Heterogeneous risk perceptions : the case of poultry meat purchase intentions in Finland

In . J. En i on. Res. Public Heal h 2013, 10, 4925-4943; doi:10.3390/ije ph10104925 In e na ional Jou nal o En i onmen al Resea ch and Public Heal h ISSN 1660-4601 www.mdpi.com/jou nal/ije ph A icle He e ogeneous Risk Pe cep ions: The Case o Poul y Mea Pu chase In en ions in Finland Jaakko Heikkilä 1,*, Eija Pou a 1,†, Sa i Fo sman-Hugg 1,† and Johanna Mäkelä 2,† 1 MTT Economic Resea ch, La oka anonkaa i 9, Helsinki FI-00790, Finland; E-Mails: [email p o ec ed] (E.P.); sa i. o sm[email p o ec ed] (S.F.-H.) 2 Depa men o Teache Educa ion, Uni e si y o Helsinki, P.O. Box 8 (Sil a uo enpenge 10), Helsinki FI-00014, Finland; E-Mail: johanna.m.m[email p o ec ed] † These au ho s con ibu ed equally o his wo k. * Au ho o whom co espondence should be add essed; E-Mail: [email p o ec ed]; Tel.: +358-29-531-7190; Fax: +358-20-772-040. Recei ed: 21 Augus 2013; in e ised o m: 12 Sep embe 2013 / Accep ed: 29 Sep embe 2013 / Published: 11 Oc obe 2013 Abs ac : This s udy ocused on he he e ogenei y o consume eac ions, measu ed h ough poul y mea pu chase in en ions, when acing h ee cases o isk. The he e ogenei y was analysed by la en class logis ic eg ession ha included all h ee isk cases. App oxima ely 60% o he esponden s belonged o he g oup o p oduc ion isk a oide s, in which he in en ion o pu chase isk ood was signi ican ly lowe han in he second g oup o isk neu als. In addi ion o socio-demog aphic a iables, he pu chase in en ions we e s a is ically associa ed wi h se e al a i ude-based a iables. We highligh ed some policy implica ions o he he e ogenei y. O e all, he s udy demons a ed ha isk ma e s o consume s, no all isk is equal, and consume ypes eac somewha di e en ly o di e en ypes o isk. Keywo ds: poul y; consume beha iou ; isk; ood sa e y; consume he e ogenei y OPEN ACCESS In . J. En i on. Res. Public Heal h 2013, 10 4926 1. In oduc ion and Backg ound Consume beha iou al esponses o isks ha e been s udied in ela ion o consume ood choices, as well as by ocusing on pu chase in en ions unde indi idual ood isk ypes. Typically, hese s udies ha e ocused on ei he p oduc ion- ela ed isks such as ho mone ea men [1,2], chemical con ol subs ances [3] and gene ic modi ica ion (GM) [4–7], o on animal disease isks such as Campylobac e [8], Salmonella [9,10] and bo ine spongi o m encephalopa hy (BSE) [11–13]. I has been ecognised ha he isks a e subjec i ely pe cei ed by consume s, e en i objec i e in o ma ion on p obabili ies and impac s is a ailable. Consume s espond o isks based on hei subjec i e pe cep ions o he p obabili y o he e en (i.e., isk pe cep ions) and on hei a i ude owa ds he e en (i.e., gene al posi i e o nega i e p edisposi ion) [14]. Addi ionally, consume eac ions depend on se e al ac o s ha a e isk ype-speci ic, including he ex en o which hey can a ec he isk hemsel es, how amilia hey a e wi h he isk, o wha ex en he exposu e is olun a y, how se e e he consequences a e (i espec i e o hei p obabili y), and so o h [15–18]. In his s udy, we ocused on he he e ogenei y o consume eac ions o a se o di e ing isk cases. In p e ious s udies, geog aphical di e ences among consume s ha e been documen ed as a sou ce o he e ogenei y. Consume s in he Eu opean Union (EU) ha e been pa icula ly scep ical ega ding gene ically modi ied p oduc s (e.g., [15,19]), as well as ho mone- ea ed bee [20], and he isk pe cep ion in gene al has been ound o be highe in Eu ope han in No h Ame ica o Asia [4]. Coun y-speci ic di e ences ha e also been documen ed [21,22]. The analysis o pe cep ions and a i udes owa d isk has o en ocused on he di e ences be ween expe s and he public [20,23–25]. Howe e , socio-demog aphic ac o s such as age, gende , whe he he esponden has child en and whe he he esponden li es in u ban o u al su oundings ha e also been ound o co ela e wi h he isk esponse (see, e.g., [6,10,26–28]). Beyond indi idual sou ces o he e ogenei y, segmen ing consume s could be use ul in isk s udies o assess how a ious ypes o isk a ec he e ogeneous consume segmen s and hei buying beha iou . Iden i ying he eac ions owa ds speci ic isks in sepa a e segmen s helps in a ge ing in o ma ion and p oduc de elopmen a hose segmen s ha a e mos sensi i e [29]. In addi ion o he design o policies, unde s anding he beha iou o he e ogeneous consume segmen s is a necessi y o he iable implemen a ion o policies such as isk communica ion. Segmen ing o he popula ion acco ding o in o ma ion needs and he de elopmen o in o ma ion wi h high le els o pe sonal ele ance o speci ic segmen s allows mo e e ec i e communica ion o he isks [20]. Addi ionally, i consume eac ions o isk di e acco ding o he segmen , he bene i s o p ecau iona y policies will also di e among segmen s, sugges ing ha equi y conside a ions a e also in ol ed. The e o e, segmen a ion