USF Libraries
USF Digital Collections

Public images about chlorine storage hazards following an accidental gas release


Material Information

Public images about chlorine storage hazards following an accidental gas release final report
Series Title:
Quick response research report ;
Physical Description:
1 online resource (16 p.) : ;
Cvetkovich, George
University of Colorado, Boulder -- Natural Hazards Research and Applications Information Center
Natural Hazards Research and Applications Information Center, University of Colorado
Place of Publication:
Boulder, Colo
Publication Date:


Subjects / Keywords:
Chlorine -- Storage -- Public opinion   ( lcsh )
Gases, Asphyxiating and poisonous -- Accidents -- Psychological aspects   ( lcsh )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Includes bibliographical references (p. 16).
Sponsored in part by the Quick Response Research Program.
Statement of Responsibility:
George Cvetkovich.
General Note:
Description based on print version record.
General Note:
"May 1988."

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 - 002021412
oclc - 428110151
usfldc doi - F57-00070
usfldc handle - f57.70
System ID:

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 nam 2200313Ia 4500
controlfield tag 001 002021412
005 20090928141643.0
006 m d
007 cr bn|||||||||
008 090728s1988 cou sb s000 0 eng d
datafield ind1 8 ind2 024
subfield code a F57-00070
RA1247.C5 (ONLINE)
1 100
Cvetkovich, George.
0 245
Public images about chlorine storage hazards following an accidental gas release
h [electronic resource] :
b final report /
George Cvetkovich.
Boulder, Colo. :
Natural Hazards Research and Applications Information Center, University of Colorado,
1 online resource (16 p.)
Quick response research report ;
v #22
Description based on print version record.
"May 1988."
Includes bibliographical references (p. 16).
Sponsored in part by the Quick Response Research Program.
x Storage
Public opinion.
Gases, Asphyxiating and poisonous
Psychological aspects.
2 710
University of Colorado, Boulder.
Natural Hazards Research and Applications Information Center.
t Natural Hazards Center Collection
4 856


rMACiIiS A8OT STO"AcaC. FoLLow. "t,;\. of IMlitu.k of Be..'" a. \J ioV'a \ S c ie.V\ce J tJ .... 0... \ H 0.. 0..'< d..s ""R. e.. ... eo..Y" c. h OJ'/\ -An 1\ c. 0.. OI\S V\{o r \'Y'I 0-. \ 01'\ Ce vl.:b:r 'Rl5pO\lS" =If-19B8


Natural Hazards Research and Applications Information Center Campus Box 482 University of Colorado Boulder, Colorado 80309-0482 PUBLIC IMAGES ABOUT CHLORINE STORAGE HAZARDS FOLLOWING AN ACCIDENTAL GAS RELEASE George Cvetkovich 1988 Quick Response Research Report #22 HOLLSr2 azard COPY __ __ -torn Institute of Behavioral Science #6 (303) 492.6818


* This search mation 0482. coding, report. PUBLIC IMAGES ABOUT CHLORINE STORAGE HAZARDS FOLLOWING AN ACCIDENTAL GAS RELEASE Final Report* MAY, 1988 George Cvetkovich Western Institute for Social and Organizational Research Department of Psychology Western Washington University Bellingham, WA 98225 (206) 676-3544 research was partially supported by the Quick Response Re Program; Natural Hazards Research and Applications Infar Center; University of Colorado; Colorado 80309Thanks are due to Frances A. Cvetkovich for her help in managing and analysizing the data presented in this


PUBLIC IMAGES ABOUT CHLORINE STORAGE HAZARDS FOLLOWING AN ACCIDENTAL GAS RELEASE Table of Contents ABSTRACT" .. \1 D II .. D Il II .. a " ........ a ... II. .. '" .. II .... .......... ........................ 1 1.. I I\ITRODUCT I Q .............. II .... a .. II It a .... II It ........ at WI .. .................................. 2 1.1. The concept of hazard image .......................... 2 1.2. Images, Judgments and ....................... 3 l 3.. Haz i.-\r-cj i mag es ........ It .. ........ II .... II ........................ a ................ II .... 3 1.4. Hazard image changes and the signal value of indL\strial .. '\cc:idfmts ............................. 4 2. IMAGES AND JUDGMENTS OF CHLORINE STORAGE HAZARD: THE PRE-RELEASE STUDy ...................... ............ 5 3. IMAGES AND JUDGMENTS OF CHLORINE STORAGE HAZARD: THE POST-RELEASE STUDy .................................. 6 3.1. The accidental gas release and its aftermath .......... 6 3 .. 2.. Su r v e y sam p 1 e.. .. .. .. .. .. .. .. II D .. a .. .. .. .. .. u .. .. ;a .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. II 7 3.3. Methods and measures .................................. 8 4. RESULTS OF THE POST-RELEASE STUDY ....... u .................. 111.8 4.1. Availability of images ................................ 9 4.2. Issue saliency and image characteristics .............. 9 4.3. Relationship of experiences to issue saliency and i m",'g&? c:h

