousday'smentionssalt,andlag2saltisthementionsofsalt2 dayspreviously.*p<.05a.Theanalyseswerecorrectedforfirst-orderautocorrelation(AR1). d. p=.0600.

PAGE 24

Table 2. Regression AnalysesaoftheEffectoftheToneofMedia MentionsonSalesofBottledWater. DependentVariable:Natural LogarithmoftheGalIons Sold Daily .114beta.805* .157*A. IndependentVariable:Lagsales Diminish ConstantR 2Adjusted R2Durbin-WatsonFirst-orderautocorrelationDiminishB.807* .041*1.761.787* .779* 1.764 .091 LagsalesLagcertainProximity Proximity*ConstantR 2Ad';ustedR2Durbin-WatsonFirst-orderautocorrelationB.Television i ab __il1 Bbeta.801* .799* .101*.592*-.107-.085Lagcertain-.082*-.472*1.899 .807* .792* 1.797 Note: Laqsalesisthe lag ofthedependentvariable,proximityisthedummyvariablecoded 0beforethesaltwaterwedgearrivedand 1afterthewedgearrived,diminishisthenumberofmentionsthatattempttodiminishthehazardorconsequence(health,taste.toxins.andsalt).andlagcertainisthenumberofmentionsthepreVlOUSdaythatclaimthatthehazardorconsequencewilloccur.uncertainmentionsinthenewspaperortelevisionnewsdidnotaffectsalesofbottledwater,certainmentionsinthenewspaperdidnotaffectsalesandmentionsthatdiminishthehazardorconsequence inthetelevisionnewsdidnotaffectsales;therefore.nomodelsareshownfortheseseries.*p<.05.a.Theanalyseswerecorrectedforfirst-orderautocorrelation(AR1).

PAGE 25

Table 3. Regression Analyses aoftheEffectofthePlaceintheMedia MentionsonSalesofBott Water. DependentVariable:Natural LogarithmoftheGallonsSold Daily .066 P.lMuemines Bbeta.852*.849* .093*.428*-.042b-.034bPlaq.-.091-.4111.396 .793* .777* 1.855LagsalesPlaq .Proximity Proximity* cpnstant RAd.;ustedR2Durbin-WatsonFirst-orderautocorrelation.060A.Newspaper Independell't1asi ab1 Bbeta.834* .832* .086*.428*-.067-.054-.075*-.352*1.570.807* .792* 1.864LagsalesN.O.Proximity Proximity*N.O. cpnstant RAdjusted R2Durbin-WatsonFirst-orderautocorre'lationLagsalesOtherPlaces Cpnstant RAdjusted R2Durbin-WatsonFirst-orderautocorrelation9ther Bbeta.785* .783* .028*.198* 1.960 .797* .790* 1.768 .111OrleansBbeta.790* .788* .028c.152c-.161c-.129c2.042 .791* .779* 1.638 .138 1.539 .782* .774* 1.713LagsalesLagPlaquemines IndependentVariable:Plaquemines Bbeta .833/.831/ .045 .129 cpnstant RAdjusted R2Durbin-WatsonFirst-orderautocorrelation.168B.TelevisionIndependentVariable:NewLagsales LagNewOrleans Proximity Cpnstant RAdjusted R2Durbin-WatsonFirst-orderautocorrelationNote:Laqsalesisthe laq ofthedependentvariable.proximityisadummyvaria coded 0beforeand1afterthe wedge arrived.otherplacesismentions conCernlnq persa hazards and consequences (hea 1th.taste.toxins.andsaIt)inSoutheasternandNorthLouislanaandoutsideofthestate.lagNewOrleansisthepreviousdav'smentlons concernNewOrleans.and laqPlaqueminesisthepreviousdaY's mentionsconcerningPlaquemines Pari*p<.05.a.Theanalyseswerecorrectedforfirst-orderautocorrelation (AR1).b. p=.057.TheFtestfortheincrement duetoproximityandtheinteraction3.01whileanFof3.23isneededforsignificanceatthe.05level.c.p<.057.TheFtestfortheincrement duetolagNewOrleans andproximitywas 3 whileanFof3.23isneededforsignificanceatthe.05level.d. p=.063.TheFtestfortheincrement duetolagPlaquemineswas3.69while i of4.08isneededforsignificanceatthe.05level.

