Contents lists available atScienceDirect
Preventive
Veterinary
Medicine
j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / p r e v e t m e d
Application
of
discrete
choice
experiment
to
assess
farmers’
willingness
to
report
swine
diseases
in
the
Red
River
Delta
region,
Vietnam
Hoa
T.T.
Pham
a,∗,
Marisa
Peyre
b,
Tuyen
Quang
Trinh
c,
Oanh
Cong
Nguyen
d,
Ton
Dinh
Vu
d,
Theera
Rukkwamsuk
e,
Nicolas
Antoine-Moussiaux
faFrenchAgriculturalResearchCenterforInternationalDevelopment(CIRAD),VanPhucDiplomaticCompound,298KimMa,BaDinh,Hanoi,Vietnam
bFrenchAgriculturalResearchCenterforInternationalDevelopment(CIRAD),AnimalandIntegratedRiskManagementResearchUnit(AGIRs),34398
Montpellier,France
cNationalInstituteofAnimalScience,Hanoi,Vietnam
dCenterforInterdisciplinaryResearchonRuralDevelopment(CIRRD),VietnamNationalUniversityofAgriculture,Hanoi,Vietnam
eKasetsartUniversity,Bangkok,Thailand
fFundamentalandAppliedResearchforAnimals&Health(FARAH),UniversityofLiège,Belgium
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received27September2016 Accepted2January2017 Keywords: Swinediseases Willingnesstoreport Surveillance ControlDiscretechoiceexperiment
Vietnam
a
b
s
t
r
a
c
t
Adiscretechoiceexperiment(DCE)iscarriedouttovaluesocio-economicfactorsinfluencingthefarmer’s decisiontoreportswinediseasesandtoassessthewillingnessoffarmerstoreportswinediseases. DatawerecollectedbetweenMarchandJuly2015intwoprovincesintheRedRiverDelta,Northern Vietnam,from196pigproducersbyface-tofaceinterview.Aconditionallogitmodelisusedtomeasure therelativeimportanceofthesocio-economicfactorsandcalculatetheexpectedprobabilityofdisease reportingunderchangesoflevelsofthesefactors.Resultsofthestudyindicatedthatthelikelihood ofcompensationandthetypeofcullingimplemented(alloronlyunrecoveredpigs)arethetwomost importantfactorsinfluencingfarmerreporting.Compensationlevel,movementrestrictionanddelay incompensationpaymentalsohavesignificantimpactsonfarmer’sdecisiontoreportanimaldisease buttheyarenotasimportantastheabovefactors.Threedifferentscenariosincludingchangesinsix differentfactors(attributes)aretestedtopredictprobabilityofanimaldiseasereporting.Underthe currentsituation(uncertaintyofbeingcompensated),only4%ofthefarmerswouldreportswinedisease outbreaktotheofficialsurveillancesystemifthecullingpolicyinvolvesallpigsinaffectedfarms.This numberisincreasedto26%ifcullinginaffectedfarmsisrestrictedtounrecoveredpigsonly.Ensuring certaintyofcompensationincreasesreportingprobabilitybyupto50%and90%ifalloronlyunrecovered pigsaredestroyed,respectively.Theresultsofthisstudyareimportantforimprovingtheperformance andsustainabilityofswinediseasesurveillancesysteminVietnam.
©2017ElsevierB.V.Allrightsreserved.
1. Introduction
Passivesurveillanceisaprocesswherebyanimalhealth infor-mation is collected regularly from different sources such as diseasereportingby livestockowners, fieldveterinarians, labo-ratorydiagnosticreports, slaughterhouses and livestockmarket records(Geeringetal.,1999).Farmers’reportingofanimaldisease playsanimportantroleinthenationalsurveillancesystemdue totheirabilitytodetectawiderangeofdiseases,theirextensive
∗ Correspondingauthor.
E-mailaddress:thithanhhoapham@ymail.com(H.T.T.Pham).
coverageoftheanimalpopulation,andtheirfrequentcontactwith animals(Cameron,2012).Theresultsofastudycomparingthree datasources ofdairy diseasesurveillancein NewZealand indi-catedthatfarmersrecordedthehighestnumberofdiseaseevents (14.6per1000cowmonthsatrisk)comparedtoveterinary prac-tices(5.2per1000cowmonthsatrisk)anddiagnosticlaboratories (0.6per1000cowmonths)(McIntyreetal.,2003).However,the willingnessoflivestockownerstoshareanimalhealthinformation withveterinaryauthoritiesisinfluencedbymanysocio-economic factors.Cameron(2012)proposedthatfearofthenegative conse-quencesofdiseasenotification,poorrelationshipwithveterinary services,andfarmers’apathytowarddiseaseoccurrenceand con-trolarethemainreasonsfor farmersfailingtoreport(Cameron, http://dx.doi.org/10.1016/j.prevetmed.2017.01.002
2012).Theresultsofasocio-psychologicalstudyamongpigfarmers in theNetherlands indicated that uncertainty aboutthe occur-renceofa given diseaseand thethreatof afalse alarm, which mighthaveanegativeimpactonthefarmandontherelationship betweenfarmersandveterinarians,wereimportantfactors influ-encingfarmerreporting(Elbersetal.,2010).Lackoftrustinthe control/eradicationmeasuresofveterinaryauthoritiesisalso con-sideredasabarriertofarmers’willingnesstoreport(Elbersetal., 2010;Palmeretal.,2009).Hadornetal.(2008)indicatedthatthe diseaseawarenessoffarmersisanimportantfactorinfluencingthe sensitivityofpassivesurveillanceandthatinendemicareas,where animalownersarelikelytobeawareofthedisease,the probabil-ityofdiseasereportinganddetectionseemstobehigherthanin areaswherethediseaseisunknownorneglected(Hadornetal., 2008).Understandingthese socio-economicfactorsiscritical for improvingtheeffectivenessofpassivesurveillancesystems.
Variousinterventionshavebeensuggestedtoimprovethe per-formanceofpassivesurveillancesystemsbasedontheinvestigated farmer’sattitudeandbehaviorinanimaldiseasereporting(Bronner etal.,2014;Dungetal.,2006;Elbersetal.,2010;HadornandStärk, 2008;Palmeretal.,2009).Themostusedmethodsareenhancing publicawareness ofpriorityanimal diseases andof the impor-tanceof earlyreporting throughmassmedia,theimprovement ofrelationshipsbetweenfarmers andveterinarystaff,provision ofincentivesforreporting,andprovisionofappropriate compen-sationforcullinganimals(Cameron,2012).Gilbertetal.(2012) suggestedasubsidyforcarcasscollectionandtelephone consul-tationsforprivateveterinariansastwosolutionstoincreasethe cost-effectivenessofpassivesurveillance(Gilbertetal.,2012). Pos-siblesolutionssuggested toimproveanimal healthsurveillance inVietnamincludetravelallowancesfordistrictveterinariansto facilitatefieldvisitingandcommunicationwithfarmers,commune para-veterinariansanddrugvendors,andimprovingdata analy-siscapacityatlocallevel(i.e.districtandprovincialoffices)(Dung etal.,2006).Inthecontextofendemicdiseases,tomaintainthe interestandmotivationoffieldactorssuchasfarmersandlocal veterinarianstoreportanimaldisease,animalhealthinformation needstobecommunicatedtothepartnerstoshowlivestock own-ersandlocalveterinariansthedirectbenefitsoftheirparticipation inthesurveillancenetwork(Doherretal.,2012).Anexampleof reinforcingpassivesurveillanceinSouthAustraliathrough subsi-diestolaboratorytesting,incentivesforprivateveterinariansin datacollection,andtheraisingoffarmers’awarenessofthebenefits ofsurveillanceactivitiesresultedinsignificantlyincreasingsample submissionrate(Cliftetal.,2006).Mostofthesestudieswerebased ona qualitativeapproachtoidentifythefactorsinfluencingthe decision-makingprocess.Toourknowledge,onlyalimitednumber ofstudies(Limonetal.,2012;Palmer,2009)haveusedquantitative approachestoassessthesocio-economicfactorsinfluencingfarmer reporting.