PU8L I C I MAGES ABOUT CHI_OR I NE STORAGE Hf.' Z ARCS FOLLOWING AN ACCIDENTAL GAS RELEASE This study is ';:Iuidec! by two r-!=seC::!I'-ch qU2'::;"\:ions: changes, if occur in the characteristics of hazard images as a result of an industrial accident? 2) How are characteristics of h2.ZE.l.I'''d r-elai:.=:?d to jLldlJments .",.bout thf2 h,,\z.0.rd? In i ... to the second question, examination of citizen responses to a local chlorine storage facility following an accidental gas release in dicate. that image vividrH?ss i:::; by pm::;onal e:::'


1.2. Judgments actions Affect-laden images have been shown to be predictors of voting for particular political candidates (Campbell! Converse, Miller & Stokes, 1960) and discrimination and prejudicial behaviors 1981). With regard tn other images have been conceptualized as consisting only of cognitive t'-epn-:=S(:2fi tat i cms. But i 1"1 th i mage'3 ,:,.,1 so have been shown to have behavioral consequences. Extensive research in environ mE?ntal ,3,nd h<3,S h(J'I'J P(:t-sonal e:-(perience such as frequency of use influences representations of spatial relationships in cognitive maps (e.g., Golledge, 1986) and the affective and behavioral consequences of these images. Similarly, urban planners following the seminal work of Kevin Lynch (1960), have talked about the urban features that make the images of particular cityscapes memorable and the ar8as v.:\r-ied initi,;:'\l for'

a1. found that the number of concrete elements in an image was predictive of political behaviors. The images of anti-nuclear activists contained more concrete elements than did the images of norl-"';'.ctivists. the public hold':::; ["-e.;:l.dily .:'.:l.vail:;I.ble (abstract) images of nuclear' and experiences a negative emotional response to the issue of nuclear Fiske (1987) arques that most people are not moved to take political action because they lack appropriate concrete images. Utilizing a methodology similar to that of Fisk2, et al. Cvetkovich and Earle (1988) investigated citizen images of a p r' op o':;;ied ten: i c 1,'j,3.":; t inc i r112r" i em p 1 ::.U\ CClfV': ,." en '::::!':::;S clf i ma.(7 ) e was found to be indirectly related to political activity. Individuals holding concrete hazard images took the action of at tending citizen meetings about the incineration plan only if their-im.:1';J'::!'::; did not ha\/e d. l-ela.tively high nU.mtJ'2r of abtl .. .::o.ct image elements. This particular pattern of r-elationships might bp accounted for by the newness of the issue of toxic incinera tion in the the United States in general and the studied com munity in par-ticular. Unlike public discussion of the nuclear holocaust issue, public discussion of toxic waste incineration has only. begun. Continued discussion is likely to pr-oduced in creased availability of concrete image elements and a decrease of abstract elements. Abstract image elements may inhibit the taking of activity concerning the hazar-d. Thus, in the future, if public discussion of toxic waste incineration continues, the pattern of relationships between image vividness and behavior should become similarto that found for the nuclear holocaust issue by Pratto, & i.e., will be directly and positively related to ticm. image concreteness hazar-d-related ac1.4. Hazard image changes and the signal value of accidents Given the apparent relationship of images to judgments and actions related to hazard adaptation it is important to learn tlOW images operate and change in a variety of circumstances. A particular-Iy important question relative to emergency r2sponse pr-eparation and public communication is hON do images and con comitant judgments change following an industrial accident. One possiblity is that accidents have a direct, straightforward ef fects on images. If a person directly experiences an accident! or receives a graphic description of it, his/her image changes by directly incorporating concrete information fr-om the experience. Thus, this line of arguement predicts that hazard images willbecome mar vivid following an industrial accident. This! is probably too simple of an expectation in that it assumes too passive of a role for the information receiver. D.lt8rnative b,:;\si;:?d Clfl the that active interpretation of information about an accident will occur-. The major key points of this expectation are 11,1 I ffi'::1Q t:.:?":; :3.t-l;.J Dr" g c\f"! i ;: i?tj \'.ji:.<. ,/'5 generated utilizing existing of thinking about a hazard). (nent.;"'.l Inoj'::01::; 2. New elements (e.g. ,information from an accident) will be in-