PAGE 26

Table 4. Regression AnalysesaoftheEffectofMediaMentionsofBusiness Consequences and ResponsestotheSaltWaterIntrusiononSalesofBottledWater. DependentVariable:Natural LogarithmoftheGallons Sold Daily .076Variable:Other Responses Bbeta.925*.922* .019 .207 .009b.007bResp.-.028-.273.685 .789* .773* 1.827 Lagsales Resp. Proximity Proximity *cpnstantRAdjusted R2Durbin-WatsonFirst-orderautocorrelation.040 Bbeta.872* .870* .146* .372*-.032-.026-.160*-.387*1.203 .797* .782* 1.907 Barging WaterA.Newspaper IndependentVariable:Lagsales Barge Proximity Proximity cpnstantRAdjusted R2Durbin-WatsonFirst-orderautocorrelation.113 IndependentVariable:BusinessEffectsBbeta.871* .871* .141c-.162-.127-.035*-.181*1.338 .798* .782* 1.736 Lagsales Laqbusiness Proximity Laq2business CpnstantR Adjusted Rl Durbin-WatsonFirst-orderautocorrelation.048 Barging Water Bbeta.838*.836* .646* 1.478*-.099-.079-.609*-1.391*1.562 .802* .787* 1.884B.TelevisionIndependentVariable:Lagsales Laqbarqe Proximity Prox. laqbarqecpnstantRAd.iusted R2Durbin-WatsonFirst-orderautocorrelation.104 Lagsales Lagresponse Lag2responsecpnstantRAdjusted R2Durbin-WatsonFirst-orderautocorrelation .indepe,ndent Bbeta.906*.906* .017.127-.027*-.198*.901 .792* .779* 1.789 Note: Lagsalesisthelagofthedependentvariable,proximityisadummyvariablecoded 0beforeand 1afterthewedgearrived,bargeismentionsofbargingwater.resp.ismentionsofotherresponsestothewedge, lagbargeisthepreviousday'smentionsofbargingwater.lagbusinessisthepreviousday'smentionsofbusinessconsequences, lag2business1Smentionsofbusinessconsequences 2 dayspreviously,lagresponseisthepreviousday'smentionsofresponsestothewedqe.andlaq2responseismentionsotresponses 2 daysbefore.* p<.05.a.Theanalyseswerecorrectedforfirst-orderautocorrelation(AR1). b. p=.0784.c.p=.0952. d. p=.0585.TheFtestfortheincrement duetoproximitywas3.84whileanFof4.08isneededforsignificanceatthe.05level.Includingproximity inthemodelcauseslag2businesstobecomestatisticallysignificant.

PAGE 27

Table 5. Regression AnalysesaoftheCombinedEffectofMediaMentions ConcerningBusiConsequences, Barging Water andOtherResponsesonSalesofBottledWater. Consequences, and Other Responses.040Barging Water, Business Bbeta.831*.831*-.133-.105.594*1.365*-.569*-1.306*.008.038-.013-.067.019.139-.017-.127 1. 670.825* .795* 1. 901DependentVariable:Natural LogarithmoftheGallons Sold DailyA.TelevisionIndependentVariables:Laqsales Proximity Laqbarqe Prox.*laqbarge Laqbusiness Laq2business Laqresponse Laq2response CpnstantR' Adjusted R2Durbin-WatsonFirst-orderautocorrelationNote: Lagsalesisthelagofthedependentvariable,proximityisadummyvaria coded 0beforeand 1afterthewedqearrived,lagbargeisthepreviousday'smentions bargingwater.lagbusinessisthepreviousdaY's mentionsofbusinessconseQuenclag2businessismentionsofbusinessconsequences 2 dayspreviously,lagresponseispreviousdaY's mentionsofresponsestothewedge, andlag2responseismentionsofrespor 2 dayspreviouslY.*pC05.a.Theanalyseswerecorrectedforfirst-orderautocorrelation(AR1).