Statedpreferencemethodsareasetofdifferenteconometric methodstocharacterizeconsumerpreferencesincaseswhere mar-ketscannotbeobserved.Theyhavebeenwidelyusedinmarketing (Lagerkvist etal., 2006; Thongand Hung, 2014;Xu and Wang, 2014),transport,andenvironmentaleconomics(Adamowiczetal., 1994;Boxalletal.,1996;Hanleyetal.,1998;ZanderandGarnett, 2011),aswellashealth economics(Howardand Salkeld,2009; Raoetal.,2013;Sadiqueetal.,2013).Morerecently,theyhave alsobeen appliedtoanimal scienceand animal health. Among thedifferenttypesofstatedpreferencemethods,discretechoice experiments(DCE)aremoreparticularlyappliedtoinvestigating consumerpreferencesforparticularqualities(calledattributes)of theproductorservice.Assuch,theyprovidequantitative informa-tiononstrengthofpreference,thetrade-offbetweenthefactors affectingthispreference,andtheprobabilityoftakeupofa prod-uct,serviceorinnovation(WHO,2012).Regardinganimalproducts,
theyhavebeenmobilizedtoquantifythedemandforunobservable attributes,suchasanimalwelfareattributes(Lagerkvistetal.,2006; Liljenstolpe,2005).DCEisalsousedtoidentifyandevaluate farm-ers’decisionsregardinganimal diseasereporting(Oparindeand Birol,2011)andanimaldiseasecontrolmeasures(Otiennoetal., 2011;Zhangetal.,2014).
Theobjectivesofthisworkaretomeasuretherelative impor-tanceofsocio-economicfactorsthatinfluencethefarmer’sdecision toreportdisease,andtoassessthefarmer’swillingnesstoreport animaldiseases.Theresultsofthisstudyhighlightthefactorsthat needtobeimprovedandtowhatextent,inordertoimprovethe probabilityofanimaldiseasereporting.
2. Materialandmethods
In this study, theDCE is used to explore farmer’s choicein regardtoanimaldiseasereportinguponoutbreakonhis/herfarm. This approach waschosen as it allows quantitatively assessing therelativeimportanceoffactorsinfluencingfarmer’sbehavioror decisionmakingandevaluatingthefamer’swillingnesstoreport intermsofpredicteduptakeunderthechangesofthesefactors (attributes).Farmer’spreferenceforanimaldiseasereportingwas elicitedbypresentingthemanumberofhypotheticalchoicesthat wereconstructedbyvaryingcombinationsofattributelevels(i.e. characteristicsofdiseasenotificationandcontrolprocedures). Par-ticipantswererequiredtoselectoneofthealternativesineach choice-setaccordingtotheirpreference.TheDCEmodelisbased onrandomutilitytheoryundertheassumptionthatthefarmer chooses the alternative that provides the highest benefit (util-ity) (McFadden, 1974).Statistical analysis of theresponse data allows toestimatehowmuch each alternativeis preferredand whichattributescontributemosttothechoice.Monetaryunitcan beincludedasoneattributeinDCEandthecombinationofcost dataprovidesanassessmentofthecost-effectivenessof alterna-tiveoptions(Kjaer,2005).Percentageoffarmer’s willingnessto reportanimaldiseasesispredictedbymodelingtheprobabilityof choosinganalternativewithspecifiedcharacteristics(attributes) ofdiseasenotificationandcontrol.Suchinformationisacritical pointofthisstudy.
ToconductDCE,anumberofdistinctstagessuchas identifi-cationofattributesandassignmentofattributelevels,designand constructionofchoicesetstopresenttoparticipants,development questionnaire,datacollection,anddataanalysiswereimplemented step-by-step.ThelinksbetweenstepsarecrucialasresultsofDCE arehighlysensitivetothesystematicandwelldesignofthewhole experiment(WHO,2012).
2.1. Identificationofattributesandassignmentoflevels
Atotalof18focusgroupdiscussionswithpigfarmersand key-informantinterviewswerecarriedoutintwoprovinces.Thetopics tackledwereswinediseases,diseasepriorityforsurveillanceand control,aswellasfactorsinfluencingthefarmer’sdecisiontoreport swinedisease.Fromtheresultsofgroupdiscussionsandkey infor-mantinterviews,sixattributeswereidentifiedandincludedinthe discretechoiceexperimentasfollow:(1)Uncertaintyofbeing com-pensatedincase ofanimal diseasereporting;(2) Compensation levelsincaseofanimalculling(priceperkgofpigbodyweight); (3)pigcullingincaseofdiseasereporting;(4)administrative pro-cedurefordiseasenotificationandgettingcompensationpayment; (5)movementcontrolincaseofdiseasenotification;(6)delivery timeofcompensationpayment.
Attribute levelswerethen assigned onthebasis of the sur-veyresultsandfromaliteraturereviewonveterinaryregulations (MARD,2008,2007)inVietnam.Theattributesandattribute
lev-Table1
Attributesandattributelevelsusedinchoiceexperiment.
Attribute Levels Probabilityofbeing compensated Uncertain Certain Compensationlevel VND24000/kgLBW VND33600/kgLBW VND38400/kgLBW Animalcullingpolicy Unrecoveredpigs
Allpigsatfarm Administrativeproceduresof
diseasereportingand compensationpayment
Simple
Complicate Movementcontrolinoutbreak
area
Nomovementban Movementban Deliverytimeofcompensation
payment
3months 6months 1year (LBW:livebodyweight).
elsusedinthechoiceexperimentaredescribedinTable1.Detail
descriptionsofattributelevelsarepresentedinAFTable1. Com-pensationlevelisaddedasamonetaryattributeandexpressedas apercentageofthemarketpriceforthelivebodyweight(LBW) ofafatteningpig,whichis thenconvertedintoacompensation price(i.e.amountofmoneyperkilogramLBWofcullingpig)using afixedmarketpriceofVND48,000kgLBW.Thispricewas consid-eredasmeanvaluesofpigpricesbetween2010and2013thatwere collectedduringfocusgroupdiscussionandkey-informant inter-views.Compensationattributelevelsweresetat50%,70%,and80% ofpigmarketpriceperkgLBW.
