Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
ITT. ANALYSIS OF THE RESPONDENT DATA As indicated in the section on sampling methodology (Appen- dix B) 107 residents were chosen at random from the current city directory in an attempt to get a representative sample of the adult city population. A separate (saturation) sample was selected in the flooded area for the purpose of testing hypotheses which might be influenced by proximity anchor exposure to threat. In the section which follows, our ~ ~ . r objective Is to give a picture ot respondent characteristics and reac- tions in the city as a whole. Therefore, the saturation sample will not be considered at this point, since its inclusion would bias the data geographically. After the interviews were completed, maps of the city were consulted and respondents in the city-wicle sample were divided into two groups by location: (1) those who lived in the area which had been inunciatecl during the week before circulation of the false report (N s 53), and (2) those who lived in the non-flooded area (N = 54~. The two groups differed significantly in certain of their re spouse s to the false report, as well as with respect to certain other characteristics. Some of these differences will not be discusser] here. The groups will be considered separately except when discrepancies between them are treated as part of the testing of the formal hypotheses in Section IV of thi s r eport. ·_0 · , I] A. Demographic Characteristic s of the Sample Relatively stable characteristics of the sample will be sum- marized here. A complete tabular presentation of these data appears as Appendix D. Of the 107 respondents, 43 (40. 2 per cent) were male and 64 (59. 8 per cent) female. The median age of respondents was 46. The s ampl e wa s p r e do minantly lo we r - mi dale cla s s, by definitio n of the interviewers. However, the socio-economic status of those respon- dents in the non-flooded area Is generally higher than that of respon- dents in the flooded area; 46. 3 per cent of those in the non-flooded area are upper-middle class. This discrepancy conforms to our expectations from what we know about neighborhoods in the community, and testifies to the representativeness of the sample. Again, the educational level of those in the non-flooded area is higher Man that of residents of the flooded area. , ~_ ~_ , 23
The population of Port Jervis appears to be highly stable. Sixty-five (60. 8 per cent) respondents had lived in the city for more than twenty-one years; no one in the sample had lived in Port Jervis for less than a year. The mean number of people who lived in the same household as the respondent was 2. S. Thirty-one respondents had children under ten years of age living with ~em. The family re- sponsibilitie s of re spondents wer e recorded because there is some indication (5, 12} that the protection and welfare of children is an effective predictor of the activities of adults in disaster situations. The s e data will be examine d whe n we c on s ide r the f or mat hyp othe s e s of the study. B. Disaster Experience of Respondents Before Circulation of False .. . . . Report . The disaster experience of respondents is indicated in Table opposite. A number of Port Jervis residents (15 per cent of the sample) had been exposed to flood disaster in previous years. Many more were directly affected by the 1955 floods; more than half the re- spondents experienced some loss, either to themselves or to close friends and relatives. The sample indicates that almost a quarter of the town's adult population was evacuated at some time during the week of the floods. Those who escaped damage in some cases served as hosts to the people who had been driven from Weir homes by flood water s. Twenty respondents (18. 7 per cent) had disaster-related jobs. Six of these worked under the direction of Civil Defense to restore the city following the flood and 7 worked with the Fire Department. A few of the latter were actually Volunteer Firemen, but most were as sisting temporarily during the emergency. C. The Situation in Port Jervis Before Dissemination of the False Report As indicated previously, the Wallenpaupack Dam seems to have been a long-standing threat in the minds of at least part of the population and rumors of dam-breaks had pervaded Me town on the days following the flood. The rumor came as the town was digging out from the flood and beginning to relax. Forty-five per cent of respon- dents indicated that they had heard rumors about a dam-break during 24
EXPERIENCE Yes No TABLE I DISASTER EXPERIENCE OF RESPONDENTS Percent IN s 107) I - I n ?* Total P devious disaster experience Flood experience Other disasters Total, previ ous disaster experience Damage from 1955 floods Maj or per sonal property lo s s Major loss to friends, family Minor lo s s, own or r elative s Personal los s to business Irons, nature unspecified Total, damage from '55 floods 15.0 13.1 ~ .. 28. 0 68. 2 15.9 23.4 4. 7 1.9 10.3 Be 7 1 00. 0 56. 1 42. 1 1.9 1 00. 0 Evacuated duri ng 1 95 5 floods Host to flood evacuees Befor e the fat s e r eport During the false report Precise time undetermined Total, ho st to flood evacue e s 23~4 6.5 5~6 1~. 9 74e 8 1 e 9 1 00. 0 14. O 85e O Oe 9 1 OOe O 1 1 1..::1 ~ * "?" means 'undetermined'' in this and later tables. 25
