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.
APPENDIX C CODING RESPONDENT PROTOCOLS One of the major tasks involved in the study was coding re- spondent protocols. In constructing the categories for this operation, a number of factors had to be considered: Some investigators in this field (20) have pointed to the lack of theory from which to derive hy- potheses for testing. It was important therefore, to construct cate- gories which would yield data which could be studied in an exploratory fashion, as well as permitting the testing of the hypotheses. (This problem is treated more fully elsewhere. (2, 21) At the same time, a certain amount of descriptive data was necessary as a means of orienting the readers and investigators to the actual sequence of events. Finally, the categories had to be defined well enough to per- mit coders to categorize respondent statements reliably. This last s ometime s means that we must sacrifice p sychologically me aningful distinctions (concerning feelings, emotions, etc. ~ for more reliable categories dealing with easily observable characteristics. The procedure used in constructing categoric s is explained in the following paragraphs. Two investigators who were completely familiar with the kinds of information needed for fulfillment of the research constructed the categories. The general areas to be coverer] were listed and a definition of each written. The definitions were based in part on first- har~d experience which one of the investigators had in Port Jervis, and in part on a thorough examination of the ''key" interview protocols. To avoid the introduction of bias into the final analysis, the respondent protocols were not used in this step. After the categories had beer defined, and various codes developed under each category, a small sample, 10, of the respondent protocols was taken to check the adequacy of the coding proceclure. Each inve stigator independently coded five protocols and then compared his listings with the other. Modifications were made and the next five were coded. This procedure was continued using the same ten proto- cols and a few of those from key interviews until agreement between the investigators was high, and it was judged that the information re- quired would be obtained with the categoric s . 101
At this point, two completely naive coders were trained. They were given a general summary of the background of the study, but were not exposed to the hypotheses, proposals, or any discussions of the specific information needed. As is indicated on the sample page in this Appendix entitled "Instructions for Coders, " each coder was directed to familiarize himself completely with the set of codes. Following this, each coded one of the ten protocols used in the pre- ceding steps. The results were reviewed by one of the code con- structors, and clarifications of various points were made. Next, they independently coded four more of the s e protocols . They r econ- ciled differences and then referred their work to the same investigator. After coding the remaining five protocols, it became apparent that further training was unnecessary, and the coding procedure began. These ten protocols could not be thrown out of the sample, although the special treatment given to them may have introduced some minimal bias into the results. The coders independently coded sets of five protocols. They then met and reconciled all differences, at the same time keeping a record of such differences so that the degree of initial agreement could be determined. This procedure was continued until about half of the protocols were coded. At this point it became evident that it was uneconomical to complete such small sets of protocols before each consultation; the size of the sets was then increased to ten and later to fifteen. The per cent agreement on each category is given in the accompanying table. The percentages range from 81. 1 per cent to 100 per cent. Those with the highest reliability primarily concern biographical data. No category can be considered unreliable. It is important to note, however, that the reliability estimates given in this table are based~upon the independent judgments of the two coders. These figures are overestimates of the reliability of the categories as they are written. The frequent meetings of the two coders resulted in their reaching mutual understanding as to the mean- ings of categories and statements in the protocols. On the other hand, the figures listed are underestimates of the reliability of the data analyzed. These data were based on the pooled judgments of the coders, which are obviously more reliable than the independent j udgment of either co de r . 102
TABLE A PERCENT INITIAL AGREEMENT OF CODERS A N = 146 - Total sample B N = 111 - Total who heard false report C N = 333 - Total for three messages* PERCENT RANK AGREEMENT CATEGORY TITLE NUMBER IN SAMPLE . 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 99.3 99.3 99.3 99.3 99.3 99.1 98.6 98.6 98.6 98. 98.2 97.9 97.9 97.3 97.3 96.6 Sex Code Number Code Number Code Number Socio-Economic Status Automobile Numbe r in Hous ehold Heard False Report T elephone Number of Children Length of Residence in the City Flood Evacuation Flood Victims Staying in Home Attitude T o war ~ P ok c e D ep t. Age Disaster-Related Job Fled as a Function of Rumor Location When Heard False Report Attitude T award Red Cro s s Schooli ng Previous Disaster Experience Expo s ed to Previous Rumor s Attitude Toward Fire Dept. When Flood Victims Stayed in Home A A A A A A A A A A A A A B A A A B B A A B B A Reliabilities for the three threat and denial messages were computed together. Therefore the N = three times the number exposed to the reports . 103