2.2. Experimentdesign
AfractionalfactorialdesignwasusedwithRsoftwareversion 2.15.3(Rcoreteam,2013)(support.CEspackage).Anunlabeled choice experiment design was created by the mix-and-match method(rotation.designfunction).Achoicesetwasgeneratedby randomlyselectingonealternativefromeachsetofNalternatives. Theselectionprocesswasrepeatedwithoutreplacementuntilall alternativeswereassignedtochoicesets(Aizaki,2012).The num-berof36 choicesetsof twoalternativeswassuggestedfromR output.Giventhenumber ofattributesand levels,this number helpstoensurethatthedesignisorthogonalandbalanced(Kuhfeld, 2010).Tolimittheburdenforrespondents,itwasdividedintothree blocksof12choicesets.
2.3. Questionnairesanddatacollectionprocess
Questionnaireswerecreatedbasedonthechoicesetsusingthe questionnairefunction in thesupport.CEspackage in R(Aizaki, 2012).Eachfarmerwasaskedtoanswer12questionsequivalent to12choicesetsinoneblock.Eachchoicesetconsistedoftwo unlabeleddisease-reportingalternativesandoneopt-out alterna-tive(non-reportingalternative).Farmerswereaskedtoconsider thesethreealternativesandchoosetheoptionthatthey consid-eredthebestfortheirsituation.Includingopt-outoption which referas“donotreport”or“donotchooseanyalternative”inchoice setsallowedustolimitthebiasoftheforcedchoiceandexplore thereasonsbehindfarmer’sdecision.
2.3.1. Studyareaandsamplingframe
TheDECsurveywasconductedbetweenMarchandJuly2015 in eight communes of four districts of two Red River Delta provincesbyface-to-faceinterviewsinVietnamese.Theselection ofprovinces,districtsand communeswasbasedonpigdensity, extentof pigproduction, diversity of pigfarming systems, and occurrenceofnotifiedswineinfectiousdiseases(e.g.porcine repro-ductiveandrespiratorysyndrome(PRRS),footandmouthdisease (FMD),andclassicalswinefever(CSF)).Thetwoprovincesare sub-sequentlyreferredtointhispaperasprovince“A”andprovince“B” aspartoftheconfidentialityproceduretoensureanonymoususe ofthedata.Stratifiedrandomsamplingwascarriedoutbasedon productiontypes(mixedfarm,fatteningfarm)andproduction sys-tems(smallorlargefarm,basedonathresholdof20sowsand/or 200fatteningpigs).Thelistofpigfarmersineachcommunewas obtainedfromcommuneveterinarians.Itwasthenstratifiedinto theabove-mentionedcategories.Pigfarmers wereselected ran-domlyfromthecategoriesandinvitedtoparticipateinthestudy withthesupportofthecommuneveterinarianandtheheadofthe village.Duetothelimitednumberoflargefarmsinthestudyarea, alllargefarmsweresubjectedtointerviews.
EquationstocomputesamplesizeforDCErequiresaninitial estimationofchoiceproportion(BliemerandRose,2005)orprior estimates ofparameter values ofattributes under investigation (Bekker-Grobetal.,2015)whichareunknownbefore implement-ingthestudy.Therefore,samplesizecalculationinthisresearch isbasedonOrme’sapproachthatconsidersthenumberofchoice sets,thenumberofattributes,andthehighestnumberofattribute levels(Orme,2009).Targetsamplesizewas120pigholdingsin eachprovince(30largefarmsand90smallholders),includingboth fatteningandmixedfarmswiththeratio(1:5)likethatof fatten-ingholdingsandmixedholdingsinarea.AccordingtotheWHO guidelineondiscretechoiceexperimentsinpublichealthresearch, aminimumsamplesizeof30isrequiredforeachsubgroup (strat-ifiedgroup)ofthemainsampletoperformeconometricanalysis (WHO,2012).
2.3.2. Datacollection
TheDCEtaskstartedbysettingouttheobjectivesofthestudy, theimportanceofthefarmers’contribution,thesecurityof per-sonalinformationandtheiroralconsenttoparticipateinthestudy. Thenascenarioofdiseaseoccurrenceonthefarmwasdescribed tocreateacontextinwhichthefarmerscouldmaketheirdecision (Table2).Detaileddescriptionsofchoicesetsaswellasattributes weremadetoensureaclearunderstandingofthechoiceprocess. Probingquestionswereusedtoexplorerelatedinformationsuchas thepossibilityofsellingsick/deadpigs,priceofsickpigsaswellas thefarmer’sperceptionofmoneyobtainedinthecaseofreporting ornotreporting.Thequestionnairesweretestedwith17pig hold-ersandwereadjustedtoadapttofieldconditionssuchasspecifying infectedpigagegroupsandthethresholdpercentageofinfected pigsinthefarmtriggeringdiseasereporting.
Generalinformation wasalsocollectedduringtheinterview abouttherespondents,suchasage,education,position,andabout thepigfarms(herdsize,productionsystem).
2.4. Dataanalysis
McFadenconditionallogitmodelwasusedtoanalysisDCEdata inthisstudyasthemodeldescribesthecharacteristics(attribute andattributelevels)ofthealternativesamongwhichthefarmer choosetoreportanimaldiseaseandrequiresthesmallestsample sizecomparedtoothermodels(Hauberetal.,2016).Thesystematic utility(Vi)boundtoascenariowasmodeledasalinearfunctionof attributelevelscharacterizingthescenario.Theexpressionofthe
Table2
Anexampleofchoiceset.
Imaginethatmorethan40%ofyourpigsaresickwithclinicalsignsassuspectedcasesfromoneofthethreenotifieddiseases(PRRS,FMD,andCSF).Yoursick pigsdidnotrespondafter3–5daysoftreatmentandtheliveweightoffatteningpigsrangedfrom20to50kg.Chooseoneofthefollowingoptionsrelatedto diseasereportingtoreportswinediseaseinyourholdingtoveterinaryorlocalauthorities.
Choiceset1 AlternativeA AlternativeB NeitheralternativeAorB
Probabilityofbeing compensated 100%(certain) 50%(uncertain) Compensationlevel (VND/kgofLBW) 24,000 (50%marketprice) 33,600 (70%marketprice) Animalculledincaseof
diseasenotification
Unrecoveredpigs Allpigsatfarm Idonotwanttoreportswinedisease Diseasereportingand
compensation procedures Simple Complicate Animalmovement controlinoutbreak area Animalmovement ban Noanimalmovement ban Deliverytimeof compensation payment 3months 6months
Yourchoice(Pleasetick onlyoneboxasyour preference)
䊐 䊐 䊐
utilitymodelisasfollows:
Vi= 0+1∗possibilitycompensation+2∗compensation
+3∗animaldes+4∗administrativeprocedureof
diseasereportingandcompensation+5∗movementban+
6∗deliverycompensation+ (1)
The utility coefficients (1–6)were calculated using clogit
functioninthesupport.CEspackageinR(Aizaki,2012).
Qualita-tiveattributeswerecodedasdummyvariables,andthemonetary parameter (i.e. compensation levels) as a continuous variable. Parameterestimationwasrunseparatelyforthetotalsampleand thenaccordingtoprovince(AandB)andproductionsystem(small vs.large,mixedvs.fattening).Thegoodness-of-fitofthemodels ismeasuredusingMcFadenpseudo-Rsquared(p2)whichcanbe definedbytheloglikelihoodsofthemodelwithandwithout pre-dictors.Valuesfrom0.2to0.4forp2 indicatethegoodfitofthe model(McFadden,1977).