that week; others had heard such rumors in years before. respondents tract heard some discussion about the possible consequences If the Wallenpaupack went out. D. Spread of the False Report - Almostall The percent of respondents who heard the false report and those who evacuated are shown in Table 2. Seventy-eight respondents almost 75 per cent, heard the false report on Saturday night before they heard loudspeakers or other sources denying the rumor. Some of those who did not hear the report were out of town; either they had been evacuated, or had left for other reasons. A third of those who heard the false report evacuated subse- quently, while many others prepared to leave but learnedthe report ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Of the entire city sample, was false before their plans were completed. 24. 3 per cent fled. This fact illustrates the heavy impact of the false report; i. e., if the sample is accurate, approximately 2100 adults out of a total population of 9, 000 fled. The official report of the Police Chief provides an empirical check here on the representativeness of TABLE 2 PER CENT OF RESPONDENTS WHO HEARD FALSE REPORT AND PER CENT WHO EVACUATED EXPOSURE TO AND REACTION TO THE FALSE REPORT Per Cent Total Ye s No Total Number ~100.0 1i 75. 7 1 00. 0 66. 7 100.0 78 Heard the false report (per cent of total sample) Evacuated (per cent of total sample) Evacuated (per cent of those who heard false report) 72. 9 24.3 33.3 107 107 26
the sample. This report states: ". . . It is estimated that between 2, 500 and 3, 000 persons and several hundred cars took part in the panic. 'I The addition of the number of children to our figure of 2, 1 00 brings it within the range specified by officials at the disaster scene. Almost 90 per cent of those who evacuated were individuals living in the area which had been flooded the week before. This fact will be discussed at length when we consider the formal hypotheses of the study. We will now look more closely at the situation and activities of the seventy-eight respondents who heard the false report on Satur- day night. I. Situation of respondent when he heard first threat message: The time respondents first heard the false report is shown in Table 3 below. The false report began to circulate sometime after 10:30 on Saturday night, and it continued to spread until after 1:00 in the morn- ing, though most of the respondents had heard the rumor by midnight. The median time for receipt of the first threat message was 11:30. Note that this provides us with another empirical check of the validity of the sample. As indicated in the description of events, the fire TABL`E 3 TIME RESPONDENT FIRS T HEARD FALSE REP OR T Time Before 11 p.m.... 1 1: 1 5 .............. 1 1 30 ~ e e e ~ ~ e ~ ~ e ~ e 1 1 45 ~ ~ ~ ~ ~ e ~ ~ e e 1 2 00 ~ e 1 2 1 5 ~ e e ~ ~ ~ ~ ~ ~ ~ e After 12 1 5 ~ e ~ ~ ~ e ~ ~ ~ Undete r mined . . ~ e ~ ~ ~ e ~ e e e Total P ercent (N = 78):k 1 Be 4 10e 3 26e 9 1 Be 4 23e 1 2 ~ 6 Be 8 2~6 1 00. 0 Percentages here refer only to the group of 78 respon- clents who heard the false report. 27
truck dissemination of the rumor took place for a period of minutes around 11:15 that night. Shortly after this official statement of the thr eat, the bulk of re spondents received it. The lo cation of r e spondent s i s shown in T able 4 below. Since the false report circulated late in the evening, it is not surprising that three-quarters of the respondents who heard it were at home at the time. It may seem strange that no respondent in the sample was at any place of entertainment such as a bar or movie when he first heard the false report. However, we know from key interviews that in one bar and in the only movie, the bartender and the manager, respectively, did not transmit the message to people who were there. Respondents activities at the time the false report was heard is indicated in Table 5 below. TABLE 4 LOCATION OF RESPONDENT WHEN HE HEARD FALSE REPORT Location - At ho me e ~ ~ ~ ~ ~ ~ ~ ~ ~ e ~ ~ ~ ~ ~ e ~ e e ~ ~ e ~ ~ e e Friends, relative s, or neighbor s' house On the street (walking or driving) . e e e Working at regular job eeeeeeee.~e Working at disaster job . eeeeea.~e Other location. ~e ~.e. I~ocation undeterminf`~1 Percent (N = 78) . .. . 75e 6 S. 1 7e 7 2~6 7. 7 Be O 1 ~ 3 1 OOe O A large proportion of respondents were with their families when they received the threat message. We had hoped to be able to estimate the means of communication people use in large scale disas- ters when they are separated from their families. The number so separated in this case, presumably e - lained by the fact that most 28
TABLE 5 ACTIVITY OF RESPONDENT AT TIME HE FIRST HEARD FALSE REPORT Activity In bed, asleep ~ e ~ ~ ~ ~ ~ ~ ~ o. o. e ~ e e ~ ~ ee ~ e e .e e ~ e ~ oe ~ a. ~ e ~ ~ Preparing for bed, or in bed, not asleep..ee.e...e.e..ee Li stening to r adio or televi s ion . . . . . . . e e ~ e ~ ~ ~ ~ ~ ~ ~ e ~ ~ e ~ e ~ e Talking, interacting with other individual s . . . e e e ~ e ~ e ~ ~ e ~ Other activities e .. ~.~. ~.~. Activities undetermined Total Percent (N = 78) _ 33e 3 7. 7 24.4 1 1 e 5 2 1 ~ 8 le 3 1 00. 0 respondents were at home, was too small to allow an analysis to be made. The location of respondents' families are indicated in Table 6 below. TABLE 6 LOCATION OF RESPONDENT'S FAKD:LY WHEN HE HEARD FALSE REPORT Location With respondent, or respondent was sure his family was safe ..eeeeeeeas Separated from re spondent With means for a quick check ........ Without means for a quick check ....... Percent (N = 78) . . . . . . . . . . 73. 1 14.1 6.4 Undetermined . ~2.6 Total, separated from respondent 23. 1 Loc attain of r e spondent' s family undetermined . . . . . . . . . . 3. Total, respondents who heard false report . 29 ......... 1 00. 0