corporated into an image to the extent that they fit existing ment,::"l model '':;. 3. Mental models about industrial facilities may in clude not only thinking about how the facility operates but also thinking about management and safety both within the plant and in the surrounding community. this leads to the conclusion that the signal value 1987) of an accident, that is, its usefulness in portending future is affected by more than the concrete sensory information received. The supposed the accident, the nature of responses to the accident, the judged effectivenoss of hazard management, moves to reduce the 1 i kel i hood o'f -f Lttur'3 acc i den t s <::,md ot her' II con t 1/ ii-If ot-mat ion will play a part in determining the judged significance of the accident and its influence on generated hazard images. This supposition leads to the expectation ttlat under some circumstances the experience of an industrial accident will not increas9 the vividness of hazard images. Such is expected to be the case when people's mental models suggest that the processes to a hazard are well known and that the hazard is being managed so as to (ninimi2e tl'i8 potentii::i.l for (Slovic, 1987). these circumstances an accident will assume a low value as a signal of future problems and will not induce more vivid hazard i ma.ges. 2. IMAGES AND JUDGMENTS OF CHLORINE STORAGE HAZARD: THE PRE-RELEASE STUDY The Cvetkovich and (1988) telephone of toxic i.ncinet'-.:.-..tion ha:-.:ard imc!,ges I'la,s cCJIlducted in Nc:r',/em!:Jet1986 at the height of public controversy over this issue. An existing chlorine storage facility was selected as a comparison hazard. The chlorine facility, part of a pulp-processing plant operated by the Georgia Pacific Corporation, has been located near the main business district of Bellingham! WA for nearly forty years. It made an interesting comparison to the toxic waste incineration p 1 21,n a) it wz;\'s not the -f DCUS o'f wi mad i a at t(::?ntion::'1t th.:::! time that the to:dc 1,_).O,1,':;te incirH?f-i:.1tion pL:Hi I'I),:':;'.S being debated; and b) it did not inspire public Dr political .::;'.c1:.:. i '.,/i t \;/ \I Study participants consisted of a sample of 207 individuals t-'andOI'iil':/ .,cr om Bf211inqrl'::Hll! t.elepl'lc.m.::= .. This number represents eighty-two percent of those contacted. People who were were asked questions first about the toxic waste incineration plans and then about ttle chlorine plant. In both cas8s? only who reported that they were aware of the activity were used in the analysGs. Comparisons of the obtain0d and judgments for the two hazards show that respondents spent more time thinking about the toxic waste incineration plan than about the facility. Images of waste incineration were more readily available and less abstract than were images of the chlolrirl8 'fi:Jcility. Thet-f:.' no -found in COII-