appea s o be a p ominen app oach in a ge ing isk managemen policies, communica ion and esea ch. Howe e , only a ew s udies ha e epo ed consume segmen s based on hei ood isk pe cep ions, a i udes o beha iou al esponses. In a e iew o consume eac ions o a ood c isis, Wansink [30] o med heo e ical consume segmen s om a combina ion o isk pe cep ions and p e-exis ing a i udes owa ds isk. Va ying he le el o isk pe cep ion and a i ude in a ma ix o ou ields, he illus a ed he beha iou al esponse o isk in hese ou segmen s. In an empi ical, s udy Kennedy e al. [31] explo ed he con en o ood sa e y a i udes, which included componen s o conce n, us , In . J. En i on. Res. Public Heal h 2013, 10 4927 he desi e o a high le el o egula ion, accep ance o he numbe o su e ing people, and p e e ence o he igh o pu chase sa e o unsa e ood. Based on hese a i ude componen s, he au ho s iden i ied i e segmen s o consume s, wi h one o hem (app ehensi e consume s) bene i ing he mos om sa e y in o ma ion. Focusing on C eu z eld -Jacob disease, Payne e al. [29] used beha iou al, a i ude and isk pe cep ion in o ma ion in an expe imen al se ing o segmen consume s. They obse ed ha g oup membe ships depend on consump ion le els, bu demog aphic a iables also p o ide p o ile in o ma ion on he segmen s. These p e ious s udies ha e segmen ed consume s based on hei isk pe cep ions, isk a i udes o beha iou al esponses owa ds isk. Howe e , none o hese s udies has ocused on segmen a ion based on beha iou al esponses (i.e., pu chasing decisions) unde a ious isk ypes. The e ha e also been a ew s udies compa ing he eac ions o a ious isk ypes among he gene al public [32–34], including he case o ood isks [35–39]. Howe e , mos o hese s udies ha e ocused on simila i ies and di e ences be ween isk ypes, no he he e ogenei y o people in hei eac ions. This implies ha in ood sa e y issues he gene al p o ile o indi iduals sensi i e o a ious isk ypes is unknown. In he case o ood pu chasing decisions, i would also be bene icial o iden i y hose indi iduals who a e pa icula ly sensi i e o any o he haza ds. In his s udy, he segmen s o consume s we e o med based on di e ences in beha iou al in en ions, bu hey we e iden i ied acco ding o isk- ela ed a i udes and socio-demog aphic o geog aphic ac o s. Mo eo e , we simul aneously conside ed se e al ypes o isk. By ecognizing ha belie s p ecede a i udes, which u he a ec pu chasing in en ions, we ollowed he ideas o Ajzen [9,40–42] in a simpli ied o m, whe e he a i udes owa d isk we e assumed o p ecede pu chasing beha iou s. Ins ead o ocusing on subjec i e no ms, which in he ood-buying se ings a e ypically inside household no ms, o on pe cei ed beha iou al con ol, which in he case o hypo he ical isks is challenging o conc e ize in a su ey, we ocused on a ious ypes o a i udes ha can be assumed o a ec he buying in en ions o poul y unde a ious ypes o isk. Al hough he e is ex ensi e li e a u e on a i udes ela ed o ood isks ( o a e iew, see [43]), he con en o a i ude cons uc s applied o model subjec i e isk e alua ions by consume s a ies conside ably [5,20]. We applied he segmen a ion concep o he beha iou al eac ions o Finnish consume s o a ious isks om poul y mea . The case o poul y is in e es ing o se e al easons. The consump ion o poul y mea in Finland has inc eased signi ican ly in ecen yea s. The inc eased demand has pa ly been co e ed by impo s, meaning ha in e na ional sa e y isks can apidly en e he ma ke s due o globalisa ion o he ood chain. A he same ime, consume s ha e become inc easingly in e es ed in how mea is p oduced and how sa e i is. I sa e y is comp omised, sho - o long- e m educ ions in demand a e o be expec ed, as demons a ed in Eu ope, o ins ance, by a ian in luenza H5N1 [44,45]. Fu he mo e, poul y mea can be associa ed wi h se e al isk ypes and hus i s he aims o he s udy. We p esen ed he esponden s wi h h ee di e en isk cases ela ed o a biological haza d (inc eased isk o Salmonella in poul y), a chemical haza d (chemical ea men o poul y mea ) and a echnological haza d (use o gene ically modi ied eed), and analysed how a ious pe cei ed isks a ec pu chase in en ions, and whe he consume segmen s exis ha eac uni o mly o all o hese isk cases. By analysing he ole o a i ude cons uc s such as heal h o ien a ion, domes ic p e e ence, GM nega i i y and sa e y o ien a ion, we also aimed o p o ide in o ma ion on he cha ac e is ics o he segmen s. In . J. En i on. Res. Public Heal h 2013, 10 4928 2. Me hods and Da a 2.1. The Su ey Da a Since sa e y ea u es ha e no obse able a ia ion o di ec p ice in s o es, expe imen s [6] o su eys e ealing pu chase in en ions need o be used o es ima e hei demand. He e, consume su ey da a on poul y mea consump ion we e used. An online In e ne su ey conduc ed in No embe 2007 p o ided in o ma ion on he eac ion o consume s o isk ypes. The da a we e collec