Of the eighteen respondents (23. 1 per cent of total) who were separated from their families at the time they first heard the false re- port, 11 (61. 1 per cent) reported that they had a means for checking quickly on the safety of their families; five (27. 8 per cent) reported no such means. In nine of the cases (50 per cent) in which the respondent was separated from his family, he made no attempt to communicate with them in the perioc] which preceded his receipt of denial. In six cases (33. 3 per cent) he mover] physically to his family, and in the re- ~aining three cases (17. 5 per cent) telephoner] his family. 2. Source and channel of the false report: a. Exposure to sirens and other noise: As indicates] pre viously, one major source of the rumor was a fire-truck (or possibly more than one fire-truck) which broadcast Me rumor over its loud- speaker and blew its siren. The siren in Matamoras, across the river, may have been heard in Port Pelvis and perceived as a danger signal. These and the noise, confusion, and movement in the streets Inlay have sensitized people to the extent they were expecting bac! news before they actually heard the false report. Forty-two (39. 3 per cent) of the 107 respondents in the sample stated that they heard noise or sirens (separately frosts any threat mes- sages) on the night of the report. (See Table 7) This number probably floes not represent the total of respondents who heard such generalized threat mes sage s, because this question was not specifically asked in the interview. TABLE 7 EXP OS URE T O SIRENS AND O THER NOISE Kind of Exposure Siren or noise as first exposure to threat . Siren or noise after verbal threat message received, but before receipt of denial .. Did not report hearing sirens or noise, but heard verbal threat message Did not report hearing sirens, noise, or verbal threat message 30 Percent (N = 78) .... 24e 3 .......... 1 5. 0 ...... 2 3. 4 1 00. 0
There were, then, eighty-two respondents who heard sirens anc! not s e, ve rbal thr eat me s s age s, or both. Of the s e, s evenly -eight heard a verbal message before receiving denial. Four respondents heard sirens or noise and did not thereafter hear a threat message. In these cases, either the respondent did not attempt to find out the cause of the disturbance or, when he did, was answered with a denial me s sage rather than a threat me s sage. b. Exposure to verbal threat messages: Seventy-eight respondents were exposed to threat messages of this sort. Only the first, seconc! and third messages which they received were recorded; since only seven respondents heard even three threat messages, the cutting point appears to be a reasonable one. A total of 106 threat messages were heard by these respondents. From Table (3 which fol- lows, it c an be s e en that the thr e at me s s ag e s we r e lar g ely tr an s mitte c] through unofficial sources. The pattern of sources for all three threat messages is quite similar. A detailed comparison of these sources with the sources of denialmessages willbe made whenthelatter are pre sensed. 3. Content of the false report: We know the content of ninety- two of the 106 threat me s sage s which the respondents remembered hav- ing received. Eighty-six of these mention that a dam had broken, while the other six were warnings to evacuate without the specific cause being stated. Of the messages whose content we could] determine, 45. 7 per cent included advice to evacuate, either with or without adcli- tional information about a dam having broken. For a detailed] analysis of the content of the threat message, see Appendix D, Table G. A study of an evacuation of Panama City, Florida, (7) con- clucled that more beach area residents left their homes than did town residents because the former received more strongly worded admoni - tions to leave. In Port Jervis, there was no increase in the tendency to flee when the respondent was advised to do so in the threat message he received. This may be explained by the fact that Port Jervis resi- dents had been discussing the consequences of a dam-break for some time before the false report gained currency. Thus the situation may have been predefined as one which would necessitate flight, so that advice on this issue was almost irrelevant. Fifty-one per cent of the persons in the previously flooclec3 area who received threat messages said that the messages included 31
TABLE 8 SOURCE AND CHANNEL OF VERBAL THREAT MESSAGES SOURCE AND CHANNEL Percent (N=78) Receiving Fir st Second Me s s age Me s s age 6.4 5.1 3.8 0.0 6.4 2.6 16. 7 7. 7 15.4 2.6 39.7 6.4 24.4 10.3 79. 5 19.2 3.8 0.0 00.0 Z6.9 _ 178121 ; Third Me s sage Total Number of Me ss age s . Official Sources Fire truck loudspeakers Officials at communica t~on centers Officials away from com- munication centers Total, official sources Unofficial Sour ce s Relative s Friends Stranger s Total unofficial sources Source Undetermined 1. 3 O. O O. O 10 3 7 1. 3 2.6 2. 6 2. 6 20 16 38 29 . 7. 7 O. O 83 3 Total, percent respondents who heard threat messages 9. 0 Total, number respondents who heard threat messages 7 106 32
advice to evacuate, whereas only 32 per cent of those in the non- flooded areas who received threat messages said that they received advice to evacuate. This difference is not statistically significant. (Chi-square - 1. 71; P - ~ 20> . 10~. Although the flooded population reported receiving advice to flee most often, and also fled most often, flight was evidently not a result of this difference in message. For those who did evacuate, both between and within the flooded and non- floodecl groups, the inclusion of advice to flee does not increase the p robability of flight. E. Reactions of Respondents to the False Report · . . . ... . 1. Belief in the false report (See Table 9 below): Although . . the false report was in most cases transmitted through unofficial channels, often by a stranger in the street, only eight (10. 3 per cent) respondents maintained consistent disbelief; six of these eight are from the non-flooded area of the city. Sixty-two per cent of TABLE 9 BELIEF IN THE FALSE REPORT Degree of Belief At least partial belief Complete belief after one exposure to thre at me s s age . . e · · ~ · · · e Complete belief after two exposures .. Complete belief after three exposures ... Partial belief (regardles s of number of exposures) ............ Total, at least partial belief . Consi stent di sbelief, r egardle s s of numbe r of expo sure s Percent (N- 78) .. 61.5 5. 1 2. 6 85. 9 Degree of belief impossible to determine ........3.~3 Total per cent hearing false report 100. 0 33