dents were more concerned expressed less trust. that it about the incineration would be safely managed. pl.::3.n 3. IMAGES AND JUDGMENTS OF CHLORINE STORAGE HAZARDI THE POST-RELEASE STUDY 3.1. The accidental gas and its aftermath and On the afternoon of October 19, 1987. almost one year after the completion of the above described study, overheated chlorine gas ate its way through a pipe in a huating system, causing g d i sch :!.Ir (J >!: wf f el'" ric c h 1 (Jr" i oj I:;:; from t. h.:::: GC?Olr'q i a p.;;) .:: if i c: Dt-' i::\ t. ion facility. Acting under the assumption that th.:::: large yellowish cloud slowly being blown througt, the main downtown area might contain deadly chlorine. emergency officials cordoned off the area and executed a partial evacuation. Several individuals were hospitalized as a result of lung and throat irritations from exposure to the chloride. Since the affected geographic area, which is approximately at sea-level, is surrounded by higher-elevation areas, the cloud was clearly visible from many sections of th.:::: city. During the period of the cloud's movement the town, verbal descriptions of it and its possible dangers w.::::re broadcast on local Visual representations of the release in the form of photos, :;chema.t i C';5 c\f1cl map':; CloS 'tH?;! 11 <:1.S descr i p t i ons were al':io pr"e':sented in 10c.3.1 and in coverage by both local and regional stations. A large propor tion of the local population saw the cloud directly or wer8 ex posed. to one or more graphic and/or verbal descriptions of it either-'.:il11e of thr2 1'",;?li2<:3.'::38 or within 24 hou.r"s 'lion In addition to being of the gas release e:-:posed t:o pu.blic the actuality or a descripwas also exposed to informa-tion about the causes of the responses to it, evaluation of .3.nd fu.tur"l:;:; S[;l.tE:-:ty i0.illj' fclJ.:::i 1 i ty ma.n3.(;)emE?nt, .::l.nd adequacy of emergency There were several sources of this information reported in the media. statements about events progressing during the release and the decision to evacuate were mi;o.d e by t.I'-, e d i t"':::.:c tor" .:;1.1"1 cl (Jt h 1 DC:.;:11 c '/ '::i vic: er" :';;:CII"lnel. The basic message given by these individuals concerned the difficult_'! of making decisiDi'j';::; qi\ien un.::el"-" tainities such as the cause and composition of the release but! th.:at I'"e':;pclnsi:::;: SE'em(2rj tCl bl;? .:lPPt"ClpY-i.,=-,ti2 tel the hazard. Another chief source of information was the manager Clf the c:hlot"ine plant. Hi:.;:, delivel"er:j in .,-:\ combative manner in the weeks following the suggested th.::.\t not ovel'--Ire;;:\ct tCl thl:;? r-"aleasE!. The presented was that the community was not 'placed in grave danger by the release (chloride, not chlorine. had Y-ele':"',5 i2-!;j), thc.,t the r"isk':::; 0+ U-jl:? ':3itirlg of .::1. ct1"lcj't:-in:::? pl.;::H1t: the business is justified since chlorine is found in many clt.h('2tp 1 .:..ces b 8':; i d E!S the i: '::;'.c iIi t '/ 3.rl cl t: h iJ.t. C h 1 Clt-i f1 P rov i d:::;';;; so many bc!!":e-fit" (,:?';:J. "Chlc:wirle i..:; a b.:;l::;;ic bu.ilding block (::J-F 0 c i Y e:i i1 ;j it' S i:J 0 i. II g t c:J b e 'iJ i t h u:;. II II ( I t ) .::; CJ b:l. Dod '/ u ';:; ':2 t U. 1 it.,;;:, !'"(::!ally ,,-\11 tI1(;2 cCJrnmurlit,y'.'1 1.':0.,) \'I.:;\tC?I'-!:. r-t. tTl (:? fl t:. Ll + t i n q CJ + P 1 .::; t. i C:5) t .1'"-k 1 :.1). d .. -Ie::;::; thi.::; l.'J.3::0 nu"!:: lj\i :;.;umc::! cor')"' '-;


cerned about future safety of the plant. These concerns seemed to be supported by chlorirl8 f2cility workers who publicly complained about how they were notified about the release and evacuated from the plant (i.e., into the potential path of the gas cloud) and that safety procedures were not being enforced and followed. As a result of the release, and as part of Superfund Amendments and Reauthorization Act Title III relatGd activities, appraisals of emergency response plans and changes in the community warning system did take place. There was no organized citizen effort expressing apprehension about the facility and overt public expressions of concern r'apidly diminished. 3.2. Survey sample The post-release survey of chlorine storage hazard images was conducted from 17 to 19 November 1987, one month after the accidental gas release. The sample studied cqnsisted of two ';Jt .. ClUP:=;: 1) 1.')1'10 1-10.lj p21.1-ti.cip,:.,ted in "I.:I-Ie first study and who had agreed at both times to be reinterviewed and 2) Two hundred and twenty individuals who were contacted for the first time to participate in the post-release study. As was true of the pre-release surveYj this secClnd group was randomly selected from the local telephone directory. The response rates for the two sub-groups are 78% and respectively. Characteristics of the two subgroups of the post-release sample in comparison to the pre-release sample are shown in Table 1. The two total samples are fairly comparable with the exception that the post-release sample contains fewer long time residents. The reinterview sub-group is for the most part similar to the two total except that it contains somewhat fewer new resi and 26 to 39 year oldsj but more 40 to 45 year aIds. Age L.E!nqth 1'-e':; i d Grl C G Education Table 1. Descriptive characteristics of samples. m.;;).l e f em.o\! e 18 to 25 26 to 39 F' tG., .-. ,.-1:2 1 e ::.:;ampl E? (n ::: 2(7) 44% 24. 40 to 65 66 or 01 3c)" 3 yrs or less 26.6% 26 yrs or more 31.01. Hi (Jh ';E;dI0Cll atless <'")C.") '"':" .. ..:....w'" ...:.../"" Post-release sample re-interview total sample (n :::: 87) l 0'::';8 21 a 36.4% 16. n': 39.4% (n =: 3(5) 56% 21 .. 34. 29. 1.1. 15. 1 22. The expectation that 0 large proportion of the population was exposed either directly to the gas release or" information about it is supported bv the results presented in 2. OVi.'::-t-<:1. U'l:Lr-;j uf the:-t:ot .::,1 s.,)iliplE.-h"'.d p>Jr-cma.lly '0;(::::8n tr-ie