PERCEN T RANK AGREE MENT CATEGORY TITLE NUMBER IN SAMPLE 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 96. 4 95.5 94.6 94. 6 94. 0 93.8 93. 7 93. 7 93. 4 92. 8 91. 9 91.9 91.0 91. 0 91. 0 91. 0 90.1 90.1 89. 8 89.2 89.2 89.2 89.2 89.2 88.3 88.3 88.3 88.3 87. 4 87. 4 86. 5 85. 6 85. 6 84.7 81.1 Mean Reliabilitv = 93. ~ nor ren1: 104 Group with Re spondent - Strange r s Denial Communi c ation Activity when Heard False Report Total Denial Sources Threat Content Damage by Flood Group with Respondent - Farnily Attitude T o war d O the r Or our s Source anti '_ ., ~ 1 Channel of Threat Group worn Re spondent - Neighbo r s Time Re spondent Heard Fal se Repo rt Communication with Family Estimated Time Until Inundation C onfir mation Channel Re actions of Othe r s Attitude Toward Civil Defense Speed of Flight Attitude Toward Unspecified Groups ~_ ~. ~. ~ -O ~ource en cl ~nanne1 ot tne L'enla Belief in the False Report Stated Personal Danger Attempts to As sist Community No. of Denial Sources to Belief Total Denials Heard Location of Family During Threat Estimated Danger to Property Things Taken in Flight Denial Action Confirmation Source Content of the Denial s Advice to Others F amity Re sponsibilitie s Nature of Flight Number of Confirmation Attempts Denial Source Leading to Belief B B B B C A B C B B B D B B B B C B B B B B B B B B B C B B B B B Median Reliability = 93. 9 per cent
In addition to the sample codes in the Appendix, the reader will find a copy of the coding sheets used to record the coders' judg- ments. The use of a separate code street has two primaryadvantages: (1) It permits the coding operations of each coder to take place inde- pendently. The practice of recording codes on the protocols obviously has an influence on the judgments of subsequent coders. (2) This format speeds up the punching operation when the codes are to be re- corded on IBM or similar cards. The punch operator can quickly and accurately punch and verify codes recorded in thi s fashion. ANALY ZING THE KEY INTERVIE WS Because of the varied nature of the key interviews, and be- cause a different type of information was desired, it was impos sible to use the same set clef codes for their analysis as for the respondent protocols. Basically, we wanted an accurate description of the actions of officials and semi-officials during the period of the false report, rather than the testing of hypotheses. Data of this sort do notlend themselves well to the latter type of treatment. One of the category constructors and one of the coders, who was by now sophisticated with respect to the study ant} its objectives, worked cooperatively in pulling the information from the key protocols. First, lists of actions taken in various centers were made by the coder. These were checked for accuracy by the category constructor. Then communications charts were constructed and these were also checked. In some instances, data were missing from the protocols, while in others factual disagreements between respondents in the same center were noted. In such cases, both investigators agreed to avoid infer- ences and to report only those data which could be supported by state- ~ents in the protocols. Finally, descriptions of the action were written along with the conclusions drawn from them, and checked for accuracy by the second investigator as well as by a third investigator who had not taken part in the phases described above. 105
SAMPLE PAGES FROM INSTRUCTIONS FOR CODERS Introduction In the pages which follow, you will find a complete set of in- structions for coding protocols from Operation DAM. It is extremely important that you follow these instructions very carefully. If you have any questions concerning the nature of the various categories, please consult with either Dr. Danzig or Dr. Thayer. In order to complete your assignment you will need the follow- ing equipment: (1) this set of instructions for coding, (2) the set of interview protocol s as signed to you, (3) a copy of the interview s chedule, and (4) a set of coding sheets. After studying the instructions for coding carefully and settling the answers to any questions, glance over one or two of the protocols so that you may become familiar with the material with which you will work. Next, select your first protocol for coding and follow the instructions below. Please do not mark the protocols themselves in any way. The only place you are to make any notations is on the coding sheets themselves. The adequacy of this study as a piece of research will, to a large extent, be determined by the coding job which you are doing. All hypotheses will be tested with the coded information that you will supply. The conclusions reached and recommendations made will also be based primarily on your work. It is evident, therefore, that your job is a most important one and must be performed with care. Coding Categories - You are to code the face-sheet data first. After this you will code the information contained in the rest of the protocols. You will note in the category definitions given below that there is a uniformity throughout with respect to coding the response ''none" and coding where no information is given. In any column where the re- sponse is "none" or its equivalent (the category is not applicable), you will always code zero (0~. In any column where there is no response; i. e., no information is given, you will code "X. " In other words, "X'' is equivalent to missing data; i. e., it looks like the interviewer should have included this information but failed to Lo so. Please make sure that you follow this procedure in all cases as it will simplify the com- puting and interpreting job. (Sample codes are reproduced on the following two pages. 106
COLUMN NUMBER CATEGORY DEFINITION SEX: On the coding sheet in the space opposite column 1 indicate the sex of the respondent as follows: 1. Male 2. Female RESPONDENT CODE NUMBER: Opposite column 2 write the first digit of the three-digit code number given on the face she et. RESPONDENT CODE NUMBER: Write the second digit. 4. RESPONDENT CODE NUMBER: Write late third digit. 5. 6. SES: 28. AGE: Code the respondent age as follows: 2. 3. 4. 1. 20 ~1 26 31 or under to 25 to 30 to 35 5. 36 to 40 6. 41 to 50 7. 51 to 60 8. 61 anti over 1. A 2. B 3. C 4. D SOURCE AND CHANNEL OF THREAT MESSAGE -- FIRST MESSAGE: For columns 28 through 30 the same coding will be used. You will note that two major distinctions are being made with respect to the source and channel of the threat and denial messages. First, we are distinguishing between official and unofficial sour ce s . Se coed, we are di stingui sh- ing among various types of official sources: impersonal, central, and personal official. The code "Impersonal Offi- cial" refers to such sources as the fire-truck loudspeaker, the radio, or television. These sources are obviously 107 .
official, yet the respondents cannot interact with such sour- ces. They must accept or reject the information supplied and cannot ask for clarification or additional information. The code "Central Official" refers to a source at a commu- nications center where the respondent has a chance to inter- act with such a source. Such interaction may take place either in a face-to-face conversation or through the tele- phone. There are two basic characteristics of such a source; (1) He is at a communications center and as a function of this may be perceived by the respondent as having up- to-ciate information. (2) The re spondent has the opportunity to interact with him and ask for reassurance, further clarification, or addi- tional information. The code "Personal Official" refers to a source that is offi- cial but is not at a communications center. NOTE: In counting threat messages, do not count any after the respondent heard the denial message and believed the denial. All columns 28 through 30 would therefore be coded as follows: 1. Impersonal official -- fire-truck, loudspeaker 2. Impersonal official -- radio or television 3. Central official -- word of mouth communication fro civil defense, police, fire, etc., at a communica tions center . 4. Central official -- telephone source from civil defense, police, fire, etc., at a communications center. 5. Personal Official -- policeman, fireman, civil defense, etc., on the street and removed from a communica tion s c ente r . 6 . Unofficial - - source is a relative . 7. Unofficial -- source is a neighbor or friend. g . Unofficial - - source is a pas s er -by unknown to re spondent. 108
OPERATION DAM - CODE SHEET Column Column Number Code Category Number Code Category 9. 10. 11. 12. 13. 14. Sex Code Number Code Number Code Number Age 6. SES 7. Car Telephone Disaster Job No. in Household No. of Children Length Residence S chooling Flood Evacuation 1 5. Damage by Flood 16. Victims With 17. Victims When lS. Previous Experience 19. Fled_ Rumor 20. He ard Report 21 . L`o cation 22. Activity 3 . Re sponsibilitie s 24. Group With - Family 25. Group With - Neighbors 26. Group With - Strangers 27. Where Family 28. Threat S&C - 1st . 29. Threat S&C - 2nd 30. Threat S&C - 3rd . 31. Threat Content - 1st 32 . Threat Content - 2nd 33. Threat Content - 3rd 34. Time Heard Report 35. Belief in Report 36. Est. Danger - Personal - 37. Est. Danger - Property 38. Time Until Flood 39. Nature of Flight 40. Confirmation Source Institute for Research in Human Relations 109 41. Confirmation Channel 42. No. Confirmation Attempts 43. As sist Community 44. Communicate with Family 45. Things Taken With 46. Speed of Flight 47. Reactions of Others 48. Denial S&C - 1st 49. Denial S&C - 2nd 50. Denial S&C - 3rd 51. Denial Source-Belief 52. No. Denial Sources-Belief - 53. TotalDenials 54. Total Denial Sources 55. Denial Content - 1st 56. 1st (continued) 57. Denial Content - 2nd 58. 2nd (continued) 59. Denial Content - 3rd 60. 3rd (continued) 61. Denial Action 62. Denial Communication 63. Previous Rumors . 64. Attitude re CD 6 5. Attitude re Police 66. Attitude re Fire 6 7. Attitude re Red Cro s s 68. Attitude re Other Group s - 69. Attitude re Unspecified 7 0. Advice to Othe r s 7 1 . Sirens and Noi se 72. 73. 74. 75. 76. 77. 78. 79. 80.