ed by sampling 2,500 esponden s om he In e ne panel o a p i a e su ey company, Talous u kimus. The panel comp ised 20,000 esponden s who had olun ee ed o pa icipa e in he panel [46]. The o he wise andom sample was balanced owa d popula ion by adding 70 esponden s ep esen ing he ca ego y o olde esponden s. The consume da a se (N = 1,312), wi h a esponse a e o 51%, was a ep esen a i e sample o Finnish In e ne use s be ween he ages o 18 and 79 yea s. The sample was ep esen a i e o he gene al popula ion ega ding gende , age, income and geog aphical loca ion (Table 1), bu he educa ional le el in he sample was somewha highe and he p opo ion o indi iduals wi h child en in he amily somewha lowe han in he gene al popula ion. In he da a se , 95% o he consume s included poul y in hei mon hly die and almos hal o hem pe cei ed ha he p opo ion o poul y had inc eased in hei die du ing he p e ious i e yea s. Table 1. Desc ip i e s a is ics o he esponden s in he da a se and he 18- o 79-yea -old popula ion in Finland. In da a In popula ion * P opo ion o emales, % 51 51 Mean age, yea s 49 47 P opo ion o people wi h a highe educa ional le el (college o uni e si y), % 38 26 P opo ion o people li ing in households wi h a g oss income unde €40,000, % 42 42 P opo ion o people wi h child en (<18 yea s) in he amily, % 29 42 P opo ion o people li ing in no he nmos Finland (Lapland), % 4 4 P opo ion o consume s ha ing poul y in hei mon hly die , % 95 N/A P opo ion o consume s who ha e inc eased he sha e o poul y in hei die du ing he las i e yea s, % 49 N/A * Sou ce: www.s a . i, 2009. N/A deno es da a no a ailable. Fo dependen a iables, he su ey measu ed he condi ional pu chase in en ions o he esponden s ega ding poul y p oduc s unde h ee di e en isk cases. These h ee cases a e he e collec i ely called “ isk p oduc s”, bu i should be no ed ha he las wo cases a e objec i ely only po en ially iskie han s anda d p oduc s. A isk p oduc he e hus e e s o isk as pe cei ed by he consume . Case 1 conce ned biological isk, he zoono ic isk ela ed o Salmonella, wi h a quan i a i e explana ion. The le els o mo bidi y and mo ali y de i ed om he Finnish Na ional Salmonella Con ol P og amme we e p o ided o he esponden s. The esponden s we e old ha 3,300 pe sons annually become mildly and 400 se iously ill ( equi ing a en ion om a doc o o hospi al ea men ) a e ea ing poul y mea . They we e asked o conside a si ua ion whe e 19,800 pe sons become mode a ely ill, 2,400 se iously ill and addi ionally ou pe sons die annually due o ea ing poul y mea . Hal o he esponden s, chosen andomly, we e asked how he isk would a ec hei consump ion In . J. En i on. Res. Public Heal h 2013, 10 4929 decisions i he poul y we e hal o he cu en p ice. These esponden s we e p o ided a esponse scale o ou al e na i es o inc eased use (25%, 50%, 75%, 100% inc ease), one al e na i e o “no e ec ”, and ou al e na i es o dec eased use (25%, 50%, 75%, no use). Hal o he esponden s we e asked abou hei eac ions i he p ice emained unchanged. Fo hem, he esponse scale was “no e ec ” and ou al e na i es o dec eased use (25%, 50%, 75%, no use). Case 2 conce ned chemical sa e y, and he esponden s we e p esen ed a scena io whe e chemical ea men is an al e na i e app oach o main ain p oduc sa e y: “The sa e y o poul y mea in Finland is ensu ed wi h good p oduc ion hygiene h oughou he p oduc ion chain. An al e na i e app oach is o ea he mea p oduc s be o e hey each he consume wi h chemicals o elimina e po en ial pa hogens. In e na ional ade nego ia ions may lead o he ma ke en y o chemically ea ed mea in he EU.” The exac name o he possible chemical (chlo ine) was no men ioned in he su ey. A e his in o ma ion, a ou -le el scale was p esen ed o he esponden s o indica e hei willingness o choose he p oduc : (a) i i was cheape han he con en ional p oduc ; (b) i i had he same p ice as he con en ional p oduc ; (c) e en i i was mo e expensi e han he con en ional p oduc ; o (d) would no choose he p oduc a all. Case 3 ela ed o echnological isk, and i was explained o he esponden s ha gene ically modi ied soya was no a he ime o he su ey used in poul y p oduc ion in Finland bu was a u u e al e na i e ( eed con aining GM soy en e ed he Finnish eed ma ke s in 2013). The pu chase in en ion was asked i GM eed was used in p oduc ion. The same scale was implemen ed as in Case 2. In all h ee cases, he esponden s who would s ill buy he p oduc (al hough possibly a a lowe p ice) we e coded in he “yes” ca ego y, and he esponden s who would s op buying al oge he we e coded in he “no” ca ego y. In o he wo ds, in Case 1, a dec eased amoun o use was also coded in he “yes” ca ego y. We es ed al e na i e modelling app oaches. Fi s , we used he whole ange o scales o he dependen a iables, i.e., he o iginal ou -le el scale o chemical isk and echnological isk and he o iginal i e-le el scale