gas cloud and nearly everyone had received information from both formal and informal sources. Seven percent of the sample had been evacuated and nearly eleven percent reported having per sonally experienced some physical effect of the release such as burning eyes or throat. Table cloud

ship of experiences to issue saliency and image characteristics; 4) Relationship of image characteristics to hazard judgments; 5) Changes in hazard image characteristics following the gas the e:-:ception o-F '::;,::?ction ..jhich reports on tho: re-interview sub-group, the reported results are for the total post-release sample. 4.1. Availability of images The amount of timG taken to begin the hazard image report varied from almost immediately after the prompt question was asked to 25 seconds later. The mean time was 3.3 seconds. This same measure of image availability failed to show a relationship to hazard-related judgments or actions in either the Fiske et ala or the Cvetkovich and Earle surveys. One reason for this may be that some people are reporting as soon as the first image element occurs to them while others are waiting until a more complete image is formed. When asked which was true for them, seventy-one o'f thf? '::;ai d t.ha,t th,=y had l'-'2pcwted i 01mediately and twenty five percent said that they had waited for a complete image. These differences in reporting style were significantly correlated to availability measures p=.OOOl). Th is i nd i tt1a,t f utut'e ri2se,:Zl.r"ch ':5hDul d e'f f or-,ts tel tl,a t C':in'::.; i. st:.ancy i n t. i ng i s i. rlr.;) Beca,use of th is d i 'Fference, this measure is not given further cDnsideration in the n:=maining resl_llts. 1\10 sir,;)ni{ic:cl.fYI.:. con'-eli:;\tions t'i0.'re found bc:!tween length of elapsed time before image repDrt and the image charac teristi.cs. Nor were any significant CDr-relations found between time of report and image characteristics, except that people who waited for a complete image to form tended tD give more passive emotions (r = -.11; P = .05). 4.2. Issue seal iency a.nd image c:ha.ra.cteristics Saliency of the chlorine facility was measured by the ratings of how much the facility had been thought about during the month on a from 1, "not at all" tD 10, "a. preclCCUp2\tion." individua,ls il,dic.:\ted th,::at they had not thought about the plant and 24 indicated that it had been a major preoccupation. The mean rating of saliency was 4.28 (8.D.=2.51). Tab 1 e 3 sholto)s abstract elements and images. The reported the frequency of reported concrete active and passive emotional reactions images can be characterized as ,i:1.nd to


--------------------------------------------------Table 3. Frequency oF concrete and abstract elements and active and passive emotions. Number of responses none one two three four five Percent of total sample en = 305) -----------------------------------------------Abstract elements 15. 8 36. 4 28. 6 14. 5 7 1 0 Concrete elements 57. 9 13. 0 u 0 O. 0 O. 0 Active emotions 68. 7 29. 1 k O. 0 o. 0 O. 0 Passive emotions 6. 0 I 56. 7 2 8. 0 1 c u O. 0 predominantly consisting of abstract elements evoking passive emotions. A mean of 4.59 abstract elements, 0.62 concrete elements, 0.34 active emotions and 1.48 passive emotions were given per individual. In addition to the above categories hazard images were coded according to their specific content. The coding were whether the images 1) actions as evacuation, escape and seeking shelter, 2) consequences such as panic, illness or effects on safety, 3) the seeking information such as listening to the radio, 4) characteristics of a gas release such as descriptions of a release cloud or explosions, 5) illicited nlemories of past human tragedies such as the qas release at Bhopal, India or the Vietnam War and 6) seeking shelter by remaining in a building and sealing it to prevent gas from enter ing. A frequency count of these themes is presented in Table 4. Total number of image elements is clearly dominated by the theme of consequences, f9110wed in frequency by the theme of actions initiated in response to the gas release. The themes of information seeking, physical characteristics of the release, induced memories of other hazards and staying sheltered are relatively infrequent.


'. Table 4. Fr8quencyof reported image themes (n = Number of One Two Three Four Five Total Action 74-15 ":!' '-.' 0 0 <) 92 Th'?in8 Info Canse-Character Memory Stay! seek quence of release seal 12 109 9 15 ''':!' 71 4 "'Q '-' .::. .:.. 0 4 1 0 1 "":!' .L. ,_, 0 11 <) 0 1 0 5 0 <) <) (l 1 0 0 <) 15 254 2 1 1 71 An important aspect of a hazard image is whether it contains ell.::!merl't::; rela.ted to beh'::l.vior-. As can b;2 seen from Table 4 about 30% of the image elements included the theme of evacuation. Even fewer, about included the theme of stayinc in a building and sealing it against the gas. These relatively law frequencies plus that only about 5% of the image elements included the seeking of additional information indicates that spontaneous images for this hazard do not include optimal strategies for self-protective behavi 4.3. Relationship of experiences to issue saliency and c:harac:teristics Pearson product moment correlations between reported personal related to the gas release and image saliency and characteristics are given in Table 5. Saliency of the gas release hazard was significantly related to all of the measured reports of personal experience. The hazardous facility was thought about frequently by peopla who had seen the cloud, who had been evacuated and who had physical effects from the gas. It was also thought about frequently by those who had been frequently exposed to messages from the media and acquaint.:!.rlce::; that e:.: hi I;)h Relatively few of the image characteristics were related to personal experiences. People exposed to higher levels of expressed concern in the messages of both media and friends 1'-f2porti2d a hi (;]her nLlmbel'-' o'f abstr-act i maJJ2 el ernent':;;. Concenl expressed in friends' messages was also correlated with a higher frequency of expressed active emotions as was the experience of physical effects of the gas release. lJ.


-------------------------------------------------e c::....J. i ellC';/ Correlations between and image characteristics pel-.;5 0 r <:'1. 1 Image characteristic --------------------------------------------E:-:perience Saliency Abstract Concrete Active ---------------------------------------------DIRECT 8'.:e cloud (l==Y; 2=N) Evacuated ( 1 ='(; 2::::1\!) E:(peri phy. e-Fff2cts ( 1 ==Y; 2==N) INDIRECT of -14"* 15)!media exposure .13* Impact of medic., conc.;9=v.conc).13* Freq. ti:.".l k t:o f r i Impact talk to ft-iends p .01 .** P .0001 02 05 06 0:;:; <)2 .08 .00 .00 16* --.05 n06 -.04 17* .00 -04 02 -09 n 05 -12-*..... 01 .04 .. 00 .01 .06 .04 A ():2 17-)1---.08 Saliency was significantly related to two image characteris tics. Both number of abstract elements and number of active emotions were positively correlated to number of reported times in the last month that the hazardous faci.ity was thought about (r's == .13 and .14 respectively, p=.Ol). Number of concrete elements and number of passive emotions were not related to saliency er's --.04 and -.03, respectively). The pattern of results presented in this section suggests t ha t t 12f f ect of oj i r-ec t p er-son a. J. e:< per i C e;:;; orl i rni::l.';l C? C 1'.::'I.r-'::.".c: .. teristics is mediated through saliency. People have to think about a hazard relative to their direct personal experiences to give the experiences meaning and for the experiences to have an effect on generated images. Messages from others such 2S fr-iends and the media seem to have a more direct effect on image charac-ter-ist.ic',3" F",-:,-hiJ.p'-:i this i'3 fr-om other-',;:; is "pr-E0pr-occ:?':;':::;ed" in already formed conclusions about because the tha.t t h "-'It"". d c::orlt.3.i n 4.4. Relationship of image characteristics to hazard judgments B C) t h r p C) rt rJ n u m t.1 e r-(.Jf iJ. b 'oS t rc t m n t -:; ,:1. r1 \j rP D r-t. '''" d p -0 .. :;sive emotions were correlated to judgments about possible futur8 y''::;''s (T"",blE' 6). n hilJher-I"lumi:)(:.?r-