respondents believed the report the first time they heard it. Table 10 indicate s that half e stimated that they and their f amilie s we re in s er - ious danger of bodily injury; an additional 28. 2 per cent thought there was some possibility that they were personally in danger. Twenty- three per cent thought that the waters would arrive in an hour or less; pre sumably thi s group may have wondered whether or not there was time for them to e scape. TABLE 1 0 RESPONDENTS' ESTIMATE OF DANGER Type of Danger Physical Property Percent (N=78) estimating danger as _ No Doubtful Dange r E stimate 28.2 19.2 26. 9 1 9.2 Certain . 50. 0 J 51.3 2.6 _ 2.6 Total Percent (N = 78) 100.0 100.0 Those respondents who lived in the area which hack been flooded the week before estimated twice as often as the others that they were definitely in physical danger, as well as in danger of property loss. TABLE 1 1 ESTIMATED DANGER -- FLOODED AND NON-FLOODED GROUPS Type of Dancer Physic al Property Flood Expe rience Flood Non-flood Floo d Non-flood P er c enter e stimating dange r as I No Dange r 37 41 No Certain 37 35.1 41 63.4 35. 1 _ 65.9 Doubtful 35.1 22. 0 24. 3 14. 6 35. 1 24. 3 1 9. 5 14.6 *Values in each row of the table are percentages of the number of re- spondents indicated in the "In" column of that row. Values do not total to 100~o in the "flood' group since in two cases it was impossible to determine the degree of danger perceived. 34
Re spondents were generally awar e of the approximate loc ation of Wallenpaupack, but there was a good deal of disagreement as to how long it would take the waters to reach them if the Dam actually had bro- ken. Respondents' estimates of time to inundation are found in Table 12 below. This disagreement is reasonable because respondents had no way of estimating how long the Dam had been broken before they re- cei-ved a warning. In addition, we may reasonably elect that very few people would agree as to how fast the water would travel. Fourteen respondents (17. 9 per cent) were unable to make any estimate of how long they thought they had before inundation. Previously flooded re- spondents e stimated somewhat le s s time before inundat, on. TABLE 1 2 RESPONDENTS' ESTIMATES OF TIME TO INUNDATION Time P e r c ent giving e stimate in . Non-flooded Flooded E stimate Group Group (N= 37) (N=41 ) 0-15 min.2.7 19.5 30-45 ruin.10.8 12.2 1-1-1/2 furs.5.4 Z2. 0 2hrs.13.5 22.0 3 { furs.S. 1 9. S No danger or belief24. 3 4. 9 Cannot estimate29. 7 7. 3 Undetermined5. 4 2. 4 Total100.0 100.0 . . 2. Activities before flight: After receiving the threat message, some residents fled immediately. Others attempted to contact absent family members, to aid other community members, or to confirm the truth of the false report. a. Confirmation attempts: Table 13 shows that of the seventy-eight respondents who heard the message, thirty-three (42. 3 35
per cent) made an active attempt lo verify the report before flight. More than three-quarters of those who tried to verify chose an official source of information. TABLE 1 3 RESPONDENTS SEEKING TO CONFIRM THREAT MESSAGES Tone of Reaction Respondent attempted to confirm before flight Re spondent bell eve d, but did not attempt to confirm before flight . Respondent did not believe sufficiently to attempt confirmation . ~ ........ Total respondents receiving threat messages Percent (N = 78) 42. 3 50.0 7. 7 · · - ... 100.0 TABLE 1 4 MAJOR SOURCE OF CONFIRMATION Source Chosen for Confirmation Official source ......... Semi-official source .... Unofficial source (family, friends, pas ser shy) Source undetermined.... Total r e spondents s e eking confirmation Percent IN = 33) 75. 8 3. 0 .. 18.2 3. 0 1 00. 0 There were three major channels used by respondents for verification, each of which seems to have been employed as often as the next. (See Table 15 following. ~ 36
TABLE 1 5 MAJOR CONFIRMATION CHANNEL Confirmation Channel Percent (N = 33) Face-to-face contact 33. 3 T elephone contact .. . . . .. . . . . . . .. . . . . 3 0. 3 Turning on local radio station 33. 3 Listening to short-wave radio broadcasts 3.1 Total respondents attempting confirmation ........... _ · · · e · ~ ~ · 1 00e 0 Those who turned on the ractio were in some cases met by silence, since the radio station hac3 not as yet gone back on the air. In other instances, they heard the denial message. Of the thirty-three respondents who attempted confirmation, eighteen (54. 5 per cent) made only one such attempt; thirteen more (39. 4 per cent) made two or three attempts, while the number of attempts macle by the two others (6. 1 per cent) is not known. b. Attempts to assist other community members: An important variable in determining the Begs e e of di sturbance in the community is the extent to which people tried to help each other in the situation. There is evidence in the literature (5, 15) that community orientation tends to break down as a disaster situation reaches panic proportions . We ther efore attempted to dete rmine how re spondents tried to assist others before they received denial messages and before flight. Twenty-eight respondents (35. 9 per cent) attempted to assist other community member s in one way or another . it Thi s is shown in sit is interesting to note that one code "Attempted to assist others in preparation for disaster, but not in flight ' was void. This code would have been used to describe respondents who attempted to assist others in non-flight measure s for withstanding the anticipated disaster ; e . g., moving people into attics to avoid the waters. 37