o biological isk. We unde ook o dinal eg ession (o de ed logi model). Howe e , he es o pa allel lines implied ha he slope coe icien s in he models we e no equal ac oss all esponse ca ego ies. Fo such da a, he app op ia e model would be he mul inomial logi model ins ead o o dinal eg ession. We es ima ed he mul inomial logi models, bu in he case o he ou -ca ego y dependen a iable, no s able solu ion was ound because o he low p opo ion o esponses in one ca ego y. The model wi h a h ee-ca ego y dependen a iable did no p o ide us wi h conside ably mo e in o ma ion han he bina y app oach, and because he mul ica ego y app oach made he compa isons be ween he isk ypes mo e di icul , we ended up using he bina y coded a iables, as de ailed abo e. To p o ide independen a iables o modelling, he su ey included ques ions ela ing o consume pa e ns o using poul y mea , se e al a i ude and belie ques ions ela ing o poul y p oduc ion, and socioeconomic backg ound a iables. The ela i e amoun o poul y mea in he die was cons uc ed om he i e-class scale measu ing he equency o poul y consump ion. The inal a iable was exp essed ela i e o he consump ion equency o o he ypes o ood ( ege a ian, ish, bee , po k). The a i ude a iables included heal h o ien a ion, domes ic p e e ence, GM nega i i y and sa e y o ien a ion (Table 2). S uc u ed ocus g oup discussions [47] on poul y mea wi h di e en ypes o consume s we e u ilized in de eloping he measu es. The salien issues om he ocus g oups we e In . J. En i on. Res. Public Heal h 2013, 10 4930 also e lec ed wi h measu es a ailable om he li e a u e (e.g., [48–51]). The eliabili y o heal h o ien a ion, domes ic p e e ence and GM nega i i y measu es was analysed wi h C onbach’s alpha (Table 2). Heal h o ien a ion was measu ed wi h eigh s a emen s om he measu es o gene al heal h in e es in he ood se ing p esen ed by Roininen e al. [48] using a Like scale om one ( ully disag ee) o i e ( ully ag ee). One o he i ems was d opped o ob ain an accep able C onbach’s alpha and se en i ems o med he inal a iable, summa ised by he mean o he i em esponses. The s a emen s we e as ollows: (1) Fo me i is impo an ha my daily ood is low in a ; (2) Fo me i is impo an ha my daily ood con ains plen y o i amins and mic onu ien s; (3) The heal hiness o snacks is i ele an o me ( e e sed); (4) I do no a oid any oods, e en i hey aise my choles e ol le el ( e e sed); (5) I make su e ha all he ood I ea is heal hy; (6) I ea wha I wan and do no ca e much abou he heal hiness o ood ( e e sed); (7) I always ollow a heal hy and balanced die . Table 2. Desc ip i e s a is ics o he poul y use a iable and he a i ude-based a iables, including C onbach’s alphas whe e applicable. Minimum Maximum Mean S d. de ia ion C onbach’s alpha Rela i e amoun o poul y mea in he die 0.26 2.27 0.81 0.26 A i ude-based a iables Heal h o ien a ion 1.00 5.00 3.63 0.73 0.857 Domes ic p e e ence 1.17 5.00 4.17 0.64 0.786 GM nega i i y 1.00 5.00 3.94 1.03 0.931 Sa e y o ien a ion 0.26 2.37 1.14 0.18 Based on he ocus g oups and p e ious s udies on coun y o o igin e ec s [50,51], he a i ude owa ds Finnish p oduc ion was measu ed by six belie s a emen s ega ding domes ic poul y p oduc ion: (1) By buying Finnish poul y mea I can a ec employmen in Finland; (2) I buy Finnish poul y mea , because I alue Finnish p ima y p oduc ion; (3) I do no pay a en ion o he o igin o he poul y mea ( e e sed); (4) Finnish poul y mea canno be dis inguished om impo ed poul y mea ( e e sed); (5) I buy Finnish poul y mea because I us Finnish ood p oduc ion; (6) Finnish poul y p oduc s should be ma ked wi h a label o in o m he consume s abou he o igin. A Like scale om one ( ully disag ee) o i e ( ully ag ee) was used. The inal a iable o domes ic p e e ence was he mean o he esponses o he six s a emen s. Ins ead o gene al a i udes owa ds GM (e.g., [49]), he GM nega i i y s a emen was a ge ed speci ically a he use o GM eed in poul y p oduc ion. The measu e was o med as a mean om ou s a emen s measu ing belie s ela ed o GM eed in he poul y p oduc ion chain wi h a Like scale om one ( ully disag ee) o i e ( ully ag ee): (1) The impac s o mea g own using GM eed on humans a e no known; (2) The impac s o GM eed on p oduc ion animals a e no known; (3) The impac s o GM eed on he na u al en i onmen a e no known; (4) The gene ic modi ica ion o ag icul u al plan s is e hically ques ionable. The sa e y o ien a ion was cons uc ed om a ques ion measu ing he impo ance o se e al quali y cues [52–54] aced when buying poul y p oduc s. In he inal a iable, he esponden s’ a ing o sa e y om one (no a all impo an ) o i e (ex emely impo an ), was di ided by he mean o he In . J. En i on. Res. Public Heal h 2013, 10 4931 impo ance o he o he a ibu es: as e, heal hiness, du a ion o use, p ice, easiness o use, a ac i e appea ance, sui able package size, wide ange o a ailable p oduc s, and clea ly ma ked coun y o o igin. In o he wo ds, sa e y o ien a ion measu ed he impo ance o sa e y ela i e o he impo ance o he o he a ibu es. The co ela ions be ween he a i ude a iables we e ela i ely low, being unde 0.3 be ween all he a iables. Al hough low, he co ela ions we e signi ican be ween domes ic p e e ence and all he o he a i ude a iables, indica ing a gene al us in domes ic ood, as also epo ed in many p e ious s udies. The socio-demog aphic a iables es ed in he models we e gende , age, educa ion, income, amilies wi h child en, and he esiden ial egion. Income and age we e analysed in classes, as epo ed la e in Table 5. 