PAGE 28

REFERENCESAdams,WilliamC.1986."WhoseLivesCount?:TVCoverageof Natural Disasters." Journal ofCommunication36:113-122. Alexander,David.1980."TheFlorence Floods--What the Papers Said." EnvironmentalManagement4(1):27-34. Carter,1.Michael.1980."CommunityWarningSystems:TheRelationshipsAmongthe BroadcastMedia,EmergencyService Agencies,andthe NationalWeatherService."Pp.214-228inDisastersandtheMassMedia,editedbytheCommitteeonDisastersandtheMassMedia,EverettM.Rogers,Chairman.Washington, D.C.: NationalAcademyofSciences. Cleary,PaulD.1987."WhyPeopleTakePrecautions Against Health Risks."Pp.inTakingCare: UnderstandingandEncouraging Self-Protective Behavior, editedbyNeilD.Weinstein.N.Y.:CambridgeUniversity Press. Glaser,BarneyG.andAnselmL.Strauss.1967.TheDiscovery ofGroundedTheory:Strategiesfor Qualitative Research. Chicago: Aldine PublishingCompany.Helwig,JaneT.1978.SASIntroductoryGuide.Cary,NorthCarolina:SASInstitute,Inc. Kepplinger,HansMathiasandHerbertRoth.1979."Creating aCrisis:OilSupplyin1973-74." PublicOpinionQuarterly 43(3):285-296. Kreps,Gary.1980."ResearchNeedsandPolicy IssuesonMassMediaDisaster Reporting."Pp.35-74inDisastersandtheMassMedia,editedbytheCommitteeonDisastersandtheMassMedia, EverettM.Rogers,Chairman.Washington, D.C.: NationalAcademyof Sciences.Lang,GladysEngelandKurtLang.1980."NewspaperandTVArchives:SomeThoughtsAboutResearchonDisasterNews."Pp.269-280inDisastersandtheMassMedia,editedbytheCommitteeonDisastersandtheMassMedia,EverettM.Rogers,Chairman.Washington,D.C.: NationalAcademyof Sciences. Larson,JamesF.1980."AReviewof the State of the ArtinMassMediaDisaster Reporting."Pp.75-126inDisastersandtheMassMedia,editedbytheCommitteeonDisastersandtheMassMedia,EverettM.Rogers,Chairman.Washington, D.C.: NationalAcademyof Sciences. Leik, RobertK.,1.MichaelCarter,andJohnP.Clark.1981."CommunityResponseto NaturalHazardWarnings:SummaryFinal Report." Minneapolis: University of Minnesota,Departmentof Sociology. Mileti,DennisS.andJohnH.Sorenson.1987."NaturalHazardsandPrecautionary Behavior."Pp.inTakingCare: UnderstandingandEncouraging Self-Protective Behavior, editedbyNeilD.Weinstein,N.Y.:CambridgeUniversity Press.

PAGE 29

Needham,D.andJ.G. Nelson.1977.IINewspaperResponsetoFloodandErosionHazardsontheNorthLakeErie Shore.IIEnvironmenta1Management1(6):521-540.Nigg,JoanneM.1982.IICommunicationUnderConditions of Uncertainty: Understanding Earthquake Forecasting.IIJournal ofCommunication(Winter):27-36. Quarantelli,E.L.1987.IISocialScience studyCommunication:Preliminary Paper.IINewark,Delaware, Disaster Research Center. of DisastersandDE:University ofMassRobinson,JohnP.andMarkR.Levy.1986.IIInterpersonalCommunicationandNewsComprehension.1IPublicOpinionQuarterly 50(1):160-175. Rogers, EverettM.1987.liThediffusion of Innovations Perspective.1IPp.inTakingCare: UnderstandingandEncouragingSelf-ProtectiveBehavior, editedbyNeilD.Weinstein,N.Y.:CambridgeUniversity Press. Rogers, EverettM.andRahulS.Sood.1980.IIMassMediaCommunicationandDisasters:A Content Analysi s ofMediaCoverageoftheAndhraPradeshCycloneandtheSahelDrought.1IPp.139-157in DisastersandtheMassMedia,editedbytheCommitteeonDisastersandtheMassMedia,EverettM.Rogers,Chairman.Washington, D.C.: NationalAcademyof Sciences.Ruch,CarltonE.andLarryB.Christensen.1981.IIHurricaneMessageEnhancement.IICollege Station,TX:TexasA&M University,SeaGrant CollegeProgram.Sandman,PeterM.,NeilD.Weinstein,andM.L.Klotz.1987.IIpublicResponseto theRiskfromGeologicalRadon.IIJournalofCommunication37(Summer):93-108. Slovic, Paul,BaruchFischhoff,andSarah Lichtenstein.1987.IIBehavioral DecisionTheoryPerspectivesonProtective Behavior.1IPp.inTakingCare: UnderstandingandEncouragingSelf-ProtectiveBehavior, editedbyNeilD.Weinstein,N.Y.:CambridgeUniversity Press. Sorenson,JohnH.andDennisS.Mileti.1987.IIProgramsthatEncouragetheAdoptionof Precautions against Natural Hazards:ReviewandEvaluation.1IPp.208-229inTakingCare: UnderstandingandEncouraging Self-Protective Behavior, editedbyNeilD.Weinstein,N.Y.:CambridgeUniversity Press. Stauffer, John, Richard Frost,andWilliam Rybolt.1983.liTheAttention FactorinRecallingNetworkTelevisionNews.1IJournal ofCommunication(Winter):29-37.

PAGE 30

Turner,RalphH.1980."TheMassMediaandPreparation for Natural Disaster."Pp.281-292in DisastersandtheMassMedia,editedbytheCommitteeonDisastersandtheMassMedia,EverettM.Rogers,Chairman.Washington, D.C.: NationalAcademyof Sciences. Weinstein,NeilD.1987."Cross-Hazard Consistencies: Conclusions aboutSelfProtective Behavior."Pp.inTakingCare: UnderstandingandEncouraging Self-Protective Behavior, editedbyNeilD.Weinstein,N.Y.:CambridgeUniversity Press.