Table 16 below. TABLE 1 6 ATTEMPTS TO ASSIST OTHER COMMUNITY MEMBERS T~^ Of ~ ^=f~;~= Attempted to as sist community member s by conveying the rumor to them 1 5. 4 by calming them ~ ~ 15. 4 by seeking confirmation for theist 5. 1 Total re spondents receiving threat me s s age s . Percent (N = 78) Total respondents attempting to assist 35. 9 Believed, but did not attempt to assist others 51. 3 Did not believe sufficiently to attempt to assist 13. 8 100.0 Later on it will be seen that twenty-one respondents spread the denial message, as opposed to the twelve who spread the rumor. This seems to imply that residents were more active in spreading denials than in spreading the rumor. Such an inference is untenable for the following reasons: (1) These totals do not reflect the percent- age of all me s s age s spread by non-official community member s, and (2) transmittal of the rumor is limited to the relatively short period of time between receipt of threat and denial messages, while the time for spreading the denial message could have been much longer. 3. Respondent impressions of the reactions of others: Re- spondents were asked to indicate their estimate or tne reactions of others in their vicinity to the threat messages. - ~ ~ ~~ Because it was diti~- cult to judge responses which described the emotional states of others, and because the validity of such responses is questionable, we recorded only whether or not the respondent perceived others in his vicinity evacuating. We must consider whether or not these data are contaminated by the r e spondent' s flight behavior. Ther e i s strong age e ement 38
between the answer to whether many others were fleeing, and the answer to whether one fled: Non-flooded group: 3 fleci; 5 claimed many others were fleeing Flooded group: 23 fled: Z5 claimer} many others were fleeing Saturation group: 23 fleck: 19 claimed many others were fleeing Entire sample: 49 fled: 49 claimed many others were fleeing An examination of Table 17 below will confirm the fact that the two re- sponses are dependent. The same people who fled, in general, are those who report that many others in their vicinity were fleeing. In the first part of the Table, geographical location is held constant, and in the second part, belief is held constant. TABLE ~ 7 REPORTS OF OTHERS FLEEING, AND THEIR RELATION TO RESPONDENTS' OWN BEHAVIOR Sample Examined A. Flooded area re spondents B. Those with some belief in false report C. City-wide sample . . ' 1 .. It's Own B ehavior R fled R did not flee R fled R did not flee No. reported fleeing None T A few ~ Many 0.0 4.369.6 32. 1 21.442. 9 ~- 2. 0 4. 167. 3 40. 5 23. 826.2 . _35.9 12.838.5 ? 26. 1* 3. 6 26. 5* 9. 5 12.8 Total Percent 100.0 1 00. 0 100.0 100. 0 100.0 Tota . Numbe r 46~* 28** 49: 42** 78 The large number of cases in which respondents who fled did not estimate the reactions of others in their vicinity (resulting in a code of "undetermined") is explained by the fact that fleeing residents often reported that they were "too agitated" to notice what others were doing. *:kThe large totals here, as contrasted with values presented in Tables 2 and 9, are a result of the inclusion of the "saturation group" sample as well as the city- wide s ample . Three hypotheses can be offered to account for the dependency of these responses: (1) the respondent wished to justify his flight by exaggerating the threat he perceived, and claiming that he was not 39
alone in his response. (~) Flight was closely limited to small, proxi- mal geographical areas, within both the previously flooded and the previously non-flooded sections of town. (3) the sight of others fleeing affected the flight behavior of respondents. While we question the~validity of the responses to this ques- tion for the reasons just given, we present below the statements of respondents in the city-wide sample in part C of Table 17. 4. Flight behavior of respondents: As indicated in Table 2, twenty-six respondents fled as a direct result of hearing the threat messages. We wish now to examine four aspects of the behavior of this group: ( 1 ) With whom did the re spondent flee ? (2) How quic kly did he flee ? ~ 3) What did he take with him in flight ? (4) Unde r what conditions did he return to Port Jervis ? Table 18 indicates that no one of the respondents evacuated alone. Of those who did flee, six (23. 1 per cent) attempted to assist community members (who were not in the party fleeing with the re- spondent) before they left. TABLE 1 8 NATURE OF FLIGHT Nature of Flight - Solitary flight . . . e ~ ~ ~ e ~ e ~ ~ e ~ ~ e e e ~ ~ e ~ ~ e e e ~ e e e ~ ~ e e Flight with family (no attempt to assist others) . . Flight with f amity and/ o ~ neighbor s (no attempt s to assist others). e ~ ~ e e ~ e ee e e e e e e ~ e ~ ~ ~ e e ~ e Flight with family and/or neighbors (after attempts to as sist others) . e ~ ~ e e e e e e e e e e e e e e ~ e ~ e ~ e ~ e e e ~ e e e Information not obtained. . e e e e e e e e e e e e e e e e e ae ~ e e ae ~ e ~ e Total respondents fleeing Percent (N = 26) Be O e e ~ e e e e e ~57. 7 e e e e ~2 3 e 1 Be 8 40 1 00. 0