2.2. The S a is ical Models Fo each o he isk p oduc s, a model o pu chase in en ion was cons uc ed. As he dependen a iables we e dicho omous, he me hod used o modelling each pu chase in en ion sepa a ely was logis ic eg ession analysis [55]. The dependen a iable in he models was whe he he esponden would con inue buying poul y mea unde a change in i s sa e y o s op buying i . The independen a iables we e selec ed based on he assump ion o a i udes explaining beha iou al in en ions. We s a ed he model speci ica ion wi h co ela ion analysis iden i ying mu ually co ela ing independen a iables and a iables ha ing a signi ican co ela ion wi h each dependen a iable. The ull se o al e na i e independen a iables was used as a s a ing poin o he model de elopmen . All non-signi ican a iables (p- alue o e 0.10) we e indi idually emo ed om he model, s a ing wi h hose ha ing he highes p- alues. In his manne , he in ended pu chase was explained by he p e iously desc ibed a i ude a iables, bu also by he socio-demog aphic a iables. In addi ion, he ela i e amoun o poul y in he die was included in he model. A ew me hods a e a ailable o s udy he e ogenei y, bu i he e a e no s ong, heo y-based assump ions ega ding he sou ces o he e ogenei y, segmen a ion h ough la en class models is a good me hod o he analysis. I does no equi e a p io i assump ions ega ding he sou ces o he e ogenei y and deals well wi h mul iple a iables. The po en ial he e ogenei y o he esponden s ega ding he pu chase in en ions was he e o e analysed by la en class logis ic eg ession o bina y choices o all h ee isk p oduc s. I was also expec ed ha la en class models would imp o e he explana o y powe o he logi models. The basic assump ion in a la en class model o bina y choice da a is ha he pa ame e s o he eg ession model di e ac oss he es ima ed classes [56]. In he gene al case, he model speci ica ion is: (1) The dependen a iable is a dummy a iable o pu chasing decisions unde isk (yi ). The indexes i and e e o he indi idual and eplica ion, i.e., h ee pu chasing decisions (one o each isk p oduc ) (T = 3). Behind he obse ed a iables exis s an unobse ed nominal a iable x ha indica es K sepa a e classes, each ha ing hei own dis ibu ion o obse ed a iables, y. In he la en class model, ),()(),( 1 1 co co p ed i T i K x i p ed iii zxy zxPzzy i     In . J. En i on. Res. Public Heal h 2013, 10 4932 p io assump ions o he easons o he e ogenei y a e no needed. Ins ead, he a i udinal and socio-demog aphic a iables zco associa ed wi h he class membe ship a e empi ically examined. The index indica es each backg ound a iable. zp ed is a se o Q p edic o a iables in luencing he dependen a iable in each o he unobse ed classes. In ou case, he p edic o s we e he case-speci ic dummy a iables (Cases 1 and 2), while Case 3 (GM eed) was he e e ence le el (Q = 2). In he case o a bina y dependen a iable, he p obabili y o y = 1 ob ains a logis ic o m: (2) whe e α and β a e model es ima es. A de ailed desc ip ion o he me hodology is a ailable in Ve mu and Magidson [57]. In es ima ion, he co a ia es (i.e., a i ude and socio-demog aphic a iables) we e ac i e, hus a ec ing he solu ion, and hey we e selec ed in he model i he signi icance le el o he coe icien was below 0.10. The model p o ides p obabili ies o class membe ship o each obse a ion, and he obse a ions we e classi ied in o he mos p obable class. The la en class model o bina y choices in La en Gold so wa e was used o es ima e he model. The Bayesian in o ma ion c i e ion (BIC) indica ed ha he op imal model was a wo-class model. Al hough he Akaike in o ma ion c i e ion (AIC) con inued o dec ease, he imp o emen om he wo-class model was mino (Table 3). Table 3. Bayesian (BIC) and Akaike (AIC) in o ma ion c i e ia o selec ing he numbe o classes. Numbe o classes BIC(LL) AIC(LL) L² R² 1 4,264 4,248 3,583 0.24 2 3,571 3,478 2,783 0.56 3 3,604 3,433 2,707 0.66 4 3,668 3,419 2,664 0.57 5 3,715 3,388 2,603 0.58 6 3,786 3,381 2,566 0.58 3. Resul s The measu es o pu chase in en ions e ealed he eac ions o he esponden s o he h ee isk cases in poul y p oduc ion. In Case 1 (biological isk), abou hal o he esponden s s a ed ha hey would dec ease hei consump ion o poul y i he isk o