ABSTRACT AKIUMEINTATIVE MESSAGE STRUCTURE AND PRIOR FAMILIARITY AS PREDICIURS 0]? SOURCE CREDIBILITY AND AITI'IUDE CHANGE By John R. Weston Persuasive messages that also take into account arguments which oppose the source' 8 position have been shown to be generally nore effective than messages that present only arguments consistent with the some ' 3 position when receivers are initially Opposed to the source ' 3 position and familiar with the message t0pic . This study investigated the relationship between a receiver ' 3 specific familiarity with argxments that oppose a cammmicator ' 3 position and the extent to which such arguments are taken into account in the message. 'I‘wo dimensions of persuasive messages were studied. A message (1) presented only arguments which were consistent with the advocated position (consmxctive argumentative structme) or also presented argmnents antagonistic to the advocated position (rebuttal. argumentative structure) on (2) fewer, the same, or rnore t0pic-related issues than those with which the subject was familiar prior to being exposed to the message . Prom previous research and theoretical considerations it was hypothesized that attitude change is directly related to the nunber of topic-related issues presented in the message. It wasalso predicted that John R. Weston rebuttal argumentative structure elicits more attitude change than constructive structure, provided none of the antagonistic arguments in the rebuttal message were previously unfamiliar to the receiver . Similar predictions were made about the source credibility induced by the various message types . The differentially induced credibility was advanced as a major determinant of the predicted differences in attitude change between the messages . Thus, it was logically consistent to also hypothesize that the removal of the effects of credibility from attitude change would significantly reduce the attitude change differences elicited by the various messages , when credibility was not controlled . It vas further reasoned that including unfamiliar antagonistic argunents in a rebuttal message strengthens rather than weakens the position the receiver held prior to the cozmunication . Thus , it was hypothesized that a rebuttal argunent that takes into account all familiar antagonistic arguments is nore persuasive than a rebuttal argtment that also takes into account previously unfamiliar arguments. When the effects of source credibility are statistically removed from attitude change, an increase in the difference in attitude change was hypothesized. 2M0 college undergraduates took part in two experinental sessions . Subjects were familiarized with argunents at time-one that would be antagonistic to the position advanced at time-two . The first session also successfuny instilled an attitude tovard the experimental t0pic that opposed the evaluation advocated in tine-two experimental message . V One week later messages were presented which varied on the two dimensions (i.e., en argunentative structure and relative issue familiarity). John R. Weston After the message presentations, at both sessions, attitude toward the topic was determined by sunning across masked versions of the same five evaluative sanantic differential scales . Attitude toward the message source (time-two) Was determined by source ratings on semantic differential scales measuring perceived safety and qualification. The data were analysed by analysis of variance, individual t-test comparisons, and zero-order as well as part correlation techniques . The hypotheses relating attitude change to argunentative stricture were supported . The prediction that also including unfamiliar antagonistic argunents in a rebuttal message would reduce the attitude change elicited was also supported . The prediction that also including unfamiliar issues in a constructive argunent would be more persuasive than including only familiar arguments was not supported. None of the hypothesized effects of the message treatments on source credibility were supported. Neither did removing the effects of source credibility from attitude change alter any of the message effects on attitude change . Previous research has led to the conclusion that the "two-sided" message is more persuasive than the "one—sided" when the receiver initially disagrees with the conmmicator ' 3 position. The results of this study suggest that this traditional proposition should be modified by stipulating that the "two-sided" message not include Opposing argunents with which the receiver is not already familiar. ARGUIVIENTATIVE MESSAGE STRUCTURE AND PRIOR PAl‘flllARI'I'Y AS PREDICIURS OF SOURCE CREDIBILITY AND ATI‘I'IUDE CHANGE By \. . (J 0‘ :‘\ John RX‘Weston A THESIS Submitted to The Department of Communication Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY 1967 Accepted by the faculty of the Department of Conmunication, College of Cannmication Arts, Michigan State University, in partial fulfillment of the requirements for the Doctor of Philosophy degree. afiw/flfi/a Director of Thesis Guidance Cannittee: [flaAfl/firé Chairman ACMOWIEIISEMEN‘I‘S This project was undertaken and conpleted with the considerable encouragement, guidance, and assistance of a number of people. Jo, in particular, was patient, understanding and very kind. Dr. David Berlo, the chairman of my. guidance committee, was unfailingly generous of his time and energies at all stages of the research. For this I am greatly indebted. Doctors Erwin Bettinghaus, Randall Harrison , Eugene Jacobson and Hideya Kumata contributed greatly to the final chapter . I am also indebted to a large nunber of my fellow students who assisted in the construction and administration of the eXperimental materials, the analysis of the data, and generally improving the thesis. TABLE OF CONTENTS CHAPTER Page I INTRODUCTION . . . . . . . . . . . . . . . . l Hypotheses . . . . . . . . . . . . . . . . 12 II METHOD . . . . . . . . . . . . . . . . . . 20 The Message Topic . 20 Message Construction . . . . . . . . . . . . 21 Independent Variables . . . . . . . . . . . . 21+ Control Variables . . . . . . . . . . . . . 26 Dependent Variables . . . . . . . . . . . . . 27 Subjects . . . . . . . . . . . . . . . . 29 Experimental Design . . . . . . . . . . . . . 29 Data Analysis . . . . . . . . . . . . . . . 32 Experimental Procedures . . . . . . . . . . . 3 3 III RESULTS 0 O O O O O 0 O O O O O O O O O O 36 TestsofManipulations. . . . . . . . . . . . 36 Tests Of Hypotheses O O 0 O O O O O O O O O O 38 IV SUMMARYANDDISCUSSION . . . . . . . . . . . . 1+7 SumnaryofResults............. 47 DistSj-On 0 O O O O O O O O O O O O O O O l+9 TABLE LIST OF TABLES Mean Tl attitudes toward civil defense for the six message-treatment groups, and analyses of differences amongthemeans............. Mean safety evaluations for the six message-treatment 7 groups, and the analysis of variance among the means . Mean qualification evaluation for the six message- Ueatment groups, and the analysis of variance among the means 0 o o o o o o o o o c o o o o 0 Mean attitude change scores on the experimental concept for the six message-treatment groups . . . . . . Analysis of variance and simple-effects of mean attitude change scores for four of the message-treatment groups iv Page 37 39 H0 #1 1+2 LIST OF FIGJ'RES FIGURE Page 1 Constructive Segments for Each Issue . . . . . . . 22 2 The Six Familiarity Inductioanessages . . . . . . 25 3 TheBasicExperimentalDesign . . . . . . . . . 30 APPENDIX NUDE! LIST OF APPENDICES 'IEX'IS OF THE CONSTRUCI‘IVE MESSAGE SEGMENTS (TM‘ONE AND TM.M) o c o c o o o o IEXTS OF THE REBUTI‘AL MESSAGE SEGMENTS . . . THE TIME-ONE QUESTIONNAIRE . . . . . . THE TIME-TWO QUESTIONNAIRE . . . . . . BASIC EXPERIMENTAL DESIGN: INDIC‘ATING MESSAGE COUN'I'ERBALANCING PROCEDURE AND DISPOSITION OF TIDE—ONE AND TIME-Two WSW o o o o o 0 vi Page 60 67 75 C 8“ 93 CHAPTER I INTRODUCTION One of the earliest message variables to be studied as a potentially useful predictor of influence acceptance was “message sidedness”. These early studies were derived more from the intuitive belief that communication ”sidedness” was a relevant factor than from any well developed theoretic considerations. The studies compared the relative persuasiveness of messages which contained only arguments and infernetion favoring the communicator's point of view with messages which also considered an alternative point of view. The initial investigation of the relative effectiveness of “one-sided” versus ”two—sided” communication was conducted during World war II (Hovland, Lumsdaine 8 Sheffield, 1949). These investi- gators found no overall difference in the persuasive influence of the two types of communication; however, When both education and initial position were considered, the communication giving both sides proved to be more effective among the better educahai, regardless of initial position. The ”one—sided” presentation was primarily effective for the less well educated when they were already favorable to the advocated position. Since this original study, a number of researchers have investigated the persuasive effectiveness of the two types of messages (e.g., Lumsdaine 8 Janis, 1953; Paulson, 195H; WOlfinger, 1955; Thistlethwaite 8 2 Kamenetzky, 1955; Thistlethwaite, Kamenetzky 8 Schmidt, 1956; Crane, 1962; Insko, 1962; Chu, 1966). The findings of some of these studies will be discussed in subsequent sections. Per the present, it is sufficient to note that, in general, the findings are inconclusive and often contradictory. Sometimes the ”one-sided” message is more persuasive; sometimes the ”two—sided” message is more persuasive; sometimes there is no difference. The one-sided/two—sided issue, however, continues to be of research interest, not because the findings are consistent, but because fairly consistently there are findings. This suggests that message ”sidedness” might be operating in conjunction with other factors in the communication situation which have not been controlled or in some other way accounted fOr in the research design. Any investigation of such a conjunctive relationship presumes that the sidedness concept has been explicated. Unfortunately, very little attempt has been made to explicate the concept. Typically, the variable simply has been dichotomized into ”one—sided” and ”two—sided”. A message has been called one—sided if it presents arguments for a given point of view without considering arguments for the Opposing point of view. .A.message that also presents arguments fOr the Opposing point of view has been called two-sided. Thus, one can conclude only that the two-sided message contains an indeterminate amount of unspecified content which is antagonistic to the source's point of view, and that the one—sided message does not. Considering the conceptual and Operational freedom permitted by such vague terminology, it is not surprising that the research which has used these labels has produced confusing findings. Failure to conceptualize the variable has precluded the develop- ment of an adequate theoretic rationale and limited the predictive use- fulness of the "sidedness" notion. Hovland and his associates, in ad- vancing a rationale for the two-sided message, originally suggested that presenting argrmlents on both sides can be defended on the grounds of "fairness" or, the right of an individual to have access to all relevant information in making up his mind (Hovland, Lumsdaine, and Sheffield, 19%). The two-sided message, they suggest, would likely be viewed as less biased than the one-sided and hence would be more persuasive. Later researchers seem to have accepted this interpretation with little refinement. Insko, for example, in predicting that the two-sided message would be more effective than the one-sided, refers to "the cannon sense grounds of the presumably more impressive impact of the two-sided communication " (1962, p. 203) Yet, the research evidence does not provide unqualified support for this presumption. Several studies have shown that the increased effectiveness is often related to two states of the receiver prior to the presentation of the message—evaluative predisPOSition or prior attitude, and issue familiarity or the receiver's information level on the issue. Prior Attitude The studies that have considered evaluative predisposition of the issue and the relative effectiveness of one-sided and two-sided messages have demonstrated that for individuals initially in agreement with the advocated position, including antagonistic arguments in the 4 message decreases persuasive effectiveness (Hovland, Lumsdaine 8 Janis, 19H9; McGuire, 1962; McGuire 8 Papageorgis, 1961). One inter- pretation of this finding is fairly obvious. When an individual is already in agreement with the advocated position, including antagonis- tic arguments raises doubts and thereby reduces the persuasiveness of the message relative to the same message with the antagonistic argu- ments amitt . Issue Familiarity Laboratory experiments also repeatedly have demonstrated that the-subject's familiarity with an issue reduces the persuasive effect- iveness of any message attempting to change the subj ect' s evaluatims (e_.g., Coffin, 19Hl; McGuire 8 Papageorgis, 1961; Crane, 1962; Manis 8 Blake, 1963; McGIire, 1962). Insko (1962) investigated tre effect of prior familiarity on the persuasiveness of one-sided and two-sided messages. He found that familiarity reduced the effectiveness of both types of messages but that the reduction was greater for the are-sided message. In summary, these studies have demonstrated that antagonistic arguments in the message increase persuasiveness when the receiver (1) is initially opposed to the cannunicator's position, or (2) is already familiar with antagonistic arguments. These two findings are quite separate in the literature . No studies have siJmJltanecus ly deter- mined or varied both prior attitude and prior familiarity in investi- gating the relative efficacy of the one-sided and two-sided message. The development of a raticnale that would encompass both sets of findings begins with the question, "Why should including antagonis- tic arguments increase the persuasiveness of a message?" Such arguments 5 Oppose the communicator's position and should, in and of themselves, represent a negative vector, i.e. , a pull away from the position advocated by the communicator. The research findings, however, suggest that the sign of the vector is not necessarily negative but is depen- dent upon at least two factors external to the message—the receiver' s lx‘ior attitude and prior familiarity. To this point, the discussion has focussed on the effect of antagonistic arguments in the message and on prior attitude and infor- mation level of receivers . Considerable research has indicated that a third type of variable, the evaluation of the source, also affects persuasion. If it can be argued that inclusion of antagonistic argu- ments affects source evaluations , and that these changes in evaluation then affect persuasiveness, then it would follow that source evalua- tion must be considered as an explanatory variable operating in con- junction with tte inclusion of antagonistic arguments. TTe Effect of Message Variables on Source Evaluation While it has not been empirically demonstrated that variations in antagonistic message content affect source evaluation, a number of researchers have shown that a diversity of other message variables do affect source evaluation. These variables include the source's verbal fluency (Miller 8 Hewgill, 19 61+) , interphrase rate and Euse time (Leitner, 1962), word familiarity (Carlson, 1960), and the use of socially acceptable language (Harms, 1960). Other writers have demon- strated that source evaluations may also be affected by such factors as the perceived manipulative intent of the message (Walster 8 Festinger, 1962), the perceived objectivity of the scarce as determined from the 6 message (Hovland 8 Mandell, 1952), or whether the message is m tative or conciliatory (Ludlem, 1956). This sample of research does indicate that several message attributes affect sauce evaluations; however , attempts to relate these message induced differences in sauce evaluation to accompanying differences in influence acceptance have met with little success. On the other hand, the presence or absence of antagonistic arguments in a message is a message variable which has been shown to be related to influence acceptance but the effects of this variable upon sauce evaluatim have not been empirically investigated. The importance of such an investigation is underlined by the weight of evidence indicating that sauce evaluation has considerable effect on influence acceptance. Sauce Evaluation and Influence Acceptance The differential effectiveness of various kinds of message sauces on influence acceptance has been repeatedly demonstrated in laboratory situations . These investigations have been conducted under a variety of descriptive labels used to denote the varying effectiveness of sauce attributes on influence acceptance. Prestige, status, charisma, image, repitation and sauce credibility are among those labels employed. This research has shown rather consistently that sources who are favorably evaluated on a variety of attributes are more effective in their persuasive attempts than those less favor- ably evaluated provided the source differences are sufficiently extreme (e.g., Haiman, 1949; Mausner, 1953; Hovland 8 Weiss, 1951; Kelman, 1953; Kerrick, 1958; Hollander, 1961). 