Flight generally took place immediately after receipt of the f fir s t warning me s s age . Twenty- one ~ 8 0. 8 p e r c ent) of tho s e who fled left immediately after hearing the first false report. Two (7. 7 per cent waited for a second threat message, and two waited for a third. One respondent fled after appreciable delay, but after hearing only one thre at me s s age . Many Port Jervis residents had made preparation for evacua- tion because of the generally prevailing flood conditions. In determin- ing things people think of taking with them in flight, we made a distinc- tion between respondents who indicated that they had made prior preparation and those who had not. Some of those who contemplated flight were stopped before they left by receipt of denial; in many of these cases, respondents had decided what they would take with them in flight. When this was so, we recorded the things the respondents had planned to take, even though they did not finally flee. In thirty-eight cases we were able to deter- mine the kind of thing respondents decided to take with them in flight. These findings are presented in Table 19 below. TABLE 1 9 MATERIALS RESPONDENTS PLANNED TO TAKE IN FLIGHT AND ITS RELATION TO PRIOR PREPARATION Prior Preparation Survival Value Mate rial Percent of Respondents Taking: _ Only Non- survival Value No Mate r tat Mate rial T o tal 100. C 100. C foot _ Ye s, prepared No, unprepared , All, prepared & unprepared 60. 0 63.6 63. 2 40. 0 9. 1 1 3. 2 41 ~ Percent N (N s 38) 5 33 . . 38 1 3. 2 86. 8_ 100.0
If we consider flight separately from other manifestations of belief, we can draw at least one interesting comparison. Of those who heard the threat message, approximately one-quarter (26. 9 per cent) evacuated immediately after receipt of the first warning. The likeli- hood of flight drops tremendously if the respondent does not flee im- mediately after receiving the first warning. When those who had fled heard the fir st denial me s s age, only 2 3. 1 pe r cent returned home immediately. More than one denial message was generally necessary to reassure the population which had fled. In the discussion which fol- lows more detailed consideration will be given to respondents' activi- tie s afte r denial . F . Di s s emination of the Denial Me s s age s _ 1. Frequency: The denial message began to spread at approx _ imately 11:40 that night. In collecting data on these messages, it wasim- possible to get respondents to give a clear estimate of the number of times they heard denials. Unlike the threat messages, denials were given through mass media broadcasting at frequent intervals. Had a respondent tuned his radio to the local station, or were fire-trucks equipped with loudspeakers passing through his neighborhood, he might hear a denial message every few minutes for a period of hours. Re- spondents could give only a general idea of the number of these messages they heard, although they could recall with some accuracy the different sources from which they heard denials. From the general statements of respondents, we attempted to develop reliable categories for coding the number of messages received. We found that it was possible only to make a few broad distinctions; i. e., respondent heard one, a few, or many denial messages. There- fore, mo st of the information concerning denial s was broken down c are - fully by source and channel of transmission, making use of the accurate responses in this area. The minimum average number of denials heard by respondents (before or after belief in the denial) is well over three. Denial mes- sages are, then, far more frequent than threat messages. Most respon- dents heard fewer than two threat mes sage s. This information is pre - sented in Table 20 below. 42
TABLE 20 NUMBER OF DENIALS HEARD BY RESPONDENTS ~I. . Number of Denials One 2 -4 5 or ~ore.... Total Percent (N=78) of Respondents 1 9. 2 33. 3 47. 4 100.0 2. Source and channel of the denials: . TABLE 2 1 NUMBER OF SOURCES OF DENTAL AND NUMBER OF DENIAL SOURCES BEFORE BELIEF Number of Denial Sources Never 2- 3~ ~~ 4-6 Believed ? . Number of Denial Source s 33. 3 Numb e r of Denial Sour c e s Before Belief 44. 9 Total Percent (N = 78) 55. 1 37. 2 10.3 1.3 100.0 6.4 9.0 2.6 100.0' . . . Table 21 indicates that only 44. 9 per cent believed the denial after hearing it from one source (and they may have heard numerous messages from any one source), while 61. 5 per cent of respondents believed the first threat message. (Table 9) Sources for the denial message (Table 22) differed greatly from threat sources (Table 8~. Of the 189 messages whose source we can determine, 79.4 per cent were transmitted by officials, while only 18. 9 per cent of the threat messages were transmitted by officials. 43
We had hoped to be able to compare sources which led to belief in the threat with sources which led to belief in the denial. However, because of the difference in the actual sources, there are insufficient data for such a comparison. TABLE 22 SOURCE AND CHANNEL OF DENIAL MESSAGES Percent (N=78) Receiving Source and Channel Official Sources Fire-truck loudspeaker Lo cat r adio s tation ~ WDLC Central official (face-to- face contact) Central official (telephone c outact) Official on the street Total, official sources Unofficial Sour c e s Relative Friend Stranger Total, unofficial source s Wate r did not c ome Undetermined _ Total, percent respondents who heard denials 5. 1 14.1 10.3 O. O 100 0 RO a Fir st Message 25. 6 1 5. 4 9. 0 Second Message 26. 9 33.3 O. O Third Mes sage 24. 4 29. 5 T otal Numbe r of Messages 60 61 Message Le acing to Belief 1 5. 4 29. 5 1. 3 7.71. 32.69 3.8 11.53.80.012 3.8 69.265.357.8150 62.8 1. 31.36 5.1 5. 10. 015 3. 8 3.85.115 _2.6 29.510.26.436 11.5 2.61.33 11.5 1.3 :46 1 13 11 '+ 66.7 ___ 91.0 Total number respondents | 78 | 63 52 | 193 ! ~ *Some respondents felt reassured only when the flood water did not come when they expected it; a number of respondents did not feel safe until some time the next day, although they had heard many denial messages before then. 3 8 9 ~6 3 4 10.3 44