Therelativeimportanceofthesocio-economicfactors influenc-ingfarmers’reportingwasassessedbasedonresultsoftheutility model(i.e.thestatisticalsignificance,signandvalueofthe util-itycoefficientsofattributesandthewillingnesstopayforchanges inattributelevels)(WHO,2012).Thefarmers’willingnesstopay (WTPkl)for levellof anattribute kwascalculated astheratio betweenitsutilitycoefficientklandthemarginalutility coeffi-cientof compensation(utilitycoefficientof themonetaryunit). Theeffects of changesonsurveillance werethen modeledas a probabilityofchoosingreportingalternativeunderthreedefined scenarios.Theprobabilityofchoosingthereportingalternativewas estimatedaccordingto changesof attributelevels comparedto baselines.Threebaselinesweresetup,whichdifferedinregardto culledpigcategoriesandcompensationlevels.Thefirsttwo base-linesreferredtothecurrentsituationofswinediseasesurveillance andcontrolinVietnam(MARD,2008,2007):i)cullingofallpigsat infectedholdings(uponfirstoutbreakintheregion),ii)cullingof unrecoveredpigsonly(uponre-occurrenceinthearea).Thethird baselinewasestablishedasanunrealisticscenariowiththe com-pensationlevelat50%ofthemarketvalueofthepiganddestroying dead/unrecoveredpigsonly.Thefulldescriptionsofthethree base-linesareasfollows:
–Baseline1:uncertaintyofbeingcompensated,cullingallpigsat farm,movementrestriction,complicatedadministrative proce-dures,compensationlevelequalto70%marketpriceofpig,and gettingcompensationpaymentafter6months.
–Baseline 2: uncertainty of being compensated, destroying dead/unrecovered pigs, movement restriction, complicated administrativeprocedures,compensationlevelequalto70% mar-ket price of pig, and getting compensation payment after 6 months.
–Baseline 3: uncertainty of being compensated, destroying dead/unrecovered pigs, movement restriction, complicated administrativeprocedures,compensationlevelequalto50% mar-ket price of pig, and getting compensation payment after 6 months.
Theprobabilityofchoosingalternativeiwaspredictedfromthe expectedutilityViusingtheformulaasfollows(Gerardetal.,2008):
Pi=1/(1+e−Vi) (2)
TheexpectedutilityViofalternativeiwascalculatedas For-mula(1)usingthepath-worthutilitiesforthedifferentattributes correspondingwithdifferentlevels.
3. Results
3.1. Characteristicsofsamplepopulation
Ofthe240pigholdingscontactedforinterviewing,27people wereunavailableorunwillingtoattendtheinterviewand17people refusedtocompletethechoicesetswhentheinterviewerexplained thescenario.Finally,196respondentswereincludedintheanalysis. Themeanageoftherespondentsinbothprovinceswas48.5 years,rangingfrom28to70years(Table3).Mostrespondentswere male,accountingfor85%ofrespondentsinprovinceAand74%in provinceB.Halftherespondentshadlessthan9yearsof educa-tion,47%reportedhavinggraduatedhighschooland2%statedthey hadanintermediateorcollegedegree.Themajorityofrespondents (80%)hadexperienceinpigproduction,with10yearsormoreof pigraising.Mixedpigholdingsrepresented81%ofsampledfarms (keepingsows,growers,andfatteningpigs),while19%kept fatten-ingpigsonly.Smallpigholdingswith6sowsand100fatteningpigs onaverageweredominantinbothprovinces,rangingfrom79%in ProvinceAto88%inProvinceB.
Table3
Characteristicsofinterviewedpigfarmers.
Characteristics ProvinceA(n=97) ProvinceB(n=99) Totalsample(n=196)
Gender
Male,n(%) 82(84.5%) 73(73.7%) 155(79.1%)
Female,n(%) 15(15.5%) 26(26.3%) 41(20.9%)
Meanofage(min-max) 48.4(28–61) 48.7(31–70) 48.5(28–70)
Education
Lessthanyear9,n(%) 49(50.5%) 51(51.5%) 100(51.0%)
Highschoolgraduation,n(%) 47(48.5%) 45(45.5%) 92(47.0%)
Intermediate/College,n(%) 1(1.0%) 3(3.0%) 4(2.0%)
Experienceinpigraising(10yearsorover10years),n(%) 79(81.4%) 77(77.8%) 156(79.6%) Productiontype
Mixedfarm,n(%) 65(67.0%) 94(94.9%) 159(81.1%)
Fatteningfarm,n(%) 32(33.0%) 5(5.1%) 37(18.9%)
Productionsize,n(%)
Largefarm(20sowsand/or200fatteningpigs),n(%) 20(20.6%) 12(12.1%) 32(16.3%)
Smallholder(<20sowsand/or200fatteningpigs),n(%) 77(79.4%) 87(87.9%) 164(83.7%)
3.2. Factorsinfluencingthefarmer’sdecisiontoreportadisease
event
Utility coefficients, calculated according to the province
(Table4),provestatisticallysignificantinfluencesofallattributes (p<0.01), except for the administrative procedures of disease reporting and compensation payment on farmer’s decision to reportanimal disease.Theattributes’utility coefficients() for “probability ofbeing compensated”and “animalculling policy” arepositive(3.33and2.21,respectively),indicatingthatafarmer wouldbemorelikelytoreportanimaldiseasesincaseofcertainty ofreceivingcompensation,andincaseofacullingpolicyrestricted tonon-recoveringpigsonly.Thehighestvaluesoftheseattributes’ coefficientsindicatethattheyhavethegreatestimpactsonfarmer’s decisiontoreportanimaldiseases.Movementrestrictionhasa neg-ativeimpactonthefarmer’sdecisiontoreport(=−0.58,se=0.09). Compensationdeliverytimesof6monthsand1yearobtain neg-ative utilitycoefficients, indicating that delaying compensation paymentreducedtheprobabilityoffarmers’diseasereporting.The utilitycoefficientofcompensationlevelispositiveandsignificant, highercompensationsincreasingthelikelihoodofreporting. 3.3. Willingnesstopay(WTP)forchangingattributelevels
ThehighestWTPoffarmersisobtainedforrestrictingtheculling policytonon-recoveringpigsonlyandforshorteningthe compen-sationdeliverytimefrom1yearto3months,withaveragesofVND 34,100/kgandVND12,300/kg,respectively.Thereisnosignificant differenceintheseWTPbetweenprovinces(Table5).TheWTPfor theabsenceofmovementrestrictions(VND4700–VND13,700/kg) andthereductionofcompensationdeliverytimefrom6monthsto 3months(VND6500toVND8400/kg)aresimilarlydeemed valu-ablechangesbythefarmers.PigproducersinprovinceBarewilling togiveupmoremoneyfortheabsenceofmovementrestriction thanthoseinprovinceA(averageWTPofVND13,700/kgandVND 4700/kg,respectively).However,thisdifferenceisnotstatistically significantduetotheoverlapofconfidenceintervals.