7 Over the past four decades, changes in emphasis have occurred in the literature on source evaluation and communication effectiveness . In their review of the area, Clevenger and Anderson (19 63) have detailed the transition from viewing credibility as a characteristic which can be intuitively determined on an 3 225332 basis to the view that the characteristics of sources which form the basis of sauce effectiveness are determined by the perceptions of the receiver. Mertz (1966) has extended the Clevenger and Andersm analysis by tracing the accompanying developments in conceptualizing and measure ing the variable. He points out that while earlier theorists disagreed as to the specific antecedents of sauce effectiveness in attaining influence acceptance, they were unanimous in their treatment of credi- bility as a unidimensicnal attribute—usually measured by a single linear rating scale . The sauce attributes thus scaled included likableness (Saadi 8 Farnswcrth, 19310, trustwcrfi'ziness (Hovland 8 Weiss, 1951), and Ergstige (Adams, 1960). Other researchers obtained sauce ratings on the evaluative dimension proposed by Osgood, Suci 5 Tanrenbaum, (1957) by summing over a number of semantic differential scales which loaded on that dimension (e.g., Wolfinger, 1955; Berlo 8 Kumata, 1957). Since each of these perceived attributes was sham to be related to sauce effectiveness, it became clear that the basis of relevant sauce evaluation caIld be more usefully viewed as multidimensional. Within recent years , attempts have been made to specify the underlying dimensions of sauce evaluation relevant to influence acceptance and to provide reliable instruments for measuring these dimensions. The 8 investigations of a number of researchers employing factor analytic techniques yielded quite similar dimensions (Anderson, 1961; Berlo 8 Lemar-t, 1961; Berlo, Lemert 8 Mertz, 1965; Schweitzer, 1966; Macroskey, 1966). The findings suggest that there are three relatively indepen- dent dimensions upon which sources are evaluated: "qualification" or perceived general ability, intelligence and expertise, "safety" or predictability, honesty and perceived manipulative intent, and "dynamism" or perceived energy and vitality. These studies repeatedly demonstrated that the three dimensions (particularly "safety" and "qualification") account for a large amount of the variability in sauce evaluation. Typically, "safety" accamted for about one-half of the common variance, "qualification" somewhat less and "dynamism" abait one-tenth of the common variance. These studies provide an operatioel definition for source credibility which can be used to investigate the interdeperrience of sauce credibility and message variables on communication effectiveness. Pre-Message Credibility MlCh of the previais research on communicator credibility has investigated only the effect of varying sauce preconceptions on in- f luence acceptance. The typical procedxue has been to attribute the message to one of a number of sauces abart whom subjects are capable of making evaluations prior to the presentation Of the message. Either sauces are well known public figues or evaluatively relevant informa- tion about the sources is provided prior to the communication itself. Any observed difference in influence acceptance is then reasmably attributable to the differential source evaluations. However, in many 9 persuasive communication situations , there is no strong prior sauce evaluation and, in some cases, the sauce is unknown. In these situa- tions, most or all of the cues on which the subject bases his credibi- lity evaluations are limited to variables in the sarrce's message. A research emphasis which is limited to the differential effec- tiveness of sauces abait which credibility evaluations are fairly well stabilized poses the rather narrow question as to the tempering effect of credibility on the subsequent message. Neglected is the related question of the effect of message variables on sauce credibility. Furthermore, considering these questions mly one at a time precludes an investigation of the possible interdepemerce of the two factors and their conjunctive effect on influence acceptance. Antagonistic Arguments , Source Credibility , and Influence This section develops a rationale for the predicting of the effect Of antagonistic arguments on source credibility which in tun serves a mediating function in the influence process. The rationale is derived from the theoretical considerations and research evidence which have been disaissed earlier and which indicate that (1) source credibility has an effect on influence acceptance, (2) that message variables are capable of affecting sauce credibility, and (3) that messages which include antagonistic arguments are more persuasive than messages which do not include such arguments, when the receiver is initially opposed to the sauce's position or is already familiar with the antagonistic arguments. In the following discussion, it is assumed that the receiver initially is not in agreement with the posi- tion advocated by the sauce. While the comunication situation in 10 which the receiver is initially in agreement is of theoretic interest , it is outside the sc0pe of this study. When the receiver is initially Opposed to the saIrce's position it is likely that he will be aware of arguments upo1 which his own position is based. At least, he will be aware that the sauce's position is not the only one available. In this type of situation, failure on the part of the sauce to acknowledge these Opposing argu- ments is likely to result in his being evaluated low on the "safety" and/or "qualification" dimensions of credibility. For initially opposed individuals, credibility (differentially induced by messages which include or do not include antagonistic arguments) is advanced as a mediating factor in the influence process. Similarly, when an individual is familiar with arguments which are antagonistic to the sauce's position, failure to include such arguments in the message should result in the source being perceived as low in "safety" and "qualification". By including those arguments which are already familiar, the sauce is likely to be evaluated as fair or unbiased and knowledgeable . Althargh the antagonistic argu- ments, in. and of themselves, represent a negative vector, at the same time, these arguments shaild increase the credibility of the sauce. This is a positive vector. Since the antagonistic arguments are already familiar, their vector strength (information value) shaIld be low relative to the vector strength of the induced credibility. Thus, the resultant of the two vectors shaIld then be positive. This is advanced as an explanatim for the geater persuasiveness of the message which includes familiar antagonistic arguments as Opposed to the message 11 which does not irolude such argumelts. However, it is obvious that categorizing persuasive messages merely as "including" cr "not including" antagonistic messages is an oversimplificaticn of actual situations. The message may include varying amounts of antagonistic arguments familiar to the individual receiver, e. g. , the receiver may be familiar with Opposing arguments not presented in the message, or the message may include all such arguments . Furthermore , the message may include unfamiliar antago- nistic arguments. It can be argued that the persuasive consequence of including unfamiliar antagonistic arguments is quite different from that of including arguments which already are familiar. While the latter was suggested to have law information value , unfamiliar antagonistic arguments stould have high information value, and should strengthen rather than weaken the position the receiver held originally. Any advantage to the communicator in terms of a credibility increment led be me than offset by the high information value of the unfam- iliar arguments antagonistic to the sauce 's position. The relationship between an individual's familiarity with arguments that Oppose the sauce's position and the extent to which a persuasive message takes these antagonistic arguments into accamt is advanced in this study as a major determinant of the influence that the message will have. Specifically, familiar antagonistic argu- ments have low infatuation value and when presented in a message, increase influence acceptance becaise of the increased credibility which accrues. In comparison, unfamiliar antagonistic arguments have high information value which Opposes the advocated position and any 12 increase in credibility that might be associated with including such arguments does not offset their negative affect on influence. Hypotheses Presentation of the hypotheses requires prior explication of the nature of the variables included in the experiment. The hypotheses require that two variables be taken into accamt prior to the experi- mental manipulations. The first is the subject's prior evaluation of the message topic. From the rationale, it is apparent that these prior evaluations affect influence acceptance. That variable will be removed from the hypotheses by inducing a prior position fcr all subjects which is antagonistic to the position that will be advocated by the sauce. The secorl control variable is the prior information of the subject. That variable will be controlled by inducing a common level of prior information arong subjects; namely, each subject will be pro- vided one argument on each of two issues relevant to the message t0pic. These arguments will also be antagonistic to the position advocated by the sauce. Within this frame, there are two major independent variables. The first is the nature of the argumentative structure of the message. This variable has two values: (1) the message includes only arguments which support the sauce's position, and (2) the message "takes into accamt" arguments which Oppose the sauce's position in presenting the arguments which support the sauce's position. These values are referred to as (l) ccnstmctive message structue and (2) rebuttal message structure. 13 The second independent variable is the subject's familiarity with the issues and arguments within the message. For the construc- tive structure treatments, i.e., only arguments supporting the sauce's position, the familiarity variable has three values: (1) fewer issues than those with which the subject is familiar—and arguments supporting those issues which are antagonistic to arguments on which the subject had prior informatim; (2) value "1", plus the remaining issues with which the subject is familiar—and arguments supporting these remaining issues which are antagonistic to arguments on which the subject had prior information; and, (3) value "2" plus issue(s) on which the sub- ject had no prior information—and arguments supporting those issue(s) on which the subject also had no prior information. The rebuttal message structure, i.e., taking into account argu— ments which Oppose the sauce's positim, has three analogaIs values fcr the familiarity variable. The cnly difference is the inclusion of arguments supportive of the subject's prior position, i.e., antagonistic to the sauce's position. For the "fewer" case, this includes fewer source-antagonistic arguments than those on which the subject had prior information; for the "equal" case, this includes all sauce-antagonistic arguments on which the subject had prior information; for the ”more" case, this includes source-antagonistic arguments on which the subject had no prior information. This arrangement of the variables leads to six experimental treatments. For all treatments, the subject's prior evaluation of the message t0pic is tle sate, i.e., subjects are Opposed to the position the sauce will take. For all treatments, the subject's prior infome- 11} tion level is the same, i.e., subjects have knowledge of two issues, and have knowledge of an argument under each of these issues which is antagonistic to the position the source will take. The six types of message-treatment cells can be schematized as follows: Argumentative Structure Issue Familiarity Less anal re. Constructive A B C Rebuttal D E F The hypotheses are based on two major propositions concerning the way in which the independent variables operate. These are: l. The more information which the SQJI‘CG presents supporting his position, the greater will be the attitude change elicited. 2 . The more the source takes into account the arguments anta- gonistic to his position , the greater will be the attitude change elicited. With reSpect to the effects of information and "taking into accamt" on subjects' evaluations of the sauce, i.e. , source credibility, these propositions are sufficient to derive two basic hypotheses: Hypothesis I . Perceived source safe and qualification is positively relat to e extent to which the sauce takes into accamt issues and arguments opposed to his position, i.e., source credibility for rebuttal message structure will be greater than sauce credibility for constructive message structure. Hypothesis II. Perceived sauce safefl and qualification is positively related to the number of issues and arguments which the sauce provides, i.e. , sauce credibility for "more" information will be greater than for "equal" will be greater than for "less". 15 When we turn to "infOrmaticn" and antagonistic arguments "taken into account" and their separate effects on subjects' evaluations of the topic, i.e., attitude change, one of the two basic propositions needs to be modified. It still holds to argue that the source's in- fluence will be positively related to the amount of information which he presents supporting his position (Proposition 1); however, Proposi- tion 2 needs to be modified as fellows: 2a. The more the source takes into account the arguments antap gonistic to his position, the more influence he will have ... given that the source does not take.into account arguments which were previously unfaniliar to the receiver. This modification is necessitated by the opposition.of the two propositions fbr'that specific situation in which the source is pro- viding “more" infOrmation, but the inforuation.is antagonistic to his position (see message-treatment cell F). The first proposition would lead to the conclusion that the P treatment would be less effective, i.e., more information is being given that supports the receiver's position. The second proposition would lead to the conclusion that the P treatment would be more effective, i.e., more arguments antago- nistic to the source's position are being taken into account. The resultant of these two antagonistic vectors is not predictable. For that reason, cell F (“more" and "rebuttal") cannot be included in the general tests of the effects of amount of information and "taking into account" on subjects' acceptance of influence. Special hypotheses are needed to predict results in cell F. The general tests can be Specified in the fbllowing two hypotheses: 16 Hypothesis III. Given that no new issues are included in the message , . . . attitude change is positively re- lated to the extent to which the sauce takes into accamt issues and arguments opposed to his position, i.e., influence acceptance for cells (D +3) >(A +B). Hypothesis IV. Given that no new issues are included in the message ... attitude change is positively re- lated to the number of issues and arguments which the source provides, i.e. , attitude change (B +E) >(A +D). For messages which do not include any arguments antagonistic to the source' 5 position, i.e. , constructive argumentative structure, Hypothesis TV can be extended to include unfamiliar issues (Hypothesis IVa); however, this is stated as a separate hypothesis since a differ- ent statistical test needs to be used. Hypothesis IV a. Given constructive argumentative structure, a message that mesents more issues than those with which the receiver is familiar elicits ‘ geater attitude change than a message which presents only those issues with which the re- ceiver is familiar, i.e., attitude change C >B. The special test for cell F requires a comparison of the results in F with those in cell E. The F treatment includes all of E, but adds a new issue. Mame specifically, it provides an argument on that issue which supports the sauce's position and it provides an argument on that issue which is antagonistic to the source's position, i.e., it supports the receiver“ 3 prior position with an argument which was pre- viously unfamiliar to the receiver. Since the antagonistic argument is consistent with the position the receiver originally held, this argument should strengthen that original position more than the sauce-consistent argument weakens the 17 original position. The consequence of taking into accamt unfami- lia~ antagonistic arguments is tested in the following hypothesis: Hypothesis V. Given a rebuttal argumentative structure, a message that only provides information on all issues with which the receiver is familiar (cell B) will elicit more influence acceptance than will a message that also provides infor- mation on unfamiliar issues, i.e., influence acceptance for cell E. :>I-". It should be noted that a similar prediction cannot be made between cell F and cell D (a rebuttal message presenting information on fewer issues than those with which the receiver is already familiar). While cells E and F differ only in that F presents information an unfamiliar issues and B does not, i.e., both present the same familiar issues aid arguments for these issues, cells D and F differ both with respect to the familiar and the unfamiliar issues presented. The rationale that has been develoPed does not consider the difference in influence acceptance between not presenting all familiar antago- nistic arguments and presenting new antagonistic arguments in addi- tion to those which are already familiar. However, the difference between influence acceptance will be calculated for heuristic purposes. Hypotheses were derived from two major propositions which pre- dicated the effect of successive amamts of information presented and the effect of arguments antagonistic to the sauce's position "taken into accamt" on sauce credibility (Hypotheses I and II) and attimde change (Hypotheses III and IV). Both sets of hypotheses predicted a similar relationship between the message treatments and the dependent variables, i.e. , (l) a positive relationship between amamt of information presented and both sauce credibility ari atti- 18 tude change, and (2) higher source credibility and greater attitude change when antagonistic arguments are taken into account . Previalsly, a rationale was developed concerning tte function- ing of source credibility as a mediating factor in changing attitudes ad research evidence was provided which indicated that attitude change and source credibility are positively