mo bidi y and mo ali y inc eased six- old. In he case o no p ice impac , only 13% o he esponden s would con inue he use o poul y a he cu en le el and 57% a a lowe le el. I he inc eased disease isk would lead o a p ice educ ion o 50%, 20% o he esponden s would con inue o use poul y a he cu en le el. Fi e pe cen o he esponden s we e less sensi i e o isk and would conside inc easing he use o poul y i he p oduc p ice dec eased by hal . Howe e , he p opo ion o esponden s no willing o buy was he same, 30%, ega dless o he p ice e ec (Table 4). )exp(1 )exp( ),1( 1 1 p ed i q Q q qxx p ed i q Q q qxx i z z zxyP          In . J. En i on. Res. Public Heal h 2013, 10 4933 Nea ly 90% o he esponden s we e o he opinion ha hey would no choose chemically ea ed poul y mea (Case 2). Se en pe cen o he esponden s, howe e , we e willing o selec he chemically ea ed p oduc i i was cheape han con en ional mea . Table 4. Dis ibu ion o buye s based on pu chase in en ions, p opo ion o buye s, and co ela ion be ween he pu chase in en ions unde di e en ypes o isk. Dis ibu ion, % o esponden s P opo ion o buye s % Phi coe icien o co ela ion... (p- alue) Would inc ease pu chases * No e ec on pu chases Would dec ease pu chases Would no pu chase ...wi h Case 2 ...wi h Case 3 Case 1 (Biological isk) 2.3 16.3 51.4 30.0 70.0 0.056 (0.043) 0.148 (0.000) Would pu chase i cheape han con en ional Would pu chase i he same p ice as con en ional Would pu chase e en i mo e expensi e han con en ional Would no pu chase Case 2 (Chemical isk) 7.5 2.9 0.7 88.9 11.1 0.361 (0.000) Case 3 (GM- eed) 23.8 12.0 1.0 63.2 36.8 * This op ion was a ailable only o he i s sub-sample, whe e i was desc ibed ha he p ice would dec ease by hal . The column “P opo ion o buye s” is he sum o he i s h ee da a columns. O he esponden s, 63% would no selec poul y mea ed wi h GM eed (Case 3). App oxima ely 25% would selec he GM p oduc i i was cheape han he con en ional p oduc . When asked, o e 90% exp essed he opinion ha GM eed should be ma ked wi h a label. Table 4 also epo s he co ela ion coe icien s be ween he pu chase in en ions o p o ide a i s imp ession o whe he he pu chase in en ions unde a ious isks a e co ela ed wi h each o he . The co ela ion coe icien s show s onge dependency be ween he in en ions unde he wo isk ypes ela ed o p oduc ion, i.e., chemical isk and GM isk. The pu chase in en ion unde biological isk is associa ed mo e weakly wi h he in en ions unde he o he wo isk ypes. Table 5 p esen s he logis ic eg ession models and he a iables ha a ec he buying in en ion o each isk p oduc . This p o ides an oppo uni y o compa e he associa ion o pe cei ed isks wi h a ious backg ound a iables. Fo Case 1 (biological isk), he pu chase in en ion p obabili y was signi ican ly a ec ed by he amoun o poul y in he esponden ’s die . This p obably indica es ha when he p opo ion o poul y in he die inc eases, consume s pe cei e he p obabili y o in ec ion o inc ease. The e ec o he heal h o ien a ion o he esponden was sligh ly highe and had a lowe p- alue in he case o biological isk han in he models o o he p oduc s. Sa e y o ien a ion consis en ly a ec ed he choice in all models, bu he e ec was sligh ly lowe o biological isk han o he o he isks. Women we e mo e likely o educe hei pu chase in en ion as biological isk inc eased, bu he e ec o gende was e en highe o he o he isk p oduc s. Younge people eac ed In . J. En i on. Res. Public Heal h 2013, 10 4940 O e all, he s udy demons a ed ha isk ma e s o consume s, no all isk is equal, and consume ypes eac somewha di e en ly o a ious ypes o isk. Taking such di e si y in o accoun in isk assessmen , isk managemen and isk communica ion is a challenge o au ho i ies, as well as o ood manu ac u e s. Con lic s o In e es The au ho s decla e no con lic o in e es . Re e ences 1. Al nes, F. S a ed p e e ences o impo ed and ho mone- ea ed bee : Applica ion o a mixed logi model. Eu . Re . Ag ic. Econ. 2004, 31, 19–37. 2. Lusk, J.L.; Roosen, J.; Fox, J.A. Demand o bee om ca le adminis e ed g ow h ho mones o ed gene ically modi ied co n: A compa ison o consume s in F ance, Ge many, he Uni ed Kingdom, and he Uni ed S a es. Am. J. Ag ic. Econ. 2003, 85, 16–29. 3. T a isi, C.M.; Nijkamp, P. Valuing en i onmen al and heal h isk in ag icul u e: A choice expe imen app oach o pes icides in I aly. Ecol. Econ. 2008, 67, 598–607. 4. F ewe , L.; an de Lans, I.; Fishe , A.; Reinde s, M.; Menozzi, D.; Zhang, X.; an den Be g, I.; Zimme mann, K. Public pe cep ions o ag i- ood applica ions o gene ic modi ica ion— A sys ema ic e iew and me a-analysis. T ends Food Sci. Technol. 2013, 30, 142–152. 5. Cos a-Fon , M.; Gil, J.M.; T aill, B. Consume accep ance, alua ion o and a i udes owa ds gene ically modi ied ood: Re iew and implica ions o ood policy. Food Policy 2008, 33, 99–111. 6. Lusk, J.L.; Coble, K.H. Risk pe cep ions, isk p e e ence, and accep ance o isky ood. Am. J. Ag ic. Econ. 2005, 87, 393–405. 7. G une , K.G.; Läh eenmäki, L.; Nielsen, N.A.; Poulsen, J.B.; Ueland, O.; Ås öm, A. Consume pe cep ions o ood p oduc s in ol ing gene ic modi ica ion—Resul s om a quali a i e s udy in ou No dic coun ies. Food Qual. P e e . 2001, 12, 527–542. 8. Ch is ensen, T.; Mø kbak, M.; Den e , S.; Hasle , B. P e e ences o Food Sa e y and Animal Wel a e—A Choice Expe imen S udy Compa ing O ganic and Con en ional Consume s; Join O ganic Cong ess: Odense, Denma k, 2006. 9. Lobb, A.E.; Mazzocchi, M.; T aill, W.B. Modelling isk pe cep ion and us in ood sa e y in o ma ion wi hin he heo y o planned beha iou . Food Qual. P e e . 2007, 18, 384–395. 10. Golbe g, I.; Roosen, J. Measu ing Consume Willingness o Pay o a Heal h Risk Reduc ion o Salmonellosis and Campylobac e iosis. In P oceedings o 11 h Cong ess o he Eu opean Associa ion o Ag icul u al Economis s (EAAE), Copenhagen, Denma k, 24–27 Augus 2005. 11. Kaloge as, N.; Pennings, J.M.E.; an I e sum, K. Consume Food Sa e y Risk A i udes and Pe cep ions o e Time: The Case o BSE C isis. In P oceedings o 12 h Cong ess o he Eu opean Associa ion o Ag icul u al Economis s (EAAE), Gen , Belgium, 26–29 Augus 2008. 12. Mazzocchi, M.; S e ani, G.; Henson, S. Consume wel a e and he loss induced by wi hholding in o ma ion: The case o BSE in I aly. J. Ag ic. Econ. 2004, 55, 41–58. 13. La ouche, K.; Rainelli, P.; Ve me sch, D. Food sa e y issues and he BSE sca e: Some lessons om he F ench case. Food Policy 1998, 23, 347–356. In . J. En i on. Res. Public Heal h 2013, 10 4941 14. Pennings, J.M.E.; G ossman, D.B. Responding o c ises and disas e s: The ole o isk a i udes and isk pe cep ions. Disas e s 2008, 32, 434–448. 15. Slo ic, P. The isk game. J. Haza d. Ma e . 2001, 86, 17–24. 16. Ge man Ad iso y Council on Global Change. Wo ld in T ansi ion: S a egies o Managing Global En i onmen al Risks; Sp inge -Ve lag: Be lin/Heidelbe g, Ge many, 1998. 17. Fischho , B.; Slo ic, P.; Lich ens ein, S. How sa e is sa e enough? A psychome ic s udy o a i udes owa ds echnological isks and bene i s. Policy Sci. 1978, 9, 127–152. 18. Yeung, R.M.W.; Mo is, J. Food sa e y isk: Consume pe cep ion and pu chase beha iou . B . Food J. 2001, 103, 170–186. 19. Food Chain E alua ion Conso ium. E alua ion o he EU Legisla i e F amewo k in he Field o GM Food and Feed; Eu opean Commission, DG SANCO: B ussels, Belgium, 2010. 20. Ve beke, W.; F ewe , L.J.; Scholde e , J.; de B abande , H.F. Why consume s beha e as hey do wi h espec o ood sa e y and isk in o ma ion. Anal. Chim. Ac a 2007, 586, 2–8. 21. Mazzocchi, M.; Lobb, A.; T aill, W.B.; Ca icchi, A. Food sca es and us : A Eu opean s udy. J. Ag ic. Econ. 2008, 59, 2–24. 22. TNS Opinion and Social. Special Eu oba ome e 354: Food-Rela ed Risks; TNS Opinion and Social: B ussels, Belgium, 2010. 23. Sa ado i, L.; Sa io, S.; Nico a, E.; Rumia i, R.; Finucane, M.; Slo ic, P. Expe and public pe cep ion o isk om bio echnology. Risk Anal. 2004, 24, 1289–1299. 24. Me z, C.K.; Slo ic, P.; Pu chase, I.F.H. Judgmen s o chemical isks: Compa isons among senio manage s, oxicologis s, and he public. Risk Anal. 1998, 18, 391–404. 25. Hansen, J.; Holm, L.; F ewe , L.; Robinson, P.; Sandøe, P. Beyond he knowledge de ici : Recen esea ch in o lay and expe a i udes o ood isks. Appe i e 2003, 41, 111–121. 26. Adda, J. Beha io owa ds heal h isks: An empi ical s udy using he “Mad Cow” c isis as an expe imen . J. Risk Unce ain. 2007, 35, 285–305. 27. G une , K.G. Cu en issues in unde s anding o consume ood choice. T ends Food Sci. Technol. 2002, 13, 275–285. 28. Hammi , J.K.; Haninge , K. Willingness o pay o ood sa e y: Sensi i i y o du a ion and se e i y o illness. Am. J. Ag ic. Econ. 2007, 89, 1170–1175. 29. Payne, C.R.; Messe , K.D.; Kaise , H.M. Which consume s a e mos esponsi e o media-induced ood sca es? Ag ic. Resou . Econ. Re . 2009, 38, 295–310. 30. Wansink, B. Consume eac ions o ood sa e y c ises. Ad . Food Nu . Res. 2004, 48, 103–150. 31. Kennedy, J.; Wo osz, M.; Todd, E.C.; Lapinski, M.K. Segmen a ion o US consume s based on ood sa e y a i udes. B . Food J. 2008, 110, 691–705. 32. Sco on, C.R.; Taylo , L.O. Valuing isk educ ions: Inco po a ing isk he e ogenei y in o a e ealed p e e ence amewo k. Resou . Ene gy Econ. 2011, 33, 381–397. 33. Suns ein, C.R. Bad dea hs. J. Risk Unce ain. 1997, 14, 259–282. 34. Slo ic, P.; Fischho , B.; Lich ens ein, S. Fac s and Fea s: Unde s anding Pe cei ed Risk. In Socie al Risk Assessmen : How Sa e is Sa e Enough? Schwing, R.C., Albe s, W.A., Eds.; Plenum P ess: New Yo k, NY, USA, 1980; pp. 181–214. 35. Ve beke, W.; Scholde e , J.; Läh eenmäki, L. Consume appeal o nu i ion and heal h claims in h ee exis ing p oduc concep s. Appe i e 2009, 52, 684–692. In . J. En i on. Res. Public Heal h 2013, 10 4942 36. Spa ks, P.; Shephe d, R. Public pe cep ions o ood- ela ed haza ds: Indi idual and social dimensions. Food Qual. P e e . 