From the preceding discussion, it is clear that we can make no inferences from these data as to the sources which were most effective in eliciting belief, given an equal number of exposures. Var- iables other than frequency of exposure also contaminate these data: (1) The fire-trucks were generally broadcasting earlier in the evening than was the radio. Presumably, additional cues later at night, such as the sight of other cars returning, increasing quiet in the streets, and so forth, may have made it easier for respondents to accept denial from source s which transmitted the me ssage later. (2) An examination of Table 27 will show lih et, since the radio did not come on until after midnight, it was not heard as a first source for denial as often as it was a second source. In fact, for first and second messages, the numerical quantity of radio reports increases as total messages cle- crease. Since the radio report was frequently a confirmation of a pre- vious me ssage from another source, we must allow for the pos sibility that it was believed more often because it came as a reinforcement of (any) other me s sage. These same data do, however, provide us with another em- pirical check on the validity of the obtained responses. We know from official reports the times at which fire-trucks and radio were broaci- casting the denial. The increase in radio messages and the decrease in fire-truck messages as time passes is entirely consonant win the official time reports. 3. Content of the denials: Five of the 136 denial messages . whose content we could determine were ambiguous in their statement. These messages said, with some differences in wording: ''There is no basis for the rumor at the present time; we are now checking with officials at the Dam. " Such messages always occur as a first denial message; this fact provides us with an additional empirical check of the data. Presumably such reports were received between 11:30 and 11:45, the time at which we know officials were attempting to verify the false report. Consultation with radio announcers at the local radio station revealer} that they made special efforts on three counts in wording their cl enial me s s age s: ( 1 ) They avoided any mention of a dam, as suming that partial receipt of a broadcast might contribute to the disturbance. (2) They mentioned official sources with whom they hac3 checked the false report. ( 3) They rotated announcer s, so that people me ght hear 45
the voice of some announcer whom they knew and in whom they had confidence. The specific wordings which resulted from these decisions are reflected in Table H in Appendix D. Twenty - s even denial me s s age s r eport an official check for verification. Forty-seven specifically mention a dam. Twenty-five messages tell the population to remain home or to go home. G. Reactions to the Denial _ 1. Denial action: (See Table 23) Only six of the respondents who had fled (23. 1 per -cent) were willing to return home immediately TABLE 2 3 RESP ONDENTS' AC TIONS UPON REC HIP T OF DENIALS A. Fleeing Population Percent of fleeing popula tion (N=26) Percent of Total popula- tion tN=78) Returned home immediately after he aring fir st denial Returned after second denial Returned after third denial or more Returned after appreciable delay but after only one or two denials Information not obt ained ~ . B. Non-fleeing Population 23.1 11.5 50.0 1 1. 5 3. 8 Percent of non -fle eing population(N = 52 ) - _ 7.7 3. 8 16. 7 3.8 1. 3 - . Percent of Total popula tion (N=78) Had planned to flee but were stopped by receipt of denial Believed report, but had not planned to flee Stayed on regular or disaster job, regardle s s of belief 1 3. 5 Never believed, so took no action 17. 3 Information not obtained 1. 9 1. 3 Can attempt was made to develop reliable categories which would indi cate how long those who fled remained out of the city. So many respon dents were unable to estimate this time interval that this attempt was abandoned. 46 32. 7 21. 8 34. 6 23. 1 9. 0 11.5
after hearing one denial message. Three (11. 5 percent) waited for a second denial mes sage; thirteen (50 per cent) clid not return until after they had heard three or more denial messages. Three others (11. 5 per cent) returned after appreciable delay, but after having heard only one or two denial messages. The information could not be obtained for the other respondent who fled. Of the fifty-two respondents who did not flee, seventeen were in the process of evacuating when they received the denial message. Eighteen others believed the false report but had not contemplated flight. Seven respondents stayed on their jobs, regardless of belief. 2. Denial communication: Twenty respondents (25. 6 per cent) communicated the denial message to others in face-to-face contact. One fireman among the respondents helped broadcast the message over the fire -truck loudspeaker system. No respondent used arty other channel to communicate the message. Fifty-six respondents (71. 8 per cent) never transmitted the denial message to others. H. Attitudes Toward Disaster Groups 1. Civil Defense: The local Civil Defense organization was active In t ~ ooa re ' lee worn, and also took an active part in di spelling the effects of the false report. Of the seventy-eight respondents, twenty -four ~ 30. 8 per cent) spontaneously expre s sed approval of the group's disaster work. When the others were specifically asked to comment on this question, thirty-eight more (48. 7 per cent) expressed approval. Eleven respondents (14. 1 per cent) felt that they did not have sufficient information to comment. This data is presented in Table 24. TABLE 24 A T T I T UDE T O WARD CI V! L DEF ENS E A GE NC Y Respondent's Attitude Percent (N = 78) Spontaneous approval Spontaneous Hi s app royal When specifically asked, approval When specifically asked, disapproval Cannot estimate Undetermined by interviewer Total respondents 47 30.8 1. 3 48. 7 1. 3 14.1 3. 8 100.0