related. Thus, the re- moval of the effect of credibility from attitude change stould reduce the effect on attitude charge of increasing amounts of information and "taking into accamt". The general tests can be specified in the following two hypotteses: Hypothesis VI. Given that no new issues are included in the message ... if the relationship between source credibility and attitude change is eliminated, the positive relationship will be reduced be- tween attitude change and tte extent to which the sauce takes into account issues and argu— ments opposed to his position. Hypothesis VII. Given that no new issues are included in the message . . . if tte relationship between source credibility and attitude change is eliminated, the positive relationship will be reduced be- tween attitude charge and the number of issues ad arguments which the source provides . Hypothesis VIIa. Given a constructive argumentative structure ... if the relationship between source credibility ad attitude charge is eliminated, the positive relationship will be reduced between attitude change ad tte number of issues ad arguments which the source provides . The relative effect on influence of removing the effect of credibility from attitude change in cell F requires a special test comparing the results in cell F with those in cell E. Credibility is predicted to be greater in cell F than in cell B (Hypothesis II), 19 but attitude change is predicted to be greater in cell E than in cell F (Hypothesis V). Removal of the effect of credibility from influence acceptance in both cells should reduce influence acceptance in cell 1? more than in cell E. Hypothesis VIII. Given a rebuttal argumentative structLue .. . if the relationship between source credibility and attitude change is eliminated, tte difference in influence acceptance will be increased between (a) a message that provides information only on all issues with which the receiver is familiar (cell E) and (b) a message that also provides information on an unfamiliar issue (cell F). CHAPTER II METHOD This chapter describes the message t0pic and tle construction of messages; the independent, control, ad dependent variables; the sample; the experimental design and analysis techniques; and, He experimertal procedures employed in the study. The Message Topic "The current civil nuclear defense program" was selected as the message topic for two major reasons. First, it was felt that, although most people are aware of the concept "civil defense," they neither have strong Opinions about tle molear defense aspects of civil defense nor are tl'ey aware of tre majority of controversial but somewhat esoteric issues involved. Therefore, it seemed likely that opposing arguments about a number of nuclear defense issues carld be written which would be unfamiliar to the subjects. A pilot investigation utilizing a small number of graduate students ard student wives determined which of a series of such arguments were familiar or unfamiliar. Only those argumerts for which there was high agreement as to their unfamiliarity were considered for the experimental messages. Second, tte natLue of the experimental design requires that the two Opposing positions taken on the "current civil nuclear defense program" be implicit rebuttals of each other. Three separate civil nuclear defense issues were selected: 20 21 (1) Fallout Protection, (2) Firestorm Protection, ad (3) Post- attack Recovery. These three issues are sufficiently separable for arguments to be constructed for one issue that does not easily generalize to He other two . Message Segments For each of tle civil nuclear defense issues, fallart protection, firestorm protection, ad postattack recovery, one message segment was written which was clearly favorable to the "current civil nuclear defense program" and one message segment was written which was clearly unfavorable to the program. The three favorable message segments lead to the conclusion that the current program is appropriately conceived. The three unfavorable message segments lead to tte conclusion ttat the current program must be changed to reflect its purpose. An argument for 92?. issue leading to one conclusion has been termed a constructive sewn Since constructive segments make no reference to any Opposing arguments, they carld be termed "one-sided" but, because of the ambiguity associated with "message sidedress ," this terminology has been avoided. A brief description of tie six constructive segments is given in Figure l. The texts of each are included in Appendix A. Testing of several of tte theoretic hypotheses requires that antagonistic arguments also be taken into account in some of the persuasive messages. None of the constructive segments, sham in Figure 1, considers opposing arguments. However, since any favorable/unfavorable pair of the constructive segrents are implicit rebuttals of each otter, messages that consider opposing arguments can conveniently be derived from tle con- structive segments. This was accomplished by incorporating a weakened Issue 1. FALLDUT PROTECTION 1(a) Favorable to the current program. This segment asserts that fallart would be the major hazard ad evidence to support this position is provided. Specific fallait shelter operations are described ad their lifesaving potential is detailed. Issue 2. FIRESTORM PROTECTION 2 (a) Favorable This segment asserts that f ire- storms, althargh a concern, would _ngt_ present a widespread hazard ad evidence is provided to support this position. Appropriate steps which have been taken in areas where fire- storms might occur are described ad evidence to support the appropriatene 53 of these steps is provided. Issue 3. POSTA'ITACK RECOVERY 3 (a) Favorable This segment asserts that recovery programs are necessarily of lower Eiorilty than preattack programs an 3 position is defended. However, a number of important programs being develOped are disarssed and evidence to support their appropriateness is provided. 1(b) Unfavorable to the current program. This segment states tlat fallart would be a relatively minor hazard and evidence is provided to support this position. The shortcomian of tl'e current program are pointed out ad the possible consequences are described. 2 (b) Unfavorable This segment asserts that firestorms would present a ma'or hazard, would be widespread, evidenc—é— is provided to support this posit ion. Civil defense is criticized for an unrealistic program ad evideice is provided to support this criticism. The possible conse- quences of tl'e present shortcomings are described. 3 (b) Unfavorable This segment asserts that a full range of thoroughly prepared re- covery programs would be essential to survival ad evidence 18 pro- vided to support this position. Civil defense is criticized for not having developed these programs and the consequences of this failure are described. ***** Conclus ion : The current civil defense program is appropriately conceived. Conclus ion : The civil defense program must be changed to reflect its purpose. Figure l. Constructive Segments for Each Issue 23 paraphrase of the argument from one segment into the Opposing segment of any pair. The combination of constructive segments in this manner has been termed a rebuttal ‘seaent. A rebuttal segment could be termed a type of "two-sided" segment but, once again, tle ambiguous terminology has been avoided. The texts of the six rebuttal segments are given in Appendix B. The following are examples of rebuttal sewts. Issue 1. FALLDUT PROTECTION l(d) Unfavorable to the current program l(c) Favorable to tte otuuent program This segment presents a weakened This segrent presents a weakened paraphrase of the antagonistic argument presented in ca“:- structive segment l(b) (FALLDUT PROTECTION—meavorable to the c1u'rent program). Transitions were made to He rebuttal of this argument which was con- structive segment 1(a) (FALLUJT PROTECTICN-é-favorable to the current program). 1(a) attempted to destroy tle argument of l(b) in general , rather than point-by-point. paraphrase of the antagonistic argument presented in con- structive segment 1(a) (FALIDUI‘ PROTECTION «favorable to the current program). Transitions were made to the rebuttal of this argument which was con- structive segrent l(b) (FALLCXJT PROTECTION--m1favorable to the aurent program). l(b) attempted to destroy the argument of 1(a) in general, rather than pOin‘t-by-pojll‘t. ***** Conclusion: The cment civil defense program is appropriately conceived. Conclusion : The civil defense program must be changed to reflect its purpose. The fair other rebuttal segments, two for issue 2 ad two for issue 3 , were constructed in the same mamer as FAIIDUT PROTECTION rebuttal segments . 2L» Irdependent Variables A test of tle hypotheses requires that the experimental messages vary on two dimensions; (1) the number of issues presented relative to the receiver's prior familiarity with issues ad (2) the presence or absence in tle message of arguments antagonistic to tle communicator's position (i.e., rebuttal versus constructive). (1) Relative Issue Familiarity: This variable is the number of issues presented in tle message relative to an irdividual's prior familiar- ity with issues. Operationalizing tle variable required tl'at tl'e Specific information held by the irdividual be known prior to his being presented the experimental message. Prior familiarity was manipulated in the following mamer. Firet, it was assumed that the three issues were themselves unfamiliar or only minimally familiar to the subjects ad that tre specific argument associated with each issue was unfamiliar prior to any experimental manipulation. Then, at an experimental session one week before the presentation of tle treatment messages, each subject received one of tle messages designed to manipulate information level. The three favorable constructive segments previously described were combined two at a time in all possible ways to make three messages. The same procedtue was followed for unfavorable construction segmemts. This produced a total of six familiarity induction messages . 25 Favorable Unfavorable (l) Fallant ad Firestorm Protection (1+) Fallout ad Firestorm Protection segments [1(a) + 2(a)]* [l(b) + 2(b) 1 (2) Fallart and Postattack Recovery (5) Fallant ad Postattack Recovery [1(a) + 3(a)] [l(b) + 3(b)] (3) Firestorms ad Postattack Recovery (6) Firestorms ad Postattack Recovery [2(a) + 3(a)] [2(b) 4- 3(b) 1 *Coded sane as Figure l. Figure 2. The Six Familiarity Induction Messages Prior familiarity with issues that would be treated subsequently in the experimental messages was rm established. The variable relative Egg familiarity can take three values. Each of tte T2 experimental messages presented one of the following: Value 1. -- flea issues than those with which the receiver was previously familiar. (Operationally , arguments concerning only 213 issue were presented.) Value 2 . -- the game issues with which the receiver was previously familiar. (Arguments concerning only 1,1,2 issues were presented.) Value 3. -— the gage; issues with which the receiver was familiar plus an unfamiliar issue. (Arguments concerning three issues were presented.) 26 Since all of tte familiarity induction messages provided arguments on two and only two sub-t0pics, tie operationalization of the message variable "relative issue familiarity" correSponds to the classification levels. (2) Constructive versus rebuttal argumentative structure: All of the experimental messages (those presented at T2) were e itrep constructive (containing no arguments antagonistic to tre communicator's position) or rebuttal messages (containing antagonistic arguments). The constructive messages were all possible combinations of constructive segments ad the rebuttal messages were all possible combinations of rebuttal segnents. Constructive segments ard rebuttal segments have previansly been described. Since the two indeperdent variables were completely crossed, tie number of segments in both the constructive messages ad the rebuttal messages was , for each message , deperdent upon tlne other variable message condition "relative issue familiarity." Control Variables The control variables employed can be classified in three categories: (a) a test of the adequacy of the manipulation of the source's position on the t0pic; (b) those used to accamt for any differences in persuasive strengths of messages; and (c) social controls used to mask the intent of the experiment. (a) Perceived sauce favorableness to experimental concepts. To determine if the position advocated in the message was perceived as interded, subjects at both experimental sessions were asked to respond 27 to the question "Inn your opinion, would yan say that the person who wrote this message was favorable or unfavorable to 'the cnuuent civil nuclear defense program. '" Opinions were recorded on tl'e single 7-point semantic differential scale "favorable-unfavorable." (B) Persuasive strength of messages (l) Favorable versus unfaiVorable messages. Possible persuasive differences between favorable and unfavorable messages were controlled. Half of the subjects received a familiarity induction which contained arguments leading to the conclusion that was favorable to the current program. At the secord session these subjects received one of the message treatments which led to the conclusion that was unfavorable to de aurent civil defense program. This procedure was reversed for the other half of the subjects. (2) Between sub-topic segments within favorable or unfavorable messages. To control for possible persuasive differences between issue segments within favorable or unfavorable messages, all possible segment combinations for any treatment condition were counter-balanced in the experimental design. (c) Social Controls. To mask the intent of the experiment, items were included in the questionnaires at both experimental sessions which were otherwise irrelevant to tte study. These included evaluation of the clarity of the messages and tie novelty of the information provided. The actual questions asked and the scales used are included in the two questionnaires which are pre- sented in Apperdix C ard Appendix D. Dependent Variables (l) Attitude Change on the emrimental concept. 28 For those subjects who received a message favorable to the experimental concept at T2, the attitnde change score was their T2 evaluation minus their Tl evaluation. For subjects wro received an unfavorable message at T2 , the attitude change score was their Tl minus T2 evaluation. Change scores were divided by 5, the number of scales in the instrument. Mean change scores could range from -6 (maximum possible "boonerang") to +6 (maxirmmm cl'ange in the direction advocated by tre T2 message). At both experimental sessions subjects irdicated their own attitudes toward tte "current civil nuclear defense program" by ratings on the same five 7-point semantic differential scales. These scales are those suggested by Osgood, Suci and Tannenbaum (1957) to measure the evaluative dimension. The scales used were good-bad, wise-foolish, valuable- worthless, fair-unfair, ad honest-dishonest. Scale ends were randomly reversed to minimize response set. Responses were coded l (unfavorable) through 7 (favorable) and summed over the five scales. To mninimize response consistency from T1 to T2 the five instrument scales were embedded in five additional scales at T2, The additional scales were also selected from those suggested by Osgood, Suci and Tannenbaum (1957) to measure the evaluative dimension. These filler scales were strong-weak, successful-unsuccessful, important—unimportant, useful-useless and appropriate-inapprOpriate . (2) Credibility of the Message Sanrce (time-two). Subjects resporded to the question "How would you rate the goon who wrote this passage as a source of information on the issue 'the current civil nuclear defense program' '2" Evaluations were indicated on twelve 7-point semantic differential scales measuring the source's perceived 29 Safety, Qualification and Dynamisrm. The four scales used for each dimension were selected from a list provided by Berlo, lemert ad Mertz (1965). Scales were presented in mixed order with scale ends randomly reversed. Summary ratings for each dimension were computed by summing across the four scales ad dividing by four. The Safety scales were safe-dangerous, just-unjust, honest-dishonest and kind-cruel; the Qualification scales were qualified-unqualified , infomed—uninfomed, experienced-inexperienced and skilled-unskilled; the Dmamism scales were bold-timid , emphatic-he sitant, active-passive and energetic-tired. Subjects The study used 12 intact g'oups of undergraduate stndents enrolled iun a Spring course in the College of Business at Michigan State University. Each section had approximately 30 students, most of whom were junior or senior males. The experimental sessions were conducted during regularly scheduled classroom periods . guanimental Design This study was designed to compare the pasuasiveness of constructive ad rebuttal argumentative structures which present less, the same, or more issues ad arguments than those with which the receiver is already familiar. The study was also designed to compare the perceived source credibility induced by messages which varied on tlese two dimensions, ad to determine the mediating effect of the irduced credibility on message persuasiveness. Basically the study is a two variable 3 x 2 factorial design (ignoring tle treatment conditions which were employed to control for any 30 differential effectiveness attributable to message segments). The irdependent message variables were (1) constructive versus rebuttal argumentative structure, and (2) fewer versus same versus more issues presented relative to tle issues with which the individual was already familiar. The dependent variables were (1) attitude change on the experimental concept, and (2) perceived safe}: and qualification of the message source . Relative Issue Familiarity Fewer Same More I T I l Constructive ' ' ' ' . A B C Argumentative ' ' ' ' Structure ' ' 1 ' Rebuttal I D : E : F : Figure 3. The Basic Experimental Design* This experimental design required that the individuals' familiarity with specific issues and arguments be ascertained prior to presenting the various message treatments . The decision to manipulate *This is the basic experimental design. It presumes that (1) order of message presentation (favorable message first versus unfavorable message first) does not affect the criterion variable, and (2) the individual message segments are equally persuasive. The complete experimental design, without collapsing either order of message presentation or message segment counterbalanced combinations , is given in Appendix E. 31 familiarity rather than to determine familiarity through pretesting pro- cedures was based upon a number of considerations. The pretesting procedure would have resulted in subjects self-selecting themselves into treatment conditions on the basis of their prior familiarity with the content of tre subsequent message. This muld have left uncontrolled any subject variables associated with a particular information level. That would have obscured the interpretation of any findings. Also, since the rationale of the study concerns the relationship between specific known arguments and their inclusion in a message, pretesting to determine specific information would have created sensitization problems. At a familiarity induction session (Tl), subjects received one of the six messages previously discussed and represented in Figure 2. Since there were six messages and twelve intact groups of subjects, each message was randomly assigned to two groups. All subjects in a particqu group received tle same message at this session. Therefore, at the in- duction session, all goups were faniliarized with arguments concerning one of the possible combinations of two issues. One week later, all subjects read one of tl'e six treatment messages in the basic design. This message led to a conclusion which opposed that of the message which was read at tre familiarity induction session. The appropriate versions for each of tle six message treatments were radcmly ‘ distributed within intact groups . Thus , an intact group did not con- stitute a cell in the design, and any biasing effects that might be associated with any of tle classroom goups were evenly distributed across all of the 3 x 2 message treatments. 