1994, 5, 185–194. 37. Fi e-Schaw, C.; Rowe, G. Public pe cep ions o e e yday ood haza ds: A psychome ic s udy. Risk Anal. 1996, 16, 487–500. 38. S e ani, G.; Ca icchi, A.; Romano, D.; Lobb, A. De e minan s o in en ion o pu chase chicken in I aly: The ole o consume isk pe cep ion and us in di e en in o ma ion sou ces. Ag ibusiness 2008, 24, 523–537. 39. Ki k, S.F.L.; G eenwood, D.; Cade, J.E.; Pea man, A.D. Public pe cep ion o a ange o po en ial ood isks in he Uni ed Kingdom. Appe i e 2002, 38, 189–197. 40. Ajzen, I. The heo y o planned beha io . O gan. Hum. Decis. P ocess. 1991, 50, 179–211. 41. Ajzen, I.; Madden, T.J. P edic ion o goal-di ec ed beha io : A i udes, in en ions and pe cei ed beha io al con ol. J. Exp. Soc. Psychol. 1986, 22, 453–474. 42. Cook, A.J.; Ke , G.N.; Moo e, K. A i udes and in en ions owa ds pu chasing GM ood. J. Econ. Psychol. 2002, 23, 557–572. 43. Lobb, A. Consume us , isk and ood sa e y: A e iew. Food Econ. Ac a Ag ic. Scand. Sec . C 2005, 2, 3–12. 44. EU Poul y Ma ke s ‘Feeling he Squeeze due o A ian Flu’. Ag aEu ope, 3 Ma ch 2006. 45. EU Poul y S ocks Bu geoning as Demand Falls. Ag aEu ope, 28 Ap il 2006. 46. Talous u kimus. In e ne Panel. A ailable online: h p://www. alous u kimus. i/in-english/ p oduc s_se ices/in e ne _panel/ (accessed on 28 Augus 2013). 47. Isoniemi, M.; Paananen, J. Laadullinen u kimus siipika janlihan käy ös ä ko i alouksissa ja uokapal eluissa (Quali a i e S udy on he Use o Poul y Mea in Households and Food Se ices); Kulu aja u kimuskeskus: Helsinki, Finland, 2008; Työselos ei a ja esi elmiä (Repo s and P esen a ions) 113, in Finnish. 48. Roininen, K.; Tuo ila, H.; Zands a, E.H.; de G aa , C.; Vehkalah i, K.; S ubeni sky, K.; Mela, D.J. Di e ences in heal h and as e a i udes and epo ed beha iou among Finnish, Du ch and B i ish consume s: A c oss-na ional alida ion o he Heal h and Tas e A i ude Scales (HTAS). Appe i e 2001, 37, 33–45. 49. Lawson, L.; La sen, A.; Pede sen, S.M.; Gylling, M. Pe cep ions o gene ically modi ied c ops among Danish a me s. Food Econ. Ac a Ag ic. Scand. Sec . C 2009, 6, 99–118. 50. Ve legh, P.; S eenkamp, J.-B. A e iew and me a-analysis o coun y-o -o igin esea ch. J. Econ. Psychol. 1999, 20, 521–546. 51. Lobb, A.; Mazzocchi, M. Domes ically p oduced ood: Consume pe cep ions o o igin, sa e y and he issue o us . Food Econ. Ac a Ag ic. Scand. Sec . C 2007, 4, 3–12. 52. Gli sch, K. Consume pe cep ions o esh mea quali y: C oss-na ional compa ison. B . Food J. 2000, 102, 177–194. 53. Be nués, A.; Olaizola, A.; Co co an, K. Ex insic a ibu es o ed mea as indica o s o quali y in Eu ope: An applica ion o ma ke segmen a ion. Food Qual. P e e . 2003, 14, 265–276. 54. K ys allis, A.; A ani oyannis, I. In es iga ing he concep o mea quali y om he consume s’ pe spec i e: The case o G eece. Mea Sci. 2006, 72, 164–176. 55. Hosme , D.W.; Lemeshow, S. Applied Logis ic Reg ession, 2nd ed.; John Wiley & Sons, Inc.: New Yo k, NY, USA, 2000. In . J. En i on. Res. Public Heal h 2013, 10 4943 56. Wedel, M.; DeSa bo, W.S. A Re iew o Recen De elopmen s in La en Class Reg ession Models. In Ad anced Me hods o Ma ke ing Resea ch; Bagozzi, R.P., Ed.; Blackwell Publishe s: Camb idge, UK, 1994; pp. 352–388. 57. Ve mun , J.K.; Magidson, J. Technical Guide o La en GOLD 4.0: Basic and Ad anced; S a is ical Inno a ions Inc.: Belmon , MA, USA, 2005. 58. G ande, J.; Bjø ns ad, E.; Wilson, M.; Hanley, N. Assessmen o Consume Risk A i udes and Beha iou Rela ed o Coun e measu es and Radioac i e Con amina ion o Food; No d-T øndelag College: S einkje , No way, 1999. A ailable online: www.sbes.s i .ac.uk/ esea ch/p ojec s/cese / Consume _ epo .pd (accessed on 28 Augus 2013). 59. Slo ic, P. Pe cep ion o isk. Science 1987, 236, 280–285. 60. Sieg is , M.; Kelle , C.; Kie s, H.A.L. Lay people’s pe cep ion o ood haza ds: Compa ing agg ega ed da a and indi idual da a. Appe i e 2006, 47, 324–332. 61. Kahneman, D.; T e sky, A. Choices, alues, and ames. Am. Psychol. 1984, 39, 341–350. 62. Fulham, E.; Mullan, B. Hygienic ood handling beha io s: A emp ing o b idge he in en ion- beha io gap using aspec s om empo al sel - egula ion heo y. J. Food P o . 2011, 74, 925–932 63. Ca ing on, M.J.; Ne ille, B.A.; Whi well, G.J. Los in ansla ion: Explo ing he e hical consume in en ion-beha io gap. J. Bus. Res. 2012. A ailable online: h p://dx.doi.o g/10.1016/ j.jbus es.2012.09.022 (accessed on 10 Oc obe 2013). 64. Klinke, A.; Renn, O. A new app oach o isk e alua ion and managemen : Risk-based, p ecau ion- based and discou se-based managemen . Risk Anal. 2002, 22, 1071–1994. © 2013 by he au ho s; licensee MDPI, Basel, Swi ze land. This a icle is an open access a icle dis ibu ed unde he e ms and condi ions o he C ea i e Commons A ibu ion license (h p://c ea i ecommons.o g/licenses/by/3.0/).