2. Other groups: Respondents were asked to name any other groups which they thought had done a good or bad job during the disaster period. The relevant data is presented in Table 25. The American Red Cross (the only outside agency) was mentioned spontaneously twenty-nine times, and in four of these cases (13. 8 per cent) it was with disapproval. Specific groups other than the American Red Cross were mentioned spontaneously 148 times, and in five of these cases (3. 4 per cent) it was with disapproval. There is some evidence in the literature that there often is a certain amount of resistance to groups from outside the community which move in after a disaster has occurred. The infrequency of disapproval mitigates against any meaningful ex- amination of thi s i s sue with the se data. Re spondents generally felt that the local Fire Department did a good job during the disaster period; 79. 5 per cent of respondents specifically named this group, and always with approval for its activities. TABLE 25 ATTITUDE TOWARDS DISASTER GROUPS Group Police Department Fire Department American Red Cross Other Groups Unspecified Groups I. Advice to Others Approval Attitude Expressed Dis approval 14. 1 1. 3 79.5 0.0 32. 1 5. 1 59. 0 3. 8 53. 8 0. 0 _ Unmentioned 84. 6 20.5 62. 8 37. 2 46. 2 Total Percent (N = 78) - 1 00. 0 1 00. 0 1 00 e O 1 00. 0 1 00. 0 Respondents gave a variety of suggestions regarding action to be taken in case of the occurrence of further disasters of this sort. Twelve (15. 4 per cent) respondents advised others to turn on their radios, to check for confirmation, or wait for a siren before evacuat- ing. This last emphasizes the threat potential the siren had for the population. Five (16. 4 per cent) respondents advised others to check 48
with officials, but specified no channel. Nine Gil. 5 per cent) advised othe r s to "keep calm. " One r e spondent re commended evacuation. Seven (9 per cent) were unable to give any advice. Eleven (14.1 per cent) gave miscellaneous advice, such as "Next time tell them not to take down the dikes to build playgrounds. . . " In twenty cases (25. 6 per cent) the interviewer failed to collect this information. J. Summary We estimate that approximately three-quarters of Port Jervis residents heard the rumor the night of August 20, 1955. The impact of this false report on the population is indicated by the fact that a third of those who heard it evacuated subsequently. Respondents also heard siren signals, both from the fire-truck and from a siren blowing in Matamoras. Since the siren was perceived by marry as the official signal of impending disaster, it added credibility to the verbal report. Almost 90 per cent of the respondents who actually fleet Port Jervis were those who lived in the area of town which had been inundated dur- ~ng the flood. In addition to those who fled, another 23 per cent of respon- dents who heard the false report were preparing for flight when they were stopped by receipt of the denial message. An analysis of the activities of those who fled shows us that they generally left immediately after receiving the first threat message. Evidently, people who delayed evacuation for any length of time were likely to receive disconfirmation in time to keep them at home. Thirty-two per cent of respondents who heard the report made some attempt to verify it before flight. A large number of resi- dents maintained some level of community responsibility during the threat period, even in those cases where they believed the false report. Some attempted to rouse neighbor s and friends, to seek verification for others as well as themselves, and to keep the community relatively calm. Some of these same people later fled. No one in Me sample fled alone. Families and neighbors evacuated together, usually after packing cars with food, blankets and other survival material. It is significant, however, that about a third of those who fled did not take any possessions at all when then evacua- ted. This group may have felt that they had no time to gather things 49
together, that the period of inundation would be too short to necessi- tate their taking anything along, or they may have been too agitated to think of collecting their belongings. Although most of the people who fled had left after hearing one threat message, less than a quarter were content to return to Port Jervis after hearing one denial. Half waited until they heard three or more denial messages. Of the total respondent population, 61. 5 per cent believed the first threat message, while 44. 9 per cent believed the first denial message to which they were exposed. This fact is perhaps surprising when we consider the sources of threat and denial messages. Approximately 20 per cent of the threat messages were transmitted by officials, while almost 80 per cent of the denials were transmitted by officials. While threat messages were usually communicated in face-to-face contact with friends, family and strangers, the denial messages were broadcast through mass media such as loudspeaker systems and the local radio station. 50