32 Data Analysis Following the second experimental session T1 and T2 responses were matcl'ed on the basis of name and student number. Some attrition occurred between the two experimental sessions but the experimental design is such that different attrition rates across cells could only be radon. A total of 15 paired questionnaires were randomly discarded from the appropriate cells to equalize cell n's at 20, tl'e cell size of tre smallest cell (cell n = 20 with T1/T2 message order not collapsed). Analyses were performed on the data provided by 240 subjects. Hypotheses I through V are tested through various analyses of variance and t-tests comparing mean credibility scores and mean attitude change scores among the six message-treatment groups. Hypotheses VI through VIII make predictions about what the relationships between the message treat- ment goups ad attitude change would she with the effect of source credibility removed from tl'e attitude change scoes (i.e., if the per- ceived credibility of the source was the same for all groups). Tb test Hypotheses VI through VIII requires correlational techniques. Specifically, tte hypotheses are tested by comparing the difference be- tween the zero-order correlation Fab ("a" is argumentative structure or issue familiarity, as tle case may be, and "b" is attitude change) and tie part correlation ra(b.c) ("c", source credibility, is removed from attitude change), Hypotheses VI and VII predict rah ra(b.c) and Hypothesis VIII predicts the reverse. Argumentative structure is dichotomous and , while issue familiarity has three values, the hypotheses involving issue familiarity are 33 restricted by the rationale to only two of these values for any single hypothesis. Thus, the appropriate r's are point biserial. It is not possible, however, to test the significance of the difference between a zero-order and part correlation computed on tre same sample because the two are not independent. Accordingly, each treatment group was randomly Split into two analysis groups of equal size. A zero- order r is computed on one group and a part r on the other. The significance of the difference between the two independent r's is then computed to test each of the three hypotheses. Experimental Procedures Familiarity induction session (I‘ll: The following procedures were followed in sequence at this initial session. (1) The experimenter entered the classroom, was introduced by the regular instructor, solicited the cooperation of the students in a "Communication Survey" under the auspices of the College of Communication Arts, and distributed tre questionnaires. ( 2) An introduction page stated that the survey was one of a series concerned with the way college students react to various kinds of infor- mation on public issues. It also stated that the students should read carefully because they would be required to make a number of evaluations of the passage they were about to read. (3) Subjects then read the message desiged to familiarize them with Specific information and arguments about tte experimental concept. 34 (it) Subjects tten read a one page description of the way to use the semantic differential scales. (5) Subjects tlen resporded to the directive "Before you rate the passage, we would like your own evaluation of 'the current civil nuclear defense program. '" (6) Subjects then rated the source of the passage, gave their Opinion of the writer' 5 position on the experiumental concept , indicated their familiarity with the content of tre passage, ad rated the clarity of the passage. ' (7) When everyone was finisled, tle questionnaires were collected, the experimenter thanked the subjects for their cooperation, and left the room. No mention was made of a second session ad the various instructors were cautioned not to provide any clues that tfere would be a second session. Message manipulation session (T2); The following procedures were followed at this session, which took place one week after Tl° (l) A different experimenter entered tie classroom ad enlisted tie cooperation of the students in a "Communication Survey." He stated that a large number of students had taken part in an earlier pilot study ad that he was aware that some of those present had taken part, but that that was all right. The questionnaires were ten radonly distributed. (2) An attempt was made to make tree general appearance of the questionnaire different from T1: however, it was felt that any furtler attempt to dissassociate the two not only would be unsuccessful, but would 35 antagonize the subjects by insulting their intelligence. The introductory page elaborated on the experimenter' S opening remarks . It stated that the four public issues used in the pilot study had been (1) legislative Reapportionment, (2) The Office of Economic Opportunity, (3) The Civil Defense Program, and, (1+) Medicare. Subjects were informed that The Civil Defense Program had been selected because it met the requirements of the present study. A note at tle bottom of tl’e introduction page informed the student that, if he participated in the pilot study and if the t0pic happened to be Civil Defense, it did not matter. (3) Subjects then read one of tre versions of one of the treatment messages. (H) The procedures for tie remainder of tle questionnaire were the same as in the T1 questionnaire, CHAPTER III RESULTS T1 Attitudes Toward the Civil Defense Topic The experimental manipulations require three issues relevant to the tOpic of the present civil defense policy. For each issue, there is one argument supporting the policy and one Opposing it. In order to vary the subjects' familiarity at T2 with issues and arguments antagonistic to the source's position, each subject was given two arguments on one or the otlner Side Of the tOpic at T . This produced Six message-treatment groups. For control purposes, tle various arguments on one or the otter Side Of the tOpic must be equally persuasive; i.e. , there must be no Significant differences in mean attitude scores among the three groups which received favorable information, and there must be no significant differences among the three groups which received negative information. The hypotheses are limited to situations in which receivers are Opposed to the position of tie source. For control purposes, at T1 positive attitudes had to be induced for lnalf Of tte subjects ad negative attitudes had to be induced for the other half. Table 1 presents the mean T1 attitude scores for each Of the Six message-treatment groups, and the results Of an analysis, Of variance ad t-tests among the means . The pro-topic and anti-topic means are significantly different overall; furthermore, the mean difference between pro and con groups is significant for each Of tl'e three message segment treatments. 36 37 Finally, there are no significant differences among the.means within the pro or con message treatment groups, nor is there an interaction between pro-anti and segments. The two control criteria.are satisfied. Table l. message-treatment groups, and analyses Of differences among the means Mean T1 attitudes toward civil defénse fer the Six Nbssage Segments Fallout + Firestorms Fallout + Recovery Firestorms + Recovery Overall Source Of variance Favorability Segments Favorability X Segments Error '.Message Favorability Favorable Unfavorable Mean (N) than (N) to WI) 3276' T57) 5.u6 (”3) “.20 (#5) 5.38 (36) 4.37 (38) 5.H3 (120) H.15 (120) d.f. _£§i_ l 2.u5 2 .03 2 .OHS 234 .031 D 1757 1.26 1.01 1.28 OI§7 <§%&fi. 5.12 «< .01 4.11 '< .01 L ‘< .01 n.s. n.s. 38 The Attitudinal Position Taken in the Message At both T1. (the argument familiarity manipulation session) ad T2 (the main experimental session), subjects were asked to state wrether the message was favorable or unfavorable toward the present civil defense policy. This deck was made to insure trat tl'e messages were perceived as interded. Of tre 2H0 subjects, 226 or 94% perceived the T1 message as interded and 215 or 9096 perceived the T2 message as interded. Tests Of the Theoretic Hypotheses Hypothesis I. Perceived source safe aund qualification is positively relat to extent to which tie source takes into account issues ad arguments Opposed to his position, i.e. , source credibility for rebuttal message structure will be greater than source credibility for constructive message structure. Hypothesis II. Perceived source safeg and qualification is positively related to tle number Of issues and arguments which the source provides, i.e. , source credibility for "more" information will be greater than for "equal" will be geater than for "less". Hypothesis I asserts that rebuttal messages will result in evaluations of the source's safety ad qualification which are higter than the evaluations resulting from constructive messages. Hypothesis II states that the evaluations on trese two dimensions Of source credibility will increase as the number of issues and arguments presented increases. Separate two—way analysis Of variance were performed to test these hypotheses with respect to first safety evaluations and then to qualification evaluations . * *T‘hree-way analyses were performed first to test whetl'er tlne control variable of "message favorability" (i.e. , a message favorable to the tOpic vs. a message unfavorable to the topic) had any effect. It did not, ad favorable-unfavorable groups were combined (no F including favorability even approacl'ed Significance). Safety evaluations. 39 Group means on safety, and tre analysis of the effects of the message treatments on safety evaluations are con- tained in Table 2. Neither Of the hypothesized main effects were significant. There is no relationship between message treatments ad safety evaluations; i.e., Hypotheses I ad II are not supported for safety evaluations . Table 2. Mean safety evaluations for the six message-treatment groups, ad the analysis of variance anong tre means (n = 40) Issue Familiarity less Same Pbre Overall Argumentative Structure Constructive (A) 4.65 (B) 4.83 (C) 4.96 4.81 Rebuttal (D) 4.88 (E) 4.77 (F) 4.70 4.78 Overall 4.765 4.80 4.83 4.80 Source of Variance _d_'_f_ M_.S_._ _F_ _L Argumentative Structure 1 .051 .06 n.s. Issue Familiarity 2 .086 .10 n.s. Structure X Issues 2 1.169 1.37 n.s. Enror 234 .856 40 Group means and the analysis of the effects of the message treatnents on source qualification evaluations are summarized in Table 3*. Again the predicted main effects were not significant, indicating that neither message variable produced differences in tle evaluations Of source qualification. Hypotheses I and II are not supported for qualification evaluations . Table 3. Mean qualification evaluation for the six message- treatment groups, and the analysis Of variance among the means. Issue Familiarity Argumentative Structure Less Same _szna Overall Constructive (A) 4.69 (B) 4.99 (C) 5.05 4.91 Rebuttal (D) 5.15 (E) 5.16 (F) 5.15 5.15 Overall 4.92 5.075 5.10 5.03 Source Of Variance __c_1_§_ M.S. _F_ _p__ Argumentative Structure 1 3.626 2.80 n.s. Issue Familiarity 2 .772 .60 n.s. Structure X Issue 2 .732 .57 n.s. Error 234 1.294 *Means are based on two rather than four qualification scales. The scales "Skilled-unskilled" ad "experienced-inexperienced" were inadvertently omitted from the questionnaire . 41 Hypotheses III, TV, ad V are concerned with tl'e effect Of various treatment conditions on attitude change. The Six message-treatment group means needed to test Hypotheses III, IV, and V are presented in Table 4. Table 4. Mean attitude orange Scores on the experimental concept for the six message-Ueatment groups (cell n = 40) Issue Familiarity Argumentative Structure Less _Sam__na 1213.. Overall Constructive (A) .135 (B) .635 (C) .615 .462 Rebuttal (D) .590 (1:31.010 (F) .460 .687 Overall .363 .823 .538 .575 Hypothesis III. Given that no new issues are included in the message, ... attitude change is positively re- lated to the extent to which the source takes into account issues and arguments Opposed to his position, i.e., influence acceptance for cells (D + E)>(A + B). Hypothesis IV. Given that no new issues are included in tte message . . . attitude change is positively re- lated to the number Of issues ad arguments which the source provides, i.e. , attitude change (B+ E)> (A 4' D). Hypotheses III ad IV are restricted to tie four experimental treatments trat do not include new issues or new antagonistic arguments (cells A,B,D, and E). Hypottesis III asserts that rebuttal messages will elicit greater attitude change than constructive messages will, given no that no new issues are included. Hypothesis IV states that the amount of attitude change will increase as the number of issues presented in- creases, again provided that no new issues are included. A two-way analysis of variance can be used to test both Of these hypotleses Simultaneously (see Table 5)*. The two hypotteses are tested by the significance of the two main effects. Both are significant, and the Table 5. Analysis of variance and simple—effect analyses of mean attitude change scores for four Of tl'e message-treatment groups (cells C and F elimninated). A. Analysis Of Variance Summary Source Of Variance £1; M.S. _F_ _p_ Argumentative Structure 1 172.22 9.97 <: .01 Issue Familiarity 1 211.10 12.22 < .01 Structure X Issues 1 2.11 - n.s. Error 156 17.28 B. Cell Means and T—Tests for Simple Effects Differennces Argumentative Issue Familiarity _ Structure Less Sane D i Constructive (A) .135 (B) .635 .50 2.68* Rebuttal (D) .590 (E) 1.010 .42 2.26* 5 .455 .375 t 2.450* 2.02* *t.975 = 1.98, df = 155 —*Again, the data first were analyzed to test for differences among the two values of the control variable, message favorability. It had no effect and groups were combined (no F including favorability even approached Significance). 43 mean differences are in the hypothesized direction. There is no significant interaction between number of issues and constructive- rebuttal structure. Simple effects analyses (see table 5) reveal tlat the rebuttal structure is more effective than the constructive structure for both levels Of information, and that more information is more effective than less information for both constructive and rebuttal structures. Hypotheses III and IV are confirmed. Hypothesis IVa is an extension Of Hypothesis IV for constructive messages only. Hypothesis IV states that attitude change will increase as the number of issues and arguments presented increases. It was partially confirmed under the test Of Hypothesis IV when it was found that attitde change was Significantly higher for "equal" information than it was for "less." Hypothesis IVa states that attitude Orange should be higher for the "more" information treatment than it is for "equal." The data do not support that hypotl'esis. In fact, the mean for "more" information (.615) is slightly but not significantly less than it is for "equal" information (.635). Hypothesis We is not confirmed, ad, therefore, the confirmation Of Hypothesis IV is lessened tO that extent. Hypothesis V. Given a rebuttal argumentative structure, a message that only provides information on all issues with which the receiver is familiar (cell B) will elicit more influence acceptance than will a message that also provides infor- mation on unfamiliar issues, i.e. , influence acceptance for cell B>F. This hypothesis states that a rebuttal message that only takes into account all antagonistic arguments with which the receiver is famniliar will elicit more attitude change than will a rebuttal message which also presents 44 antagonistic arguments with which the receiver was not previously familiar. This hypothesis was tested by a t-test comparing the mean attitude change scores for those two cells (E and F). The mean for the "all familiar" group was 1.010 and for the "also unfamiliar" group was .460. The difference between these two means is statistically Significant (t = 2.48, df=78, p< .05). Hypothesis V is confirmed. The inclusion of an unfamiliar argument in addition to all familiar arguments antagonistic to the source' S position reduced the attitude change elicited by the message. For heuristic purposes, differences in the mean attitude change scores were compared between the rebuttal treatment group which received arguments on fewer issues tlan those with which they were familiar (mean = .590) ad the rebuttal treatment group which received arguments on more issues than those with which they were familiar (mean = .460). The two means did not differ significantly (t = .13). Hypothesis VI. Given that no new issues are included in the message ... if the relationship betneen source credibility ad attitude cl'ange is eliminated, tle positive relationship will be reduced be- tween attitude change and the extent to which tle source takes into account issues and argu- ments Opposed to his position. Hypothesis VII. Given that no new issues are included in tle message ... if the relationship between source credibility and attitude change is eliminated, the positive relationship will be reduced be- tween attitude change and the number Of issues and arguments which the source provides. 