elements and a higt18r number of active emotions were both associated with high estimations of the likelihood that: (1) a chlorine release will occur in the future; (2) injuries will oc cur because of the release; and (3) the respondent will be personally injured by the release. Reported number of abstract elements and reported number of active emotions were also associated with high levels of concern about the facility, low acceptance of the facility, and low trust in management and operation of the plant. Saliency, number of concrete elements and number of active emotions were not related to hazard jUdgments. Table 6. Correlations between saliency, image characteristics and hazard judgments. Judgments Image character. Est.of Est.of release some injury Est.of self injure Accept Concern Trust Saliency Abstract Concrete Active Passive p .05 ** P .01 --03 10* 03 16** 08 -03 02 08 12* -07 03 13* 26** -02 02 01 02 03 1 1* 10* 17** 02 02 06 21** 17** 21** 04 06 01 ChAnges release in hazard image characteristics following the gas The accidental gas release and subsequent direct or indirect experience with it clearly increased the public's awareness of the chlorine production and storage facility_ Seventy-seven percent of the total post-release sample spontaneously mentioned the plant as one of the major technological hazards in the cuunty, versus 52% in the pre-release sample. Was this increased public awareness accompanied by a change in images and judgments about the facility? In light of the increased power of longitudinal over cross sectional designs in evaluating ctlanges (Baltes & Nesselroade, 1979), Dilly the results from respondants who completed both pre and post-release surveys are used to evaluate the effect of the accidel1tal gas release on image characteristics and Judgments. Pre-and post-release saliency and image characteristic measures are presented in Table 7. Clearly the gas release produced an increase in the saliency of the hazardous facility. Significant changes were also found for all four of the image characteristics. Both the frequency of repor-ted abstract e18ments and passive emotions increased across tile points, whereas there was a significant decrease number of concrete elements and active emotiorls. two measurement in the reported Characteristics


of the hazard images changed following the gas release. images did not become more vivid. Table 7. standard deviations and analysis of variance results for preand post-release measures of saliency and image char i3.ctt:\r i st i cs. t"lea SUI'" f.? Sc.".liency Abstract elements Concrete elements t. i emot:L on'3 Passive emotions 1. 813 <). C?7 1.21 0.93 2. 11 O. 1.27 0.9."5 ()II 59 r::'ost -'r-el !vle,:Hi S. D. 2.50 1 42 1 17 <). 0.80 () 1 () a 5 (; 1..16 0.88 F 40.64 '7 w 12 .. ()Ct 40.02 p .0001 ()()-l .001 .0001 .0001 Significant changes were also found for four of the six measur'ed hazard judgments (see Table 8). There were increases in the estimated chances of a future release, that someone would be injured by the release and that the person making the judgment be injLwed. al.'00 ,"\1"'1 incn2i"\S',e in cor) cern about the facility. However, no change was found in the acceptance of the existance of the facility at its present location n or-I,"D.S t 1'\ i2r" ''2 cl).:':':\I'i g i r') t 1"'1 e tru. st (JlJ t thef iJ.C i 1 it i 85 ina.!") 12ment and operation. Table 8. Means, standard deviations and results for preand post-release measures hazard ']U,jr.;;lil1l:2I")t Est. of 1'-021 e, Est. of some injury Est. of self injury Conc::ern Tt" LISt c: c: e p t <.:Ui C ''2 0 ::-:::cy 1 42 1 1 1 44 -:!' 83 Lr".\ '-' L .. i.J. .. r.::'r.:;" ._J...J 0.:1-4 t'" '''7 C\c::, I 7...J 4 67 1 ".., '-' -. .' pCJ::;t-f"el ea:5e t'1\"22.n s. D. 5 "'(::'1 ::;:7 "';"'.J '''\ 49 .::. 4 C:'I-:: I .J .... Ci 84 :21\ 8'7" 5 "" OS ..:, t1-n 34 '-' 05 5. DISCUSSION AND CONCLUSION p 107 1 1 <)7 I::' 2 1 ,.J ll. b 0 4r '.' ..:.. <) 33 000 1 <)<)2 0:3 000 1 n':::; n':; this study are summarized in Figure 1. The results support the expectations that hazard im3ge c: h ,3r 2.C t i::=.'r" i .:; tic: s cl D p r-,:=.'d i c t. h ::''12 ar cI j I._l>:i(;jin,::?n t:; a.n d t It a t t t some aspects of) image vividness is influenced by direct Qnd indirect accident experiences. The study supports the validity of the image concept and suggest its useful. in understanding h a 2:::'. r ,jc,. d",\ P t. 3.t i Cl r-I b h -::;, v i 0 r' T h (;::, s; c ';3 U. cJ (,: .:::; t t? d I'" C 1 0. t i u I' \ S hip 0 + 1. il


course, are based on simple descriptive and correlational analyses. More sophisticated multivariant analysis further ex-ploring these relationships are planned for the near future. DIRECT EXF'EI::;: I ENCE: CDMr'IUN I C{iT IONS FFi: I ENDS/IviED I (,4 SALIENCY HI'.:jZAF;:D _____ HAZAFW II'1(.'iGE >,j UDGr1ENTS Figure 1. Model of effects of experience on image characteristics C1.nlj j udgm(;nts. The results also suggest that hazard image characteristics are not changed directly by the concrete sensory qualities of accident. The signal value of an accident, as reflected in the concreteness of a hazard image and the active emotional response it induces, is greatly colored by interpretations of the accident's context. Thus, existing mental models concerning the processes underlying a hazard and management of the hazard influence the meaning of an accident. If citizens' mental models indicate that there exists a community repertoire for understanding and effectively responding to a hazard, as seems to be the case for the hazard studied here, an accident will not be inter preted as an emergency signal. The chlorine qas release can be described as producing a state of vigilance. As reflected in hazard judgments, citizens became more concerned and increased estimates of statistical risk. The .accident did not produce an increase in image vividness, nor did concerted efforts to change the I'"isikinf:?s's elf tt-Ie ClCCUX". One implication of the si:udy is that it indicates some limits concerning efforts to get cit_ i z c ':lnCer" necj ,"Hld" i fl VO 1 ved" I,.l i th en vi r" onmen tal .i:).nd h (;)';'-1.1 ti, hazards through the use of depicted images. These efforts will not. be sl_tcce-:'isfLI.I if they fail to ( ttlinking about the hazard. 15


P. B. & Nesselroade, J. of lcmgitudinal .. F.:. ( 1979) In ,J. H. History and rationale Nesselroade and P. B. D a 1 t e s ( Ed s.) .. ... t. h cJ \'.. .. Beach, L. R. & Mitchell, T. P (1988). Image theory: Principles, 1;1 0:;1. 1 n d pIa. [1 i !"'j d c: i:; ion iii .<. kin (;J .. i n .. -:?s!:, Ca.mpbell, P,., P .. E. '! \!L E .. ., St.c)k'!:-:s., E. ( 1 Th(:! ca.!". \,1ot.i2r". Ne'rJ Yew k: \ .. J i. 1 Cvetkov:lch, G., E"-\I'''l.e, T. C. (F7'8!]). H,:I.z,?r"d imaqes,! and political activity -the case of tuxic waste tion with implications for hazard communication. evl i (.:l[1'", incinera. .... In R. E. 1< a per s Ol"j n cl F'. S t. ,:,1.1 1 fD n ( [7.: d <;Oi.) .. c .. 9.. H'72.3.1 t.h D.nd g .. r.:;: i t c) for" thcLlrni ng. EI'" i c ssm"" {-\. S i mOil, H. (1 (;84). .. .. i.':; ... .. .:3. s D a. t a C in lJ r' i d ';1 C', 1"'1 A :: Ivl I T F'r' E":; ':::; Fiske, (1987). People's reactions to nuclear war: Implica.-t ion':=; for p s y c h 0 1 C.) q :L ':; t .::.; .. .. LJ:2, :207 Fiske, c w .. i en,:::,u,)es Social T., r .... '::( 1"1. PI. (198:3). CitL:en':S' of nuclear war: Content and consequences. Journal of 3(7" 1., 41 _.. 65. f\. G. (1986) .. co(;,)nii.:ion. In D. I A 1 t en a n ( Ed s.) r", d .. .. .. 0 q.i.' N e I',J Yen-k: v.Jiley. 1 tem D. L. (1981). .. .. 8...c .. .. :!:....s.C."EE-.:tLp..2: .. n.9. D.n d Hillsdale, N.J.: Erlbaum. Lynch, 1<. (19,!::'O). The o-F t.he C'::\I11bl'''idi.j8, 1'1::\'33: 11. I T. F'r s'ss. Ivla.r k'_I.s H., I\lur ius, P. ( 19B6) Possible selves. e::Ebo 1 ::.;t. 4 9 954-969. Nisbett,., fi:. E .. L... (:Lr.?t30). .. .. of .. .. Pr't-?nt i ce-Ha.ll P. (1987). Perc8ption of risk. Science. Stark, J. (1987). G-P official clownplays chlorine risks. q Ii 1\1 0 V (2 ill b ;;2 r 3 1 9 C 7 <:"':::: ,_.1 "HW' 113 L Heuristics and biases .. N, ,J .. De 11 i r .. .. J.. . .1