3.4. Probabilityofanimaldiseasereportingbyfarmers
Theprobabilityoffarmerstoreportswinediseasesunder dif-ferentscenariosofcontrolpolicyisrepresentedinFig.1.Under thecurrentsituation(baselinesofthefirstandthesecond scenar-ios)(i.e.uncertaintyofbeingcompensated,movementrestriction, complicatedadministrativeprocedures,compensationlevelequal to70%ofmarketprice,andgettingcompensationpaymentafter6 months),4%(95%CI:0.9%–14.3%)offarmersarewillingtoreport
swinediseasesincaseofcullingallpigsinaffectedfarms,while26% (95%CI:6.2%–65.4%)ofpigholderswouldreportdiseasesintheir farmifonlydead/unrecoveredpigsinaffectedfarmsaredestroyed. Makingadministrativeprocedureslesscumbersome,allowingfree movementsofuninfectedpigsinthearea,andincreasingthe com-pensationpaymentto80%ofmarketvaluedoesnotsignificantly increasetheproportionoffarmerswillingtoreport(allbetween1 and3%ofreportingprobabilityinscenario1andbetween6and12% inscenario2).Incontrast,iffarmerswerecertainofbeing compen-sated,probabilityofdiseasereportingwouldriseupto52%(incase ofcullingallpigsinaffectedfarms)(95%CI:16.8%–85.3%)andto 91%(95%CI:59.7%–98.5%)(incaseofdestroyingdead/unrecovered pigsinaffectedfarms).
Scenario3,whichtakeslowercompensationlevel(50%market value)intoaccountcomparedtoscenario2,showsan intermedi-ateimprovementinreportingaccordingtothedifferentchangesin policyascomparedtothefirstandsecondscenarios(from16%to 25%reportingandapeakvalueat84%(95%CI:47.9%–96.8%)).
4. Discussion
Thepresentanalysistacklesthequestionofunder-reportingof animaldiseasesin thecase ofVietnamesepigfarmers,focusing ontheimpactofdiseasemanagementpoliciesontheirdecision. Twocomponentsoffarmers’perceptionofthepoliciesarepointed ascrucialtothisdecision:theconfidenceincompensation pay-mentandtheacceptabilityofstampingoutstrategy.Interestingly, thecompensationlevelisshowntohavelessinfluence.Therefore, improvementsindiseasereportingfromanincreaseof compen-sationlevelmaybeoutweighedbydefaultsinthesetwocrucial factors.Onthecontrary,complyingwithfarmers’preferenceson thesetwoaspectsresultsina potentialreportingprobabilityof 80%–90%.Similarlytocompensationlevel,movementrestriction anddeliverytimeofcompensationareminoralthoughsignificant decisionfactors.Theadministrativeproceduresfordisease report-ingdonotappearasanimportantfactor.
Themethodologicalchoiceofexpressingcompensationsasa priceperkgpigwasaimedatmakingthechoicemorerealistic andacceptableforfarmers.Therefore, attentionwaspaidtothe numberofpigsinthesamplingprocedureandseparateanalyses wereperformedforsmallholdersandlargerfarms,despitethelow numberoffarmersinthelattercategory(duetothelowfrequency ofthisfarmtypeinthestudyregions).Theveryclosecoefficient estimatesbetweenthetwo groupsindicatethatthechoicesets werewellunderstoodbyintervieweeswithsimilarstakesinthe decision-makingbeingrevealed.Giventhissimilarityin
decision-Table4
Utilitycoefficientestimatesforfactorsinfluencingthefarmers’decisiontoreport.
Attribute Coefficient(95%CI)
ProvinceA(n=97) ProvinceB(n=99) Totalsample(n=196)
Probabilityofbeing compensated Uncertaintyvs.Certainty 4.70*** (4.26;5.14) 2.56*** (2.30;2.81) 3.33*** (3.11;3.54) Cullingpigs
Allpigsatfarmvs.
Unrecoveredpigs 2.94*** (2.52;3.35) 1.89*** (1.63;2.16) 2.21*** (2.0;2.43) Administrativeprocedures Complicatedvs.Simple 0.001 (0.31;0.31) 0.13 (−0.11;0.37) 0.10 (−0.09;0.29) Movementcontrol Nomovementbanvs. Movementban −0.42*** (−0.69;−0.15) −0.70*** (−0.91;−0.46) −0.58*** (−0.75;−0.42)
Deliverytimeofcompensation
payment 6months 1year −0.57*** (−0.90;−0.24) −0.81*** (−1.14;−0.48) −0.42*** (−0.67;−0.17) −0.82*** (−1.07;−0.56) −0.44*** (−0.64;−0.25) −0.80*** (−1.0;−0.6)
Compensationlevels(per103
VND/kg) 0.09*** (0.06;0.11) 0.05*** (0.03;0.07) 0.06*** (0.05;0.08) Numberofrespondents 97 99 196 Numberofobservations 3492 3564 7056
Loglikelihoodofthemodel −600.82 −905.28 −1557.87
Loglikelihoodofthemodel
withoutpredictors
−1278.79 −1305.15 −2583.94
McFaden’spseudoR2 0.53 0.31 0.40
CI:ConfidenceInterval.
***statisticallysignificant(p<0.01).
Table5
Farmers’willingnesstopayforchangesofattributelevels.
Attribute WTPforchangeinattributes(103VND/kgofcullingpig)Mean(95%CI)
ProvinceA(n=97) ProvinceB(n=99) Totalsample(n=196)
Cullingpigsrestrictedto unrecoveredpigs 33.4 (25.7,46.2) 38.0 (27.2,60.8) 34.1 (27.7,43.8) Nomovementban 4.7 (1.7,8.7) 13.7 (8.0,24.0) 9.0 (6.0,13.0) Deliverytimeofcompensation
changedfrom6monthsto3 months 6.5 (2.5,12.0) 8.4 (3.0,17.1) 6.8 (3.6,11.1) Deliverytimeofcompensation
changedfrom1yearto3 months 9.2 (5.3,14.7) 16.3 (10.0,27.6) 12.3 (8.7,17.1) CI:ConfidenceInterval.
making,theanalysisofreportingprobabilitiesispresentedherefor
thewholesample.
TheWTPhighlightsthetruebarrierthatanimalcullingpolicy
posesonsurveillanceefficacy.Indeed,farmers wouldbewilling
topay43%–80% market value ofpig tobesurethat only
non-recoveringpigsaredestroyed.Classicalstamping-outpolicywould
leadfarmerstohidediseases,despitecompensationlevelsashigh
as80%marketvalueofpigaswellasthecertaintyofreceivingthis
money.Themotivesbehindtheseresultswereobtainedthrough
openqualitativeinterviewsafterDCEcompletion.First,notifiable
swinediseases suchasPRRSand FMDareendemic inVietnam
andthereforefarmersperceiveapolicyofstamping-outaslacking
relevanceastheyareusedtomanagingthesediseaseslikeother
endemicproductiondiseases.Mostfarmersbelieveinahigh
prob-abilityofsuccessfultreatmentandabetterprofittobeexpected
fromrecoveredpigsthanfromcompensationpayment.
Compen-sationlevelat80%marketvalueofpigwasconsideredascovering
breedingcostsandapartoffeedingcosts.Farmerestimatesthe
successrateoftreatmentsat50%,hencea50%chanceof
gener-atingprofit.Second,thefarmersexpecta riseinpigpricesafter
diseaseoutbreaks,creatinga particularspeculativeopportunity.
Third,thedestructionofclinicallyhealthypigsisconsidered
unac-ceptable,beingperceivedasawasteofresources,food,andvaluable
geneticasset.Indeed,thecullingofclinicallyhealthybreedingstock
isunacceptableduetothehighvalueofsowswithgood
perfor-mance(i.e.annualnumberofpigletsandpigletweightatweaning,
maternalbehavior).MostfarmersbelievethatPRRShasnoeffecton
thereproductiveperformanceofsows.Thereforetheywouldnot
acceptthecullingofasymptomaticsowsevenafterconfirmation
ofinfection.
Compensationisconsideredasacriticalelementofeffective
diseasecontrolas itencourages farmers’reportingand
compli-ancewiththecullingofconfirmedorsuspectedcasesofinfection,
thusreducingtheriskofdiseasespreading(WorldBank,2006).
Compensationrateshouldbenolessthan50%ofmarketvaluefor healthyanimalsandrecommendedratesrangefrom75to90%of thereferenceprice(WorldBank,2006).Lowercompensationcould beappliedfordeadanimalsthanfordiseasedorhealthyanimals toencourage earlyreporting(World Bank, 2006).In agreement withtheveterinaryregulationsofVietnam,thesame compensa-tionlevelwasproposedinourstudyforallcategoriesofdestroyed pigs(piglets,fatteners,growers,sows,andboar,deadorsickpigs) (MOF,2008).Thislackofacknowledgementofthehighervalueof breedingstockappearsasamajordisincentivefordisease report-ing.Inthecontextoftheavianinfluenzacompensationschemein Indonesia,contingentvaluationwasusedtoquantifysucha
dif-Fig.1.Probabilityofswinediseasereportingbyfarmersunderdifferentscenariosofcontrolpolicy.
ferentiatedwillingnesstoacceptcompensationfordifferenttypes ofbirds(Yakhshilikovetal.,2009).However,ourstudywasnot designedtoestimatethepotentialimprovementobtainedthrough a differentiated compensationschemeand this question would thusrequireaspecificapproach.
The uncertainty and lack of transparency of procedures for cullingand compensation paymentappear hereas barriers for diseasereporting.AccordingtoVietnamesediseasecontrolpolicy (MARD,2008,2007),cullingpigsatfarmsafterconfirmationof dis-easecanapplyforallpigsincaseofanewoutbreakoriftheoutbreak occursinalimitedarea(fewhouseholdsinthevillage/commune) toruleoutdiseaseearlier.Cullingofsolelynon-recoveringpigsis appliediftheoutbreakcoversalargeareaoranareawherethe dis-easehasoccurredseveraltimes.Thisflexibilityofproceduremakes thingsunclearforfarmers.Theseresultssuggesttheimportanceof clarifyingcullingruleswithadequatecommunicationstrategy.
Findingsofthestudyalsoemphasizethewideracceptabilityof restrictedcullingandtheinteresttoapplythisstrategyina max-imumofrelevantcases.Theeffectivenessofsuchtargetedculling onPRRS,FMD,and CSFdiseasecontrolinVietnamwould prob-ablybeverylimited butsimilartothecurrent situationofhigh levelofunder-reportingandlocaldiseasemanagementstrategy. Howeveritwouldgreatlyimprovethesensitivityofthecurrent surveillancesystemsprovidesbetterepidemiologicalinformation onfield diseasesituationand help rebuilding farmers’trust on nationalsurveillance and control policies.On thelong run this wouldprovidestrongerbasisfordefiningnovelcontrolstrategies incollaborationwithfarmersandlocalstakeholders.
Eventhoughtheadministrativeproceduresdonotshowhere asignificantinfluenceonfarmers’willingnesstoreport(p>0.05), itisexpectedthatsimplifiedandtransparentprocedureswould promoteclarityofthesystemandfarmers’trust,whicharefactors showingasignificanteffectonreporting.Adistrustinveterinary authoritiesandcontrolmeasuresis,indeed,listedasanimportant barriertodiseasereporting(Bronneretal.,2014;Elbersetal.,2010; Palmeretal.,2009).Hereagain,thetimelinessandreliabilityof compensationdeliveryinterveneascriticalelementsinbuilding trustbetweenfarmersandofficials(WorldBank,2006).
InopeninterviewsfollowingtheDCE questionnaire,farmers furtherjustifiedmovementrestrictionasanobstacletoreporting bytheextensionofpigkeepingtimeandincreaseofinputcost. Thesethenindicatedawillingnesstopayforlaboratorytestsincase ofoutbreakintheirregiontomaintaintherighttofreemovement. Thiswillingnessmightalsobefurtherexploredthoughcontingent valuationandtheinstitutionalfeasibilityofsuchanarrangement wouldthenneedtobeassessed.Continuityofbusinessisa par-ticularmodalityoffootandmouthdiseaseresponseintheUnited States,tobeappliedincombinationwithquarantineand move-mentcontrol (USDA, 2015).It allows non-affectedfarmstosell non-infectedanimalsornon-contaminatedanimalproductsoutof thecontrolarea.Thisisasolutiontomaintainagriculturalbusiness operationswhilealsomitigatetheriskofdiseasespread.However, itrequiresnotonlysamplingandtestinganimalsbutalsoassessing farm’sbio-securitybeforemovement(USDA,2015).
FewstudieshaveappliedDCEinVietnaminthedomainof ani-malscience(Costalesetal.,2008;Roessleretal.,2008;Zhangetal., 2014)anditsuseinanimalhealthsurveillanceisevenmore
lim-ited.DCEprovidesquantitativeevidenceoftherelativeimportance offactorsinfluencingindividualdecision-making,andclarifiesthe trade-offsbetweengivenattributes,thuspredictingtheprobability ofuptakeinspecifiedalternatives(WHO,2012).AlthoughDCEis basedonhypotheticalchoiceratherthanactualchoice,theresults ofstudiesontheenvironment(Adamowiczetal.,1994)and pub-lichealth(Lambooijetal.,2015)haveindicatedtheconsistency ofstatedandrevealedpreferences.Toourknowledge,thereisno studycomparingtheresultsofstatedandrevealedpreferenceson animalhealthsurveillance.Theunobservablenatureofthechoice ofnotreportingdiseasesmakesitdifficulttoconsiderprotocols abletomakesuchacomparison.DCEallowsforthisestimationby theinclusionofanopt-outalternativerepresentingnon-reporting behaviorinthechoiceset.Thereforethis methodologicalchoice revealscrucialtoapplyDCEtosurveillanceevaluationinVietnam, whereunderreportingofanimaldiseaseisanimportantfeature.
5. Conclusion
DCEprovedfruitfulintheanalysisofreportingthebehaviorof pigfarmersinVietnam.Moreparticularly,ithighlightsthe impor-tanceofgeneralortargetedcullingstrategy.Thequantificationof expectedimprovementsofreportingprobabilitymaybeusedto modeltheexpectedbenefitsofinstitutionalchangesinsurveillance policyinVietnam.
Acknowledgements
ThisstudywascarriedoutintheframeworkoftheCIRAD- REVA-SIAproject,theCIRADandIRDjointyoungresearcherteamproject JEAI-SWEIDandwiththesupportofGREASEnetwork.Wethank theveterinariansofthesubdepartmentofanimalhealthand vet-erinarystations,andthelocalauthoritiesinthestudyareafortheir supportindatacollectionandallinterviewedpigholdersfortheir timeandparticipationinthisstudy.
AppendixA. Supplementarydata
Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.prevetmed.2017. 01.002.
References
Adamowicz,W.,Louviere,J.J.,Williams,M.,1994.Combiningrevealedandstated
preferencemethodsforvaluingenvironmentalamenities.J.Environ.Econ. Manag.26,271–292.
Aizaki,H.,2012.Basicfunctionsforsupportinganimplementationofchoice
experimentsinR.J.Stat.Softw.50,1–24.
Bekker-Grob,E.W.,deDonkers,B.,Jonker,M.F.,Stolk,E.A.,2015.SampleSize
RequirementsforDiscrete-ChoiceExperimentsinHealthcare:aPractical
Guide.Patient–Patient-CenteredOutcomesRes.8,373–384.http://dx.doi.org/
10.1007/s40271-015-0118-z.
Bliemer,M.C.J.,Rose,J.M.,2005.EfficinecyandSampleSizeRequirementsfor
StatedChoiceStudies.WorkingPaperITLS-WP-05-08.Instituteoftransport andlogisticsstudies.TheUniversityofSydney.
Boxall,P.,Adamowicz,W.,Swait,J.,Williams,M.,Louviere,J.J.,1996.Acomparision
ofstatedpreferencemethodsforenvironmentalvaluation.Ecol.Econ.18, 243–253.
Bronner,A.,Hénaux,V.,Fortané,N.,Hendrikx,P.,Calavas,D.,2014.Whydofarmers
andveterinariansnotreportallbovineabortions,asrequestedbytheclinical brucellosissurveillancesysteminFrance?BMCVet.Res.10,93.
Cameron,A.,2012.ManualofBasicAnimalDiseaseSurveillance.Interafrican
BureauforAnimalResources,Available:
www.au-ibar.org/.../1546-manual-of-basic-animal-disease(Accessed:August 20,2015).
Clift,K.H.,Weaver,J.,Frazer,J.L.,2006.Rebuildingapassivesurveillanceprogram.
In:Proceedingsofthe11thInternationalSymposiumonVeterinary EpidemiologyandEconomics,Cairns,Australia.
Costales,A.,Son,N.T.,Lapar,M.,Tioncgo,M.,2008.Determinantsofparticipationin
contractfarminginpigproductioninNorthernVietnam(ResearchreportNo.
8-4).Pro-PoorLivestockPolicyInitiative(PPLPI).
Doherr,M.G.,Calavas,D.,Cameron,A.,Dufour,B.,Greiner,M.,Gustafson,L.,
Hoinville,L.,Knopf,L.,Roger,F.,Stärk,K.D.C.,Salman,M.D.,2012.First
internationalconferenceonanimalhealthsurveillance(ICAHS).Prev.Vet.
Med.105,165–168,http://dx.doi.org/10.1016/j.prevetmed.2012.03.014.
Dung,D.H.,Taylor,N.M.,MacLeod,A.,2006.Improvingveterinarysurveillancein
Vietnam-aknowledgemanagementapproach.In:The11thInternational
SymposiumonVeterinaryEpidemiologyandEconomics.International
SymposiaonVeterinaryEpidemiologyandEconomics,Cairns,Australia,
Available:www.sciquest.org.nz(Accessed:August13,2015).
Elbers,A.R.W.,Gorgievski-Duijvesteijn,M.J.,vanderVelden,P.G.,Loeffen,W.L.A.,
Zarafshani,K.,2010.Asocio-psychologicalinvestigationintolimitationsand
incentivesconcerningreportingaclinicallysuspectsituationaimedat
improvingearlydetectionofclassicalswinefeveroutbreaks.Vet.Microbiol.
142,108–118,http://dx.doi.org/10.1016/j.vetmic.2009.09.051.
Geering,W.A.,Roeder,P.L.,Obi,T.U.,1999.ManualonthePreparationofNational
AnimalDiseaseEmergencyPreparednessPlans,FAOAnimalHealthManual. FoodandAgricultureOrganizationoftheUnitedNations,Rome.
Gerard,K.,Shanahan,M.,Louviere,J.,2008.Usingdiscretechoicemodellingto
investigatebreastscreeningparticipation.In:UsingDiscreteChoice ExperimenttoValueHealthandHealthCare.Springer,pp.117–137.
Gilbert,W.,Hasler,B.,Rushton,J.,Flood,T.,Hoinville,L.,Brouwer,A.,Tearne,O.,
2012.Enhancingpassivesurveillance:aneconomicevaluationofalternative
methods.In:Proceedingsofthe13thInternationalSymposiumonVeterinary EpidemiologyandEconomics.InternationalSymposiaonVeterinary EpidemiologyandEconomics,Belgium,Netherland.
Hadorn,D.C.,Stärk,K.D.C.,2008.Evaluationandoptimizationofsurveillance
systemsforrareandemerginginfectiousdiseases.Vet.Res.39,57,http://dx.
doi.org/10.1051/vetres:2008033.
Hadorn,D.C.,Haracic,S.,Stärk,K.D.,2008.Comparativeassessmentofpassive
surveillanceindisease-freeandendemicsituation:exampleofBrucella
melitensissurveillanceinSwitzerlandandinBosniaandHerzegovina.BMC
Vet.Res.4,52,http://dx.doi.org/10.1186/1746-6148-4-52.
Hanley,N.,Wright,R.E.,Adamowicz,V.,1998.UsingChoiceExperimentstovalue
theenvironment.Environ.Resour.Econ.11,413–428.
Hauber,A.B.,González,J.M.,Groothuis-Oudshoorn,C.G.M.,Prior,T.,Marshall,D.A.,
Cunningham,C.,IJzerman,M.J.,Bridges,J.F.P.,2016.Statisticalmethodsforthe
analysisofdiscretechoiceexperiments:areportoftheISPORconjointanalysis
goodresearchpracticestaskforce.ValueHealth19,300–315,http://dx.doi.
org/10.1016/j.jval.2016.04.004.
Howard,K.,Salkeld,G.,2009.Doesattributeframingindiscretechoice
experimentsinfluencewillingnesstopay?Resultsfromadiscretechoice
experimentinscreeningforcolorectalcancer.ValueHealth12,354–363,
http://dx.doi.org/10.1111/j.1524-4733.2008.00417.x.
Kjaer,T.,2005.AReviewoftheDiscreteChoiceExperiment-withEmphasisonits
ApplicationinHealthCare,Available:http://static.sdu.dk/mediafiles//Files/
OmSDU/Centre/cistsundoke/Forskningsdokumenter/publications/ Working%20papers/20051pdf.pdf(Accessed:August18,2015).
Kuhfeld,W.F.,2010.ExperimentalDesign,Efficiency,Coding,andChoiceDesigns,
Available:http://support.sas.com/techsup/tnote/tnotestat.html#market
(Accessed:December20,2014).
Lagerkvist,C.J.,Carlsson,F.,Viske,D.,2006.Swedishconsumerpreferencesfor
animalwelfareandbiotech:achoiceexperiment.AgBioForum9,51–58.
Lambooij,M.S.,Harmsen,I.A.,Veldwijk,J.,Melker,H.,de,Mollema,L.,Weert,Y.W.,
van,Wit,de,G.A.,2015.Consistencybetweenstatedandrevealedpreferences:
adiscretechoiceexperimentandabehaviouralexperimentonvaccination
behaviourcompared.BMCMed.Res.Methodol.15,19,http://dx.doi.org/10.
1186/s12874-015-0010-5.
Liljenstolpe,C.,2005.Valuinganimalwelfarewithchoiceexperiments:an
applicationtoSwedishpigproduction.In:PresentedattheEuropean AssociationofAgriculturalEconomist,Copenhagen,Denmark,August24–27.
Limon,G.,Ruiz,H.,Balanza,M.E.,Guitian,J.,2012.Amixedmethodapproach
investigatingfactorsinfluencingreportingofanimaldiseaseamong smallholdersinsouthernBolivia|SciQuest.In:Proceedingsofthe13th InternationalSymposiumonVeterinaryEpidemiologyandEconomics, Belgium,Netherland.
MARD,2007.MinistryofAgricultureandRuralDevelopment.DecisionNo
5/2007/QD-BNNonpreventionandcontroloffootandmouthdisease(FMD),
dated22/1/2007.
MARD,2008.MinistryofAgricultureandRuralDevelopment.DecisionNo
60/2008/QD-BNNonpreventionandcontrolofPorcinereproductiveand
respiratorysyndrome(PRRS),dated15/7/2008.
MOF,2008.MinistryofFinance.CirculationNo80/2008/TT-BTConfinancial
supportforpreventionandcontrolactivitiesinanimalhealth,dated18/9/2008.
McFadden,D.,1974.ConditionalLogitAnalysisofQualitativeChoiceBehavior,In:
FrontiersofEconomics.AcademicPress,London,pp.105–142.
McFadden,D.,1977.QuantitativeMethodsforAnalyzingTravelBehaviourof
Individual:SomeRecentDevelopments.CowlesDiscussionPaperNo474. Cowlesfoundationresearchforeconomics.YaleUniversity.
McIntyre,L.H.,Davies,P.R.,Perkins,N.R.,Morris,R.S.,Jackson,R.,2003.A
comparisonbetweenthreesourcesofendemicanimaldiseasesurveillance data.In:Proceedingsofthe10thInternationalSymposiumonVeterinary EpidemiologyandEconomics.InternationalSymposiaonVeterinary EpidemiologyandEconomics,VinadelMar,Chile,p.334.
Oparinde,A.,Birol,E.,2011.FarmHouseholds’PreferenceforCash-Based
CompensationVersusLivelihood-EnhancingPrograms,Available:https://core.
Orme,B.,2009.Samplesizeissuesforconjointanalysisstudies.SawtoothSoftware
ResearchPaperSeries.SawtoothSoftware;2009.
Otienno,D.J.,Ruto,E.,Hubbard,L.,2011.Cattlefarmers’preferencesfordisease
freezonesinKenya:anapplicationofthechoiceexperimentmethod.J.Agric. Econ.62,207–224.
Palmer,S.E.,Sully,Sarah,Max,M.,Farida,Fozdar,2009.Farmers,animaldisease
reportingandtheeffectoftrust:astudyofwestaustraliansheepandcattle farmers.RuralSoc.19.
Palmer,S.,2009.FactorsAffectingLivestockDiseaseReportingandBiosecurity
Practices:AStudyinWestAustralianSheepandCattleProducers.Murdoch University,Australia.
Rcoreteam,2013.R:ALanguageandEnvironmentforStatisticalComputing.R
FoundationforStatisticalComputing,Vienna,Austria.,pp.2013(Available:
http://www.R-project.org/).
Rao,K.D.,Ryan,M.,Shroff,Z.,Vujicic,M.,Ramani,S.,Berman,P.,2013.Rural
clinicianscarcityandjobpreferencesofdoctorsandnursesinIndia:adiscrete
choiceexperiment.PLoSOne8,e82984,http://dx.doi.org/10.1371/journal.
pone.0082984.
Roessler,R.,Drucker,A.G.,Scarpa,R.,Markemann,A.,Lemke,U.,Thuy,L.T.,Valle
Zárate,A.,2008.Usingchoiceexperimentstoassesssmallholderfarmersá
preferencesforpigbreedingtraitsindifferentproductionsystemsin
North–WestVietnam.Ecol.Econ.66,184–192,http://dx.doi.org/10.1016/j.
ecolecon.2007.08.023,SpecialSection:IntegratedHydro-EconomicModelling
forEffectiveandSustainableWaterManagement.
Sadique,M.Z.,Devlin,N.,Edmunds,W.J.,Parkin,D.,2013.Theeffectofperceived
risksonthedemandforvaccination:resultsfromadiscretechoiceexperiment.
PLoSOne8,e54149,http://dx.doi.org/10.1371/journal.pone.0054149.
Thong,N.T.,Hung,N.M.,2014.PositioningVietnam’sPangasiusCatfishinthe
Frenchmarketusingdiscretechoicemodel.J.Sci.Technol.Dev.17.
USDA,2015.Foot-and-MouthDisease(FMD)Reponse:ReadyReference
Guide-quarantine,MovementControl,andContinuityBusiness,Availableat
ttps://www.aphis.usda.gov/animalhealth/emergencymanagement/
downloads/fmdrrgcobqmcplan.pdf(Accesseddate:11/6/2016).
WHO,2012.HowtoconductaDiscreteChoiceExperimentforhealthworkforce
recruitmentandretentioninremoteandruralareas:Userguidewithcase
studies.
WorldBank,2006.EnhancingcontrolofHighlyPathogenicAvianInfluenzain
DevelopingCountriesthroughcompensation:Issuesandgoodpractice.
Xu,P.,Wang,Z.,2014.Countryoforiginandwillingnesstopayforpistachios:a
Chinesecase.Agric.FoodEcon.2,1–16.
Yakhshilikov,Y.,Birol,E.,Tiongco,M.,Narrod,C.,Friedman,J.,2009.Acontigent
valuationstudyonIndonesianfarmers’willingnesstoacceptcompensationfor
poultry.ILRI/IFPRI-HPAIResearchBriefNo.18,6pp.
Zander,K.K.,Garnett,S.T.,2011.Theeconomicvalueofenvironmentalserviceson
indigenous-heldlandsinAustralia.PLoSOne6,e23154.
Zhang,H.,Kono,H.,Kubota,S.,2014.Anintegratedepidemiologicalandeconomic
analysisofvaccinationagainsthighlypathogenicporcinereproductiveand
respiratorysyndrome(PRRS)inthuathienhueprovince.Vietnam.
Asian-Australas.J.Anim.Sci.27,1499–1512,http://dx.doi.org/10.5713/ajas.