45 Hypothesis VIIa. Given a constructive argumentative structure ... if the relationShip between source credibility and attitude change is eliminated, the positive relationship will be reduced between attitude change and the number of issues and arguments which the source provides. Hypotheses VI, VII, VIIa are concerned with tle reductions in to relationship between the independent variables and attitude change that will occur when the relationship between.source credibility and attitude change is eliminated. The lack of support fOr Hypotheses I and II obviate any statistical tests Of these hypotheses. Since there is nothing other than a chance relationship between the message-treatment variables and source evaluations, "removal" of the relationship between credibility and attitude change can not reduce the relationships between the message-treatment variables and attitude change. Thus, Hypotheses VI, VII, VIIa are not supported. Hypothesis VIII. Given a rebuttal argumentative structure ... if the relationship between source credibility and attitude change is eliminated, the difference in influence acceptance will be increased between (a) a message that provides information only on all issues with which the receiver is familiar (cell E) and (b) a message that also provides infOrmation on an unfamiliar issue (cell F). This hypothesis states that eliminating the effect of credibility on attitude change will reduce attitude change less if only all familiar antagonistic arguments are included in a.message (cell E) than if uns familiar arguments are also included (cell F), i.e., source credibility is predicted to be higher in F than B (Hypothesis V). 46 The difference in tie zero-order correlation between issues and attitude change (13‘ = .259) and the part correlation with the affect of credibility on attitude change eliminated (raw c) = .257, calculated from rab = .266; rac = .001; rbc = .102) is so slight that a test Of significance was not performed. Thus, Hypottesis VIII was not supported . CHAPTER IV SUMMAIQv AND DISCUSSION This study investigated three types of hypotheses. These hypotheses concerned tl'e effects of variation (a) in the argumentative structure Of the message ad (b) the prior familiarity receivers had for the issues presented on (1) two dimenSions Of the credibility Of the message source-safety and qualification, (2) attitude channge, and (3) attitute change after renoval of the effect Of credibility on attitude change. The first two hypotleses predicted that the credibility Of tle source would increase as the number of topic-related issues presented in tle message increased, and would increase when arguments antagonistic to the source's position were taken into account. Neither Of the predicted relationships for either dimension of credibility was supported. T‘l'ere was no evidence that the message variables produced any variability in source evaluation. The second type of hypotheses concerned the effects of tle message variables On attitude change . Wlen messages presented only issues with which the receiver was already familiar, presenting all familiar issues elicited more attitude change than presenting only some Of tle issueS.* *Throughout the discussion it wl'ould be recalled that "sore" or "fewer" issues refers to one Of tle issues the receiver was exposed to at time-one. "All" familiar issues refers to t1"e same to issues presented at time-one. TTe expression "also unfamiliar issues" refers to the sane two issues presented at time—one, plus tle third issue which was not then presented . 47 48 This held for both constructive and rebuttal argumentative structures , as predicted. Also, when just familiar issues were presented, rebuttal argu- mentative structure elicited more attitude change tlan constructive struc- ture. Again, as predicted, this leld when all familiar issues were pre- sented or when less tlan all familiar issues were presented. Predictions were also made concerning the effect on attitude change of presentinng unfamiliar issues. If the message only presented arguments which are consistent with tle source's advocated position, i.e., constnuc- tive argumentative structure, Hypothesis IVa stated tlat also presenting unfamiliar issues would increase attitude change . This was the only attitude change hypothesis that was not supported. If all familiar issues were presented in the message, the addition Of an unfamiliar issue did not further increase attitude clange-when the message contained only source-consistent arguments. Tle two message types were equally persua- sive and both were more persuasive than the constructive message that presented only some of the issues known to tle receiver. T‘le prediction of Hypothesis V was Opposite to that of IVa. Hypothesis V predicted that when antagonistic arguments were included as well as consistent arguments, i.e., rebuttal argumentative structure, attitde change would decrease if unfamiliar issues were also presented (in addition to all familiar issues). The hypothesis was supported. The remaining hypotkeses predicted the effect of the message variables on attitude change with tie effect Of credibility removed from attitude change. Hypotheses VI, VII and VIIa predicted that tre removal of tie effect of source credibility from attitude change would reduce the difference in attitude change between selected message 49 conditions . Since there was no relationship between message teatments and perceived source crebility, it was not statistically possible for Hypotleses VI, VII, and VIIa to be supported; therefore, they were not tested. Hypothesis VIII predicted that the removal Of the effect of source credibility from attitude change would increase the difference in attitude change, again between selected message conditions . This hypothesis was tested but was not supported . Discussion With constructive argumentative structure, also presenting an unfamiliar issue did not elicit an attitude clange increment above presenting only all familiar issues. DeSpite the lack of support for the credibility hypotheses , an attitude change increment was expected on the grounds that the more arguments presented which are consistent with the advocated position, tle greater the persuasive impact Of the message ad the greater the attitude change. It could be argued that the lack Of support for this hypotlesis is readily explainnable for messages that were unfavorable to the 'tOpic . Introducing an unfamiliar‘issue, e.g., firestorm protection, ad proceeding to argue that tle present firestorm protection is inadequate could be expected to lave little effect on attitude change, since the receiver was not even previously aware “that a_nx firestorm protection was in existence . However, when the message was favorable to tie t0pic, this reasoning is not as conpel'ling . Moreover, the evidence does not provide any support for this interpretation. T'l'e favorability of the message toward the topic did not lnave an effect on attitude clannge, i.e., the mean for the group that received tle version favorable to the t0pic was not different from 50 the mean Of the group that received the unfavorable version. An adequate. interpretation Of the failuure Of Hypothesis IVa is not apparent . It would be unwise, however, to generalize this finding to messages wtere the ratio of unfamiliar to familiar issues presented is increased. In this study, one unfamiliar and two familiar issues were presented. It is conceivable that if more unfamiliar issues had been presented, the hypothesis would have been supported . This is an empirical question and needs furtl'er research. Tl'e predicted relationships between attitude change ad relative issue familiarity were supported wlen tle messages were Of rebuttal argu- mentative structure . When the message took into account all familiar arguments antagonistic to the advocated position, maximum attitude change resulted. Failure to consider some of tl'e antagonistic arguments known to tie receiver, or considerinng more antagonistic arguments than those with which 1e was previously aware, significantly reduced tle persuasive- ness Of the message. These results are interpreted in termus of tle predicted difference in the negative effect Of famniliar and unfamiliar antagonistic arguments- relative to tre positive persuasive effect Of the consistent arguments, all Of which were unfamiliar. The lack of support for tle credibility hypotleses is of minor relevance to this interpretation . Credibility was expected only to change the differences in attitude change resulting from the message treatments, not to eliminate the differences between them. T'le general support for the attitude change hypotheses suggests that relative familiarity with message information has predictive utility and deserves furtler investigation. The present study, lowever, is of 51 limited generalizability. There is no assurance that the general hypotheses. would have been supported had the ratios of "issues presented" to "issues known" been different, i.e., other than l:-2, 2:2 and 3:2. A number of other combinations should be tested before a great deal of confidence can be placed in the present findings. In addition to varying the ratio , varying the absolute number of issues presented in the message might also produce differing results. The maximum number of issues presented in this study was three and it can be argued that this kept the messages relatively simple and comprehensible. The addition of more issues would have increased the complexity of the message and the amount of information required to be processed. Conceivably "number of issues" might produce different results if the range of arguments was less restricted. This study also neglected the ccmmmmication situation where a message takes into account unfamiliar issues as well as familiar issues but not all familiar issues. What are the attitudinal effects of neglecting some issues which are familiar to the receiver while at the same time including issues with which he was previously unfamiliar? Do the neglected issues take on greater or lesser significance for the receiver? Since the communicator is familiar with issues not previously known to the receiver, does the receiver assume that omitted issues are actually known to the communicator but are mfimportant? These questions are important research that should be investigated . The findings concerning familiar and unfamiliar antagonistic arguments presented in a message clearly suggest the predictive advantage of defining antagonistic arguments in terms of their prior familiarity 52 to tie receiver. Pram the results obtained in this study, it is apparent that recommendations with respect to presenting antagonistic arguments (two-sided) or rot presenting them (one-sided) must be dependent on the receiver' s familiarity with the antagonistic arguments. Hypothesis III predicted that rebuttal argumentative structure would elicit more attitude change than constructive structure , provided that rone of the antagonistic arguments in rebuttal messages were previously unfamiliar to the receiver . Although the hypothesis was supported, the rationale from which the hypothesis was develoPed states that the difference in attitude change is a function of differences in the message induced credibility of the two sources. The differentially induced credibility was predicted to mediate differences in attitude change. Given the failme of the messages to differentially affect source credibility, the interpretation of the attitude change findings is obscured. One possible explanation for the attitude change findings involves tie rotion of "rehearsal" advanced by Hovland and his associates. Pre- senting only arguments which are consistent with the advocated position may stimulate the receiver to "relearse" his own position, thereby reducing his attention to the source's arguments. Since, in the present study, the receivers were opposed to the source' 5 position and were familiar with arguments antagonistic to the source's position, the "rehearsal hypotlesis" must be considered as a possible interpretation of these findings. A second kind of interpretation is analogous to "associative facili- tation" in learning theory. When two stimuli are presented in order and the first increases the extent to which the second is effectively learned, 53 this is termed "associative facilitation" . In rebuttal argumentative structure the antagonistic argument is presented before the constructive argument and may clarify the constructive assertion to some extent, tlereby increasing the persuasiveness of the rebuttal argument over the constructive argument . A third interpretation involves a consideration of the source credibility measuring instrument . It is possible that source evaluations were differentially affected by the messages but that the instrument did not detect the differences. The scales used to measure the dimensions of credibility were ctosen from a study which factor analyzed the ratings on known sources (Berlo, Lemert 8 Mertz, 1965). In the present study the communicator was not well known. The source was "loom" only through the message. Everything about the various messages which could produce variability in source evaluations was leld constant except for number and familiarity of issues, and whether or not antagonistic arguments were included . While the rating scales which were used for each dimension are capable of differentiating well known sources (e.g., John F. Kennedy, Memen Williams, and Fidel Castro), they may rot be sufficiently sensitive to differentiate between much more subtle clues. Perhaps the best example of this argument is the safety scale "kind-cruel". In retrosPect, there is little reason to expect that tie two message variables in the present study would significantly effect variability in source ratings on this scale. However, this is an extreme aample and the same explanation for the scale "fair-unfair", also used to measure the safety of tle source, is much less tenable. Therefore, it seems unlikely that this consideration 51+ is capable of accounting for the complete failure of the messages to differentiate source credibility . More likely, all or some combination of the above considerations produce the persuasive differences in the two agrumentative structures . Also, factors not yet considered may also have contributed to the differences . fur'tler work could profitably be directed at investigating the functioning of Opposing arguments in a persuasive message. At the present time there is inadequate data to cl'oose between "receiver rehearsal" and "source or message perceptions" as alternative exPlanations for the data . A resolution of this issue will require different design and measurement research strategies . Typically, attitude measures are taken prior to and immediately after message manipulations . Occasionally a delayed post test or post tests are incorporated in the research design. This design forces a consideration of complete messages rather than individual message dements as tte unit of aelysis. To determine the effect of specific message elements on any changes that occur, a metlodology is required which permits the determination of these individual effects . This necessitates that a variety of measures be taken at appropriate intervals during the actual presentation of the message. Only by adepting such procedures will it be possible to isolate tie extent to which such factors as "rehearsal", or "credibility" or otter factors might be Operating to produced any observed differences . It is the isolation of the effects of these individual factors that will lead to an understanding of the way in which argumentative structure operates . 55 Analytic schemes such as the one just proposed would be facilitated by specifying in advance the message variables assumed to be related to each theoretic consideration assumed to be relevant to the dependent variable. If "relearsal" and "credibility" do mediate differences in attitude change, then it is necessary to specify the eleIents in the message with sufficient precision that their relationship to tlese factors can be predicted and adequately measured . In the present study messages were precisely constructed in terms of constructive message segments. Tke smallest definable unit was the constructive message segment. Tte only guides that were used in develoPing these segments were (1) that only arguments and information consistent with one position be included, and (2) that the segments have equivalent effects on receiver attitudes . Additional rules for constructing types of segments for otter t0pics need to be developed . These rules slould produce equivalence in such things as semantic content , the nature of logical cmnections, etc . Within tl'ese limitations, all but one of the attitude change findings are quite clear. If arguments that oppose tl'e communicator's position are krown to the receiver, it appears that they sl'ould be taken into account in a persuasive message. Tte study, l'owever, did not provide an explanation for this phenomena. Some tentative interpretations of the finding were proffered and suggestions for future research were made. If arguments that oppose tie communicator's position are not familiar to the receiver, trey probably should not be taken into account in a persuasive message. Previous research in the area of message "sidedness" has lead 56 to the conclusion that the "two—sided" communication is more effective than tte "one-sided" when the receiver disagrees with the cammunicator's position. Tte findings of this study suggest that the generality of this traditional proposition be restricted to "two-sided" messages which include only familiar antagonistic arguments . 57 BIBLIOGRAPHY Adams, J.B. Effects of reference group and status on opinion change. JOurn. Quart., 1950, 37, HOB-H12. Andersen, K.E. "An Experimental Study of the Interaction of Artistic and Non-artistic Ethos in Persuasion," unpub. diss., Wisconsin, 1961. Andersen, Kenneth E. and Clevenger, Theodore, Jr. A Summary of Experi- mental Research in Etl'os. g1 Vol. XXX (June, 1963) pp. 59-78. Berlo, David K. and Hideya Kumata, Tie Investigator: The Impact of a Satirical Radio Drama. Journalism @arterly, Vol. 33, pp. 287-298, (1956). Berlo, D.1<. , and Lemert, J. "An empirical test of a general construct of credibility," paper presented to SAA Convention, New York, December, 1961. Berlo, D.K., Lemert, J.B. 8 Mertz, R.J. "Dimensions for evaluating the acceptability of message sources," submitted to £9.93 Carlson, E.R. Word familiarity as a factor in forming impressions. PSYleo R820, 7, 1960, 18-23. Coffin, T.E. Some conditions of suggestion and suggestibility: a study of certain attitudinal and situational factors influencing the process of suggestion. ngchol. Nbrogr., 53, 1941, lt-12. Crane, E. Immunization: with and without use of commteruargments. JOUI'n. We, 39, 1962, “45"”50. b“) .450" Haiman, F.S. An Experimental Study of the Effects of Ethos in Public Speaking. Speech anog., 2 (1949), 190-202. Harms, L.S. Social judgerent of status cues in language. Speech anog., 27, 1960, 87-96. Hollander, E.P. Some Effects of Perceived Status on Responses to Inno- vative Behavior. J. Abnorm. Soc. Psychol., 63 (1961), 2147-250. Hovland, Carl L, Janis, Irving L. and Kelley, Harold N. Communication and Persuasion, New Haven, Yale University Press, 1953. Hovland, C.I., Lumsdaine, A.A. and Sheffield, F.D., Emeriments on Mass Communication, Princeton University Press, New Jersey, 1911.9, 201-225. Hovland, C.I. , and Mandell, W. An Experimental Comparison of Conclusion- Drawing by the Communicator and by the Audience. Journ. Abnorm. Soc. Ps chol., Vol. I+7, pp. 581-588 (1952). 58 Hovland, Carl 1., and Walter Weiss. The Influence of Source Credibility on Communication Effectiveness. Public Opinion Q1arterly, Vol. 15 , pp. 635-650, (1951-1952). Insko, C.A. One-sided versus two—sided communications and countercommuni- cations. J. Abnorm. Soc. Psychol., 65, 1962, 203-206. Kelman, M.C. Reinstatement of the Communicator in Delayed Measurement of Opinion Change. J. Abnorm. Soc. Psychol., 48 (1953), 327-335. Kerrick, Jean 8. The effect of relevant and non-relevant sources on attitude change. J. Soc. Psychol., 1958, 47, 15-20. Leitner, M.A. "A study of the effects of intraphrase rate and pause time on information gain and speaker image." unpubl. diss., U. of Wis., 1962. Indlum, T.S. "A study of the techniques for influencing the credibility of a communication," unpubl. diss., Ohio State University, 1956. Lumsdaine, A.A. and Janis I.L. Resistamce to "counter-propaganda" pro- duced by one-sided and two-sided "propaganda" presentations. 39g, 17, 1953, 311-318. Manis, M. , and Blake, J .B. Interpretation of persuasive messages as a function of prior immunization. J. Abnorm. Soc. Psychol. , 1963 , 66, 225-230. Mausner, Bernard. Studies in social interaction: III. Effect of varia- tion in one partner's prestige on the interaction of observer pairs. McCroskey, James C. Scales for the measurelent of ethos. _S_Ijl_, Vol. XXXIII, (March, 1966), pp. 65-72. McGuire, W.J. Persistance of the resistance to persuasion induced by various types of prior belief defenses. J. Abnorm. Soc. Psychol. , 64, 1962, 241-248. McGuire, W.J., and Papageorgis, D. The relative efficacy of various types of prior belief-defense. J. Abrorm. Soc. Psychol., 62, 1961, 327-337. Mertz, R.J. "Acceptance of persuasive influence as related to three dimeeions of source evaluation," unpubl. diss. , Michigan State University, 1966. Miller, G.R., and Hengll, M.A. The effect of variations in nonfluency on audience ratings of source credibility. S, 50 (1964), 36-44. Osgood, Charles E., Suci, George J., and Tannenbaum, Percy H. The Measure- ment of Meaning, Urbana, University of Illinois Press, 1957. 59 Paulson, Stanley. "Experimental study of spoken ccmmrmications; the effects of prestige of the speaker ard ackrowledgerent of opposing arguments on audience retention aid shift of opinion," unpubl. diss., Univ. of Minn., 1952. Also briefly reported in SM, 11 (1954), 267-271. Pm 'Jtmg “' Saadi, Mitchell and Farnsworth, P. The degrees of acceptance of dogmatic statements and preferences for their supposed makers. J. Abnorm. Schweitzer, Don A. Factors of commmicator credibility. in Secord 8 Backman (eds.) Problems in Social Psychology (in press). Thistlethwaite, D. and Karenetzky, J. Attitude change through refutation and elaboration of audience counter-arguments. J. Abnorm. Soc. Psyghgl., 51, 1955, 3-12. Thistlethwaite, D., Karenetzky, J. and Schmidt, H. Factors influencing attitude change through refutatiye communication. Speech anographs , 23, 1956, 19-25. 33: V73 Walster, E. and Festinger, L. The effectiveness of overheard persuasive communications. J. Abrorm. Soc. Psychol., 65 (1962), 395-402. Wolfinger, R.W. Attitude change toward source and issue resulting from one-sided and two-sided communication. Unpubl. M.A. thesis, Univ. of 111., 1955, reported in Osgood, Suci and Tannenbaum, Tie Measure- ment of Meaning Univ. of Ill. Press, Urbana, 1957. APPENDIX A: TEXTS OF THE CONSTRUCTIVE MESSAGE SEGMEN'I'S (TIME-ONE AND TIME-TWO) 61 Standard Introductory and Concluding Paragraphs (Time—one) EAVORABLE.MESSAGES It may be reasonably assumed that the probability of a thermonu- clear attack on this country is extremely low. That the use of such weapons would leave no nation a clear winner is possibly the one thing upon which world leaders agree. However, as long as stockpiles of nu— clear weapons exist and international tension.remains relatively high, the probability of a nuclear attack is, unfertunately, not reduced to zero. A slight possibility, either by non-rational intent or accident remains. As long as any possibility of sudh an occurrence exists, the only responsible course of action available is to provide the nation with the necessary protective measures. It is within this frame of reference that the current civil defense program will be discussed. It is, of course, impossible in a few paragraphs to do more than present an overview of the civil defense program.. This brief analysis has merely attempted to point up the extensive nature of the current prognam. UNEAVORABLE MESSAGES It may be reasonably assumed that the probability of a thermonu- clear attack on this country is extremely low. That the use of such weapons would leave no nation a clear winner is possibly the one thing upon which world leaders agree. However, as long as stockpiles of nu— clear weapons exist and international tension remains relatively high, the probability of a nuclear attack is, unfortunately, not reduced to zero. 4A.slight possibility, either by non-rational intent or accident remains. As long as any possibility of suCh an occurrence exists, the only responsible course of action available is to provide the nation with the necessary protective measures. It is within this frame of reference that some of the more obvious shortcomings of the current civil defense programlwill be discussed. In conclusion, the evidence suggests that civil defense programs need to Change their present emphasis. This is not meant to imply that civil defense measures be intensified, but only that they be altered to more closely reflect the realities of the situation for Which they are intended. 62 (Identification code — EALP) (EALLOUT PROGRAM-—EAVORABLE) One of the major reasons underlying the effectiveness of the civil defense program is the emphasis placed on marking and stocking fallout shelters during the past six:years. Without the protection this would provide, significantly more deaths would be caused by fall- out than by any other potential nuclear hazard.. The fallout shelter program would provide greater life saving potential per unit cost then would any other alternative program. To date, 172, 000 fallout shelters have been designated which would provide space for more than 150 million people. Seventy-nine thousand of these have already been stocked with non-perishable fOod, blankets, sanitation facilities, and emergency medical supplies. In addition, ventilation kits which will make it possible for people to stay inside shelters for two weeks or more, will soon be installed. Experts have reliably estimated that after two weeks, the radiation level will drop to one one-thousandth of its initial level, that is, within human tolerance. In all, these measures would save from.15 to 30 million people who would not otherwise survive a.major nuclear attadk. Moreover, if an attack was not full scale, or if it was directed primarily at mili- tary installations, millions more lives would be saved by the civil defense shelter programm Location of shelters is highly relevant to the numbers who would be saved. Predictions are that closer to 30 million could be adequately protected because the majority of shelters (seventy percent) are located in downtown areas of cities where the population is most dense and the need would be greatest. Proportionately fewer shelters have been constructed in rural areas that will suffer neither the blast force nor such intensive fallout danger. (Identification code — FIRp) (FIRESTORM.PROGRAMé—EAVORABLE) While offering protection from.the more obvious dangers in the event of'a.thermonuclear attack, the current civil defense program has also initiated a system of interrelated measures to deal with the lesser danger of firestorms. Considerable research has been conducted which indicates that firestorms, the fires that could follow a nuclear blast, would be limuted to a.radius of approximately only one mile from.ground zero. Thus, best available estimates conclude that for the majority of the population, firestorms would, at worst, be only a secondary hazard. Nevertheless, a number of precautionary measures have been taken in those areas where firestorms might be expected to occur. Programs 63 now underway, for example, provide fOr reinfOrcing shelters in prime target areas with concrete as well as noncombustible siding and equip- ping shelters with oxygen units and air purification units. These programs are expected to be complete by 1970, less than three years from.now. Although highly desirable, even these steps would not be absolutely necessary for Hany'locations. The typical American city, with its wide streets and mainly concrete and brick buildings, does not supply sufficient quantities of highly combustible Material to naintain large-scale destructive fires. fhrther, it is estimated that only a.small percentage of persons in the nation would be potentially affected by firestorms. For that Small percentage of the population in shelters where fires might occur, sufficient high purity oxygen in special containers designed for shelter use is already being produced. It is estimated that this oxygen could supply 200 times the number of persons who.might be affected by fire- storms. These calculations are based on the 1966 figures of the U. S. Departnent of Commerce which.report that the annual production of oxygen amounts to 51 billion cubic feet, enough fer a two week supply fer one out of every 50 Americans. Thus, even though firestorms would present less of a.hazard than would fallout, thorough precautionary measures are being taken to protect against the upper limits of its possible effects. ’ (Identification code - RECp) (RECOVERY PROGRAMF-EAVORABLE) While providing protection from.the more obvious dangers in the event of nuclear attack, civil.defense is currently developing an exten- sive number of interlocking programs which would hasten the recovery of the nation. While initial shelter protection must remain the major fUnction of civil defense, post attack.neasures are being developed to greatly reduce the hazards which.would fellow. As an example of these measures, an enlarged medical treatnent ferce would be provided for the increased numbers of people who would require medical attention. A large number of fatalities would be averted by this prompt medical treatment. The progranlwould be operated by exper— ienced physicians employing emergency facilities and would be augmented by auxiliary and paramedical personnel (nurses, dentists, pharmacists and veterinarians). This operation would continue until emergency treat- ment was no longer required. However, medical recovery, like the other types of post attack recovery, depends on adequate pre attack planning and preparation to support post attack activities. For this reason, civil defense is planning the program now so that it would be opera- tional if an attack ever should occur. The post attack medical treatment program.is but one recovery progranlnow being developed. Disease and pest control, food and water contamination control, emergency housing and exposure control as well 6% as sanitation provisions are other sinilarly planned measures. In addition to preventing unnecessary fatalities, the correlated result of these preplanned measures would be to hasten the restructuring of our society. (Identification code - FALc) (ffiiiDUT PROGRAMéeUNFAVORABLE) The nation's civil defense program, with its almost total empha- sis on fallout shelters as the major source of protection from a poten- tial nuclear attack, is unjustifiable. While this danger should not be dismissed, it is a gross miscalculation to presune that fallout would represent the most danger to human life. When other hazards are con- sidered, fallout would have to be ranked well down on the list, and fallout protection would have to be rated as having low life—saving potential on a per unit cost basis. ' ' Despite civil defense' 8 preoccupation with fallout, even in this area, the program falls considerably short of providing anything but a mere semblance of the minimum.protection which.would be required. As an example of this facade of protection, 70 per cent of all shelter space is located in the downtown areas of large cities where the danger from fallout does not in anyway compare with the danger from.the initial blast. On the other hand, in suburban and outlying areas where fallout becomes a much.greater problenn ninimally protective shelter space is available fOr less than one in ten residents. Not only is present shelter space inappropriately located, those shelters Which are available are either unstocked or only tokenly stocked ‘with feed, blankets, medical and sanitation supplies. Moreover, measures fer maintaining a crucial safe air supply have not been installed which, in itself, almost completely negates any potentially realistic userlness the existing shelter systenlnight afford. Radiation experts have esti- mated that adequate shelter Space would be required for as long as three months after a thermonuclear attaCk, befOre radiation levels would be 'within human tolerance. It is fairly Obvious that the existing civil defense program is a long way fronlbeing able to provide this minimal protection. ' (Identification code - FIRc) (FIRESTORM PROGRAMF-UNFAVORABLE) The current program focusing almost solely on fallout shelters, has neglected the major hazard of firestorns which would.innediately folldw a nuclear blast. Civil defense authorities seenlto have dis— missed what was known some twenty years ago about the firestorms created by even conventional bombs. During World war II firestorms almost leveled the EurOpean cities of Hamburg and Dresden after the allied fOrces attacked with conventional bOmbs. In Dresden, fer instance, more 65 than half of the 556, 000 dwellings were destroyed, mostly by the fire- storms which reached temperatures of 1800°F. In a single night 300, 000 were killed by a "mere" 2, 000 tons of explosives. A.20 megaton bomb, internediate sized by today's standards, would deliver ten thousand tines the tonnage Dresden.received! Instead of rationalizing these histor1cal facts, civil defense planners ought to be projecting this information into the fUture and asking what safeguards would be needed for a thermo- nuclear attack, if suCh an attack Should ever occur. It is estimated that in a city the size of Chicago hundreds of thousands would perish frenlthe firestorm.effects from a 20 megaton bomb if they sought protection in the shelters now provided by our civil de- fense systenu Shelter occupants would be almost totally vulnerable to the heat, flame, carbon monoxide and lack of oxygen. Considering only the latter, firestorms consume the vital supply of free oxygen. Shelter occupants, without large quantities of emergency oxygen supplies, would suffocate, even if they Hanaged to escape the heat and flame. Yet recent U. S. Department of Commerce figures (1966) indicate annual pro- duction of high purity oxygen, only a small proportion of'whiCh is in cylinders suitable for shelter use, amounts to only a few days' supply for less than 2 per cent of the population. Even if these supplies were installed in shelters, which they are not, they would be far’frenladequate. In short, the fallout shelter~program, based upon a set of false assump- tions, is a misleading justification fOr an unrealistic civil defense program. (Identification code — RECC) (RECOVERY PROGRAM—-UNEAVORABLE) However, the most indefensible defect in the civil defense program is its lack of concern for post attack measures. The present civil de- fense program virtually stops when survivors emerge from.their shelters, two weeks after an attack. Although civil defense has had fifteen years to initiate and operationalize post attack recovery programs, none exist. It is impossible to calculate the number of "early" survivors who would die because of this oversight. However, the cost of human life would assuredly be high and the real tragedy is that these people would die needlessly. A As an example, visualize the vast numbers of initial survivors who Will die because of the lack of necessary medical attention. This number‘would multiply simply because radiated persons are more susceptible to infection and disease. Add to these those who would be suffering from a.multitude of other injuries resulting from the attack and.the point becomes too Obvious to labor. Yet, fer all our 200 million people we have only 237,000 doctors to minister to there-now, under normal con- ditions. Of all registered doctors in Hiroshima when the "small" twenty- kiloton bomb fell, 80 per cent were immediate casualties, and unfit for service. Officials in that city attributed 35 per cent of the deaths whiCh occurred in the first week to the fact that normal medical facili— ties were no longer available. No action has been taken by civil defense '66 to program.any measures to lessen the consequences of the patient/ Physician imbalance. This is not an isolated example of civil defense's failure to develop a realistically balanced program. In the unimaginable Chaos Which.would innediately follow an attack, it would be impossible to locate auxiliary skilled personnel to cope with a.whole array of emergencies--unless organization and training had occurred prior to the emergency. No such training has been conducted by Civil Defense. If the situation would be so hopeless that post attack programs would be fUtile, civil "defense" should be exposed as a cruel myth, If this is not the case, no responsible civil defense program.can afford to neglect this phase, when the real fight fer surVival would begin. There is no alternative to the development of these programs. They cannot be delayed any fUrther simply because they laCk the visible public display of fallout shelters and the comforting aura of protection these structures affOrd. ' APPENDIX B: TEXTS OF THE REBU‘ITAL MESSAGE SEGMENTS" * . . . . Throughout this Appendix "refutatlonal" 18 used interchangeably w1th "rebuttal. " 68 Standard Introductory and Concluding Paragraphs INTRODUCTION -—ALL MESSAGES (TIME-TWO) As long as the world situation remains in its present state of unrest, and as long as an increasing number of nations continue to devel- op and stockpile weapons of mass destruction, most people would agree that no nation can afford to take civil defense lightly. Certainly the chances of a nuclear attack are sufficiently remote that it would be unwise for every citizen to devote much tine and energy dwelling on such a psychologically disturbing matter. If this were to happen the end result could be an undesirable national paranoia about nuclear war. On the other hand, however, it would be distressing for the average person to think that civil defense authorities lad not developed programs to minimize the destructive effects of the "unthinkable" should it ever occur. CONCLUSION-—FAVORABLE MESSAGES (TIME—TWO) It is, of course, impossible in a few paragraphs to do more than present an overview of the civil defense program. This brief analysis has merely attempted to point up the extensive nature of the current program. CONCLUSION--UNFAVORABLE MESSAGES (TIME—TWO) In conclusion, the evidence suggests that civil defense programs need to change their present emphasis. This is not meant to imply that civil defense measures be intensified, but only that they be altered to more closely reflect the realities of the situation for which they are intended. . (Identification code — PAL-PALp) (EALLOUI' PROGRAM--REEUI‘ATIONAL FAVORABLE) The present civil defense program has come under attack from some quarters for concentrating its efforts on fallout protection. Critics of the program contend that , in the event of a nuclear attack, hazards other than fallout would represent the greatest dangers to human life, and that these hazards have been largely disregarded by civil defense officials. This criticism is , however, unwarranted. Without the pro- tection fallout shelters would provide , significantly more deaths would result from fallout than from any other nuclear hazard. The fallout shelter program would provide greater lifesaving potential per unit cost than would any other alternative enpha318 . In keeping with this analysis , 172 ,000 fallout shelters have been designated which would provide Space for more than 150 million peOple. Seventy-nine thousand, or almost half, of these have been stocked with nonperishable food, blankets, sanitation facilities, and emergency medical supplies. In addition to these measures, civil defense officials realize that the lifesaving potential of shelters is greatly reduced be— cause they are not presently equipped with air'purification facilities. Consequently, ventilation kits which will make it possible fer shelter occupants to remain inside fer two weeks or more, will soon be installed. Admittedly, there is considerable controversy over the length of time survivors will have to remain in shelters after an attack. Expert opinion varies from.two weeks to as long as three months. However, radia— tion, which would be the greatest danger to human life, drOps off in two weeks to 1/1000 its initial level, that is, to within human tolerance. In all, it is estimated.that the existing fallout shelter system. would save from.15 to 30 million people Who onld not otherwise survive a major nuclear attack. This is largely due to the fact that shelter space has been strategically located where the danger, and hence the necessary protection required, would be greatest. Nearly three-quarters of all shelter is in the downtown areas of metropolitan centers where the population is most concentrated and.where attacks, if they came, would likely occur. Critics of this allocation of Shelter space have argued that the danger from fallout in the downtown areas of cities will be less critical than the hazard.of the initial blast. The real danger from.fallout, they contend, is in suburban areas Where proportionately less shelter space is located. This position, is, of course, untenable. Net only is shelter space immediately available for approximately ten percent of all suburb banites, but the inescapable fact remains--the vast majority of Americans are concentrated within cities. (Identification code - PIRrFIRp) (FIRESTORM PROGRAMé-REFUTATIONAL EAVORABLE) While offering protection from the more obvious dangers in the event of a thermonuclear attack, the current civil defense programlhas also initiated a system of interrelated measures to deal.with the lesser danger of firestorms. The fires whiCh would fellow a nuclear blast would be intense at ground zero but extensive research has shown that the hazard of'fire ' would be limited to a.radius of one mile frcnlthis point. The widespread fires whiCh fellowed the allied bombings of Hamburg and Dresden near the end of Werld.War II resulted largely because most of the structures in these cities were of highly combustible materials whidh fed the initial flames. This, plus the fact that in these cities the streets were ex— tremely narrow and fire conditions ideal, permitted 2,000 tons of non- nuclear explosives to extensively damage the cities and cause hundreds of thousands of deaths. This would not be the case in the U. S. where the streets of large cities are wide and the buildings constructed of steel, concrete and brick, despite the fact that a 20 megaton bomb has 70 much greater explosive power'than did the bombs used at Iresden and Hamburg. While fire cannot be dismissed as a nuclear hazard, the estimate, by critics of the present civil defense programu that hundreds of thousands would perish if they were in existing shelters, is a great exaggeration. Most shelter occupants would not be vulnerable to fire- storms or the hazards of carbon monoxide poisoning and insufficient oxygen whiCh accompany fire. Prom.the best available evidence, it can be concluded that fer the majority of the population, firestorms would, at worst, be only a secondary hazard. Nevertheless, a number of precautionary measures have been taken in those areas where firestorms might be expected to occur. Programs now underway, for example, provide fOr reinforcing shelters in prime target areas with concrete as well as noncombustible siding and equip- ping shelters with oxygen units and air purification units. These pro- . grams are expected to be complete by 1970, less than three years from.now. Phrthermore, for that small percentage of the population in shelters where fires might occur, sufficient high purity oxygen in special containers designed for shelter use is already being produced. .Although not presently located in shelters, these could be quickly trans- ported if world tension were to significantly increase. It is estimated that this oxygen could supply 200 times the number of persons who might be affected by firestorms. These calculations are based on the 1966 figures of the U. S. Department of Commerce which reports that the annual production of oxygen amounts to 51 billion cubic feet, enough for a two week supply fer one out of every 50 Americans. Thus, even though firestorms would present less of a.hazard, than would fallout, thorough precautionary measures are being taken to protect against the upper limits of firestorms' possible effects. (Identification code - REC—RECp) (RECOVERY PROGRAM- -REEUTATI ONAL FAVORABLE) While shelter protection must remain the major fUnction of civil defense, an extensive number of interlocking programs are currently being developed whiCh would hasten the recovery of the nation. Civil defense has been sharply criticized in the past fOr placing the major emphasis on shelters at the expense of vital post attack programs. Critics of the program.point to Hiroshima where a.relatively small nu- clear bomb immediately incapacitated most of the medical personnel. It is true that thousands died because medical facilities were unavailable and that results in this country, without a preplanned action program, would be similarly disastrous. For this reason, a program.is currently being planned which would provide fer an enlarged medical treatment fOrce fer the increased number of people who would require medical attention. A large number of fatal- ities would be averted by this prompt medical treatment. The program 71 would be operated by experienced physicians employing emergency facili- ties and would be augmented by auxiliary and paramedical personnel (nurses, dentists, pharmacists and veterinarians). This Operation would continue until emergency treatment was no longer required. The post attack medical treatment program is but one recovery program now being developed. Disease and pest control, fOOd and water contamination control, emergency housing and exposure control as well as sanitation provisions are other similarly planned measures. In addition to preventing unnecessary fatalities, the correlated result of these preplanned measures would be to hasten the restructuring of our society. (Identification code — EAL-EALC) (EALLOUT PROGRAM-eREfUTATIONAl.UNEAVORABLE) The nation's civil defense program, with its almost total emphasis on fallout shelters as the major source of protection from a potential nuclear attack, is unjustifiable. This emphasis is the result of the erroneous assumption that more deaths would be caused by fallout than any other potential nuclear hazard. This is a gross miscalculation. When other hazards are considered, fallout would have to be ranked well down on the list, and fallout protection.wou1d.have to be rated as having low lifesaving potential on a per unit cost basis. ’ Despite civil defense' 3 preoccupation with the fallout shelter prtgramu even.within this limited area, only a.mere semblance of the minimumlprotection which would be required has been provided. .Although 172,000 fallout shelters have been designated which could accommodate 150 million people, and, according to civil defense personnel, save up to 30 million who would otherwiSe die, these estimates are not only ex— tremely misleading but also fallacious. Seventy percent of these "shelters" are in the downtown areas of major cities where, contrary to Civil Defense pronouncements, the danger from.fallout is minimal when compared with the danger'fromlthe initial blast. Also, contrary to Civil Defense calculations, the concentration of people is not in the downtown areas but in the suburbs where the danger from.fallout is potentially greatest. Paradoxically, shelter space for suburbanites is almost nonexistent. Calculated as a national average, shelter space has been provided for less than 10 per cent of this segment of the population. In addition to the inappropriateness of shelter location, the illusion of safety becomes even clearer by considering the fact that more than half of these shelters have not been stocked with any emergency pro- visions. Such disaster necessities as nonperishable fOod, sanitation facilities, and medical supplies have been.minimally distributed. But most important, no measures have been taken to ensure a.crucial safe air supply in any shelters. Although civil defense authorities have been "considering" installing ventilation kits, this talk has continued for a number Of years and no action has been taken. Without such equipment any realistic usefulness the existing shelter systemlmight potentially 72 affOrd, is completely negated. Furthermore, experts in the field of radiation have estimated that adequate shelter facilities would be required.from.two weeks to three months depending upon a large number of interrelated factors. Per some reason, civil defense persOnnel have elected to base the projections they have made on the two week figure. Despite the fact that most types of radiation drop to l/1000 of their initial radiation level in this period, without any way of knowing what the initial radiation level might be, the magic number "two" weeks represents rather curious optimism.‘ When all of these inadequacies are considered, it becomes exceedingly Obvious that the existing civil defense program.provides only mythical protection. ' ' (Identification code — PIR-PIRc) (FIRESTORM.PROGRAMé—REFUTATIONAL‘UNEAVORABLE) The nation's civil defense programu fOcusing almost solely on fallout shelters, has largely negleCted the major hazard of firestorms which would immediately follow a nuclear’blast. On the assumption that firestorms would be limited to a radius of only one mile from.ground zero, a trivial program.fOr reinforcing a.small fraction of Shelters with concrete and noncOmbustible siding has been instituted by civil defense planners. These precautions, so minimal that they will take only a couple of years to complete, provide a clear indication of just how lightly this major hazard is being taken by some nuclear defense planners. Civil defense authorities seemlto have dismissed what was known some twenty years ago about the firestorms created by even conventional bombs. During WOrld war II firestorms almost leveled the European cities of Hamburg and Dresden after the allied fOrces attacked with 2927 ventional bombs. In Dresden, for instance, more than half Of the 556,000 dwellings were destroyed, mostly by the firestorms which reached temperatures Of 1800° P. In a single night 300,000 people were killed by a "mere" 2,000 tons of explosives. 'Those who attempt to downgrade the importance of firestorms counter with the argument that U. S. cities, unlike Hamburg and Dresden, have wide streets and buildings built mainly of concrete and brick. For this reason, they contend, there would be insufficient quantities of combustible material to support widespread firestorms. This is, of course, absurd, A.20 megaton bomb, intermediate size by today's standards, would deliver ten thousand times the tonnage Dresden received! Instead of rationalizing these hiStorical facts, civil defense planners ought to be projecting this information into the future and asking what Safeguards would be needed for a thermonuclear attack, if such an attaCk should ever occur. It is estimated that in a city the size of Chicago hundreds of thousands would perish fromlthe firestormleffects of a 20 megaton bomb if they sought protection in the shelters now provided by our civil defense syStemn Shelter occupants would be almrst totally vulnerable to the heat, flame, carbon monoxide and lack of oxygen. Considering only 73 the latter, firestorms consume the vital supply of free Oxygen. Shelter occupants, without large quantities of emergency oxygen supplies, would suffocate, even if they managed to escape the heat and flame. Yet, recent U. S. Department of Commerce figures (1966) indicate annual pro- duction of high purity oxygen, only a small proportion of which is in cylinders suitable fOr shelter use, amounts to only a few days' supply for less than 2 per cent of the population. Even if these supplies were installed in shelters, which they are not, they would, despite the claims of some, be far from.adequate. In short, the fallout shelter program, based upon a set of false assumptions, is a misleading justification for an unrealistic civil defense program. (Identification code — REC—RECc) (RECOVERY PROGRAMF—REFUTATIONAi.UNEAVORABLE) The most indefensible defect in the civil defense program.is its lack of concern for postattack measures. The present civil defense pro- . gram virtually stops when survivors emerge from their shelters, two weeks after an attack. It is impossible to calculate the number of "early" survivors who would die because of this oversight. However, the cost of human life would assuredly be high and the real tragedy is that these people would die needlessly. ’ As an example, visualize the vast numbers of initial survivors who will die because of the lack of necessary medical attention. This number'would multiply simply because radiated persons are more suscepti- ble to infection and disease. Add to these those who would be suffering from.a multitude of other injuries resulting from the attack and the ' point becomes too Obvious to labor. Yet, for all our 200 million people we have only 237,000 doctors to minister to theme-now, under normal con- ditions. Of all registered doctors in Hiroshima when the "small" twenty- kiloton bomb fell, 80 per cent were immediate casualties and unfit fOr service. Officials in that city attributed 35 per cent of the deaths which occurred in the first week to the fact that normal medical facili- ties were no longer available. Civil defense officials have had more than twenty-five years to develop a postattack medical program but their output still amounts to nothing more than an "interesting set of hy- potheses" about what might be done and about who might do it. In the almost unimaginable chaos Which would surely fellow an attack, it would be impoSsible to locate auxilliary trained personnel to cope with the whole array of emergencies which.would emerge--that is, unless organization and training had occurred prior to the emergency. Failure to preplan emergency medical facilities is not an isolated ex- ample of civil defense' 3 attitude towards a balanced postattack program. These programs Simply do not exist. Civil defense will argue that "measures are being planned" but these are paper’programs. No action has been taken. I 74 If the situation would be so hopeless that postattack programs would be fUtile, civil "defense" should be exposed as a cruel myth. If this is not the case, no responsible civil defense program can afford to neglect this phase, when the real fight fOr survival would begin. There is no alternative to the development of these programs. They cannot be delayed any fUrther simply because they lack the visible public display of fallout shelters, with their comforting aura of Protection. APPENDIX C: THE TIME-ONE QUESTIONNAIRE 76 FOR RESEARCH PURPOSES ONLY Communication Survey leA Please fill in the following: Name: Year in School: Student Number: Sex: 77 INTRODUCTION This research study is one of a series concerned with the way college students react to various kinds of information on public issues. Please note that the infOrmation you provide will be held strictly confidential and will be used only fOr the purposes of this research project. In this study we are interested in a number of kinds of infor— mation. we'll deal with them one at a time and give you the necessary instructions as we go along. This study concerns a number of public issues on Which various kinds of information have been.made available by a variety of infor— mation sources. In this phase of the study we are asking you to read and make some types of evaluations of just one passage about one issue. The passage starts on the next page of this boOklet. Please read the passage carefu11y. Please turn the page and begin. . . 78 It may be reasonably assumed that the probability of a thermonu- clear attack on this country is extremely low. That the use of such weapons would leave no nation a clear winner is possibly the one thing upon which world leaders agree. However, as long as stockpiles of nu- clear weapons exist and international tension remains relatively high, the probability of a nuclear attack is, unfOrtunately, not reduced to zero. .A slight possibility, either by non-rational intent or accident remains. .As long as any possibility of such an occurrence exists, the only responsible course of action available is to provide the nation with the necessary protective measures. It is within this frame of reference that some of the current civil defense program will be dis- cussed. One of the major reasons underlying the effectiveness of the civil defense programris the emphasis placed on marking and stocking fallout shelters during the past six years. Without the protection this would provide, Significantly more deaths would be caused by fall- out than by any other potential nuclear hazard.. The fallout shelter pmtpramlwould provide greater life saving potential per unit cost than wOuld any other alternative program. To date, 172,000 fallout shelters have been designated which would provide space for more than 150 million people. Seventy-nine thousand of these have already been stocked with non—perishable fOOd, blankets, sanitation facilities, and emergency medical supplies. In addition, ventilation kits which will.make it possible fOr people to stay inside shelters for two weeks or more, will soon be installed. Experts have reliably estimated that after two weeks, the radiation level.will drop to one one—thousandth of its initial level, that is, within human tolerance. In all, these measures would save from.15 to 30 million.people who would not otherwise survive a.major nuclear attack. Mereover, if an attack was not fu11 scale, or if it was directed primarily at mili- tary installation, millions more lives would be saved by the civil de— fense shelter programn Location of shelters is highly relevant to the numbers who would be saved. Predictions are that closer to 30 million could be adequately protected because the majority of shelters (seventy percent) are located in downtown areas of cities where the population is most dense and the need would be greatest. Proportionately fewer shelters have been con- structed in rural areas that will suffer neither the blast fOrce nor such intensive fallout danger. While offering protection.from.the.more Obvious dangers in the event of a thermonuclear attack, the current civil defense programlhas also initiated a system.of interrelated measures to deal with the lesser danger of firestorms. Considerable research has been conducted which indicates that firestorms, the fires that could fellow a nuclear blast, 79 would be limited to a radius of approximately only one mile from.ground zero. Thus, best available estimates conclude that fer the majority of the population, firestorms would, at worst, be only a secondary hazard. Nevertheless, a number of precautionary measures have been taken in those areas where firestorms might be expected to occur. Programs now underway, fOr example, provide fOr reinforcing shelters in prime target areas with concrete as well as noncombustible siding and equip- ping shelters with oxygen units and air purification units. These pro- . grams are expected to be complete by 1970, less than three years from now. Although highly desirable, even these steps would not be absolutely necessary fOr many locations. The typicaermerican city, with its wide streets and mainly concrete and brick buildings, does not supply suffi- cient quantities of highly combustible material to maintain large-scale destructive fires. ‘ ' Enrther, it is estimated that only a small percentage of persons in the nation would be potentially affected by firestormS. For that small percentage of the population in shelters where fires might occur, sufficient high purity oxygen in special containers designed fOr shelter use is already being produced. It is estimated that this oxygen could supply 200 times the number of persons Who might be affected by fire— storms. These calculations are based on the 1966 figures of the U. S. Department of Commerce WhiCh report that the annual production of oxygen amounts to 51 billion cubic feet, enough fOr a two week supply for one out of every 50.Americans. Thus, even though firestorms would present less of a hazard than would fallout, thorough precautionary measures are being taken to protect against the upper limits of its possible effects. It is, of course, impossible in a few paragraphs to do more than present an overview of the civil defense program, This brief analysis has merely attempted to point up the extensive nature of the programu ****** 80 Now that you have finished reading, we'd like to get your evalua- tions Of a number of aspects of this piece of writing and of the writer. On the following pages you will find a number of rating scales upon which to make your evaluations. These are 7—point scales of which the fOllow— ing is a sample: Good : : : : : : Bad VERY . QUITE SLIGHTLY NEUTRAL SLIGHTLY AQUITE VERY or DON'T KNOW Here is how the scales work. Suppose you were asked to rate the idea: "18 year old vote in Michigan." If, in your opinion, a voting page of 18 is a very good idea, you should put a check in the extreme left-hand position of the scale (in the space closest to the adjective "Good"). If, on the other hand, you feel a voting age of 18 is a.yery_ bad_idea you should mark the extreme right-hand position (in the space closest to the adjective "Bad"). If you could not decide whether a voting age of 18 would be good or bad, or if your position of a voting Vage of 18 was neutral, then you Should mark the center position on the scale. Please fellow this procedure on every scale on the fOllowing pages. Mark each scale only once and d9_not skip any scales. If you have any questions about how to mark the scales, please ask them.now. Okay, please turn the page and begin . . . 81 First of all, befOre you rate the passage, we would like your own evaluation of the issue "the current civil nuclear defense program," Please use the rating scales below to indicate your opinion. "The Current Civil Nuclear Defense Program" worthless : : : : : : : valuable _ good : : : : : : : bad wise : : : : : : : foolish honest : : : : : : : dishonest fair : : : : : : : unfair On the basis of this one sample of the writer's work, what do you think this person would be like as a source of infOrmation? just unqualified skilled dangerous hesitant infOrmed energetic kind inexperienced bold active honest unjust qualified unskilled safe emphatic uninformed tired cruel experienced timid passive dishonest 83 In your opinion, would you say that the person who wrote this passage was favorable or unfavorable to the "current civil nuclear defense program." favorable : : : : : : unfavorable In_genera1, would you say that the information provided in the passage was previously familiar or unfamiliar to you? familiar : : : : : : unfamiliar Would you say that the writing was clear or unclear? clear : : : : : : unclear APPENDIX D: THE TIME-TWO QUESTIONNAIRE 85 Confidential — For Research Purposes Only Do Not write Here C 1-3 Project NR C 4 Card NR C 5 Phase NR C 6 Sub Deck NR C 7-9 Respondent NR Communication Survey 2—C Please fill in the following: Name: Year in School: St'llden‘t NUJ'Ilber: Sex : 86 Introduction During the past month, a large number of peOple took part in a pilot investigation conducted on campus . This investigation was concerned with the way various kinds of information Was received . It also was concerned with the way in which these kinds of information affected the way the writer was subsequently perceived. Information on four public issues was investigated. These were (1) legislative Reapportionment , (2) The Office of Economic Opportunity, (3) The Civil Defense Program, and (LL) Medicare. Of the four issues, the Civil Defense Program met the necessary research criteria for the present study. In the present study we are concerned with a variety Of kinds of information which have come from a number of sources interested in the topic. We would like you to read and make some evaluations of one of these passages. It begins on the next page. Please begin. Note: If you took part in the pilot study the general format of this questionnaire will be familiar. For the purposes Of the present study, it does not matter if the issue was the same one presented to you in the pilot investigation. 87 As long as the world situation remains in its present state of unrest , and as long as an increasing number of nations continue to develop and stockpile weapons of mass destruction, most people would agree that no nation can afford to take civil defense lightly. Certainly the chances of a nuclear attack are sufficiently remote that it would be unwise for every citizen to devote much time and energy dwelling on such a psychologically disturbing matter. If this Were to happen the end result could be an undesirable national paranoia about nuclear war. On the other hand, however, it would be distressing for the average person to think that civil defense authorities had not developed programs to minimize the destructive effects of the "unthink— able" should it ever occur. The nation's civil defense program, focussing almost solely on fallout shelters, has largely neglected the major hazard of firestorms which would immediately follow a nuclear blast. On the assumption that firestorms would be limited to a radius of only one mile from ground zero, a trivial program for reinforcing a small fraction of Shelters with concrete and noncombustible siding has been instituted by civil defense planners. These precautions, so minimal that they will take only a couple of years to complete, provide a clear indication of just how lightly this major hazard is being taken by some nuclear defense planners . ' Civil defense authorities seem to have dismissed what was known some twenty years _ago about the firestorms created by even conventional bombs. During World War II firestorms almost leveled the European cities of Hamburg and Dresden after the allied forces attacked with conventional bombs. In Dresden, for instance, more than half of the 556,000 dwellings were destroyed, mostly by the firestorms which reached temperatures Of 1800° F. In a single night 300,000 people were killed by a "mere" 2,000 tons of explosives. Those who attempt to downgrade the importance of firestorms counter with the argument that U. S. cities, unlike Hamburg and Dresden , have wide streets and buildings built mainly of concrete and brick. For this reason, they contend, there would be insufficient quantities of combustible material to support widespread firestorms. This is, of course, absurd. A 20 megaton bomb, intermediate size by today's standards , would deliver t__e_n_ thousand times the tonnage Dresden received! Instead of rationalizing these h18torical facts, civil defense planners ought to be projecting this information into the future and asking what safeguards would be needed for a thermonuclear attack, if such an attack should ever occur. It is estimated that in a city the size of Chicago hundreds Of thousands would perish from the firestorm effects of a 20 megaton bomb if they sought protection in the shelters now provided by our civil defense system. Shelter occupants would be almost totally vulnerable to the heat , flame , carbon monoxide and lack of oxygen . Considering only the latter, firestorms consume the vital supply of free oxygen. Shelter occupants , without large quantities of emergency oxygen supplies , would suffocate, even if they managed to escape the heat and flame. Yet , recent U. S. Department of Commerce figures (1966) indicate annual pro- duction of high purity oxygen, only a small proportion of which is in 88. cylinders suitable fOr shelter use, amounts to only a few days' supply for less than 2 per cent of the population. Even if these supplies were installed in shelters, which they are not, they would, despite the claims of some, be far from adequate. In short, the fallout shelter program, based upon a set of false assumptions, is a misleading justifi— cation fOr an unrealistic civil defense program. ' In conclusion, the evidence suggests that civil defense programs need to Change their present emphasis. This is not meant to imply that civil defense measures be intensified, but only that they be altered to more closely reflect the realities of the situation fOr which they are intended. 89 Now that you have finished reading, we'd like to get your evaluation of a number of aspects of this piece of writing and of the writer. On the following pages you will find a number of rating scales upon which to make your evaluations. These are 7—point scales of which the fOllowing is a sample: Good Bad ‘ VERY ‘ QUITE SLIGHTLY . NEUTRAL SLIGHTLY . QUITE VERY , KNOW OI‘ NEUTRAL Here.is how the scales work. Suppose you were asked to rate the idea "18 year old vote in Michigan." If, in your opinion, a voting age of 18 is a very good idea, you Should put a Check in the extreme leftéhand position of the scale (in the space closest to the adjective_"good"). If, on the other’hand, you feel a voting age of 18 is a very bad_idea you should mark the extreme right—hand pOsition next to the adjective "bad. " If you could not decide whether a voting age of 18 would be good or bad, or if your position on the matter is neutral, then you should mark the center position on the scale. Please fellow this procedure on every scale on the following pages. Mark each scale only once and do_not skip any Scales. Continue . . . 90 Before you rate the passage, we would like your own evaluation of "the current civil nuclear defense'program." Please use the rating scales below to indicate your Opinion. The Current Civil Nuclear Civil Defense Program strong : : : : : : weak unsuccessful : : : : : : successful honest : : : : : : dishonest useless : : : : : : useful valuable : : : : : : ‘worthless important : : : : : : unimportant . good : : : : : : bad feelish : : : : : : wise fair : : : : : : unfair inapprOpriate : : : : : : appropriate How would you rate the person who wrote this passage as a source of infOrmation on the issue "the current civil nuclear defense programW? dangerous qualified timid unjust infOrmed dishonest cruel emphatic active energetic inexperienced skilled safe unqualified bold just uninfOrmed honest kind hesitant passive tired experienced unskilled 92 In your Opinion, would you say that the person who wrote this passage was favorable or unfavorable to "the current civil defense program"? favorable : : : : : : unfavorable WOuld you say that the writing was clear or unclear? clear : : : : : : unclear Before you read the passage, how much Of the infOrmation.was previously unfamiliar to you i.e., how much of the infOrmation was novel? (check one) ____ All of it _____Quite a bit of it Some of it ____ A little bit of it None of it APPENDIX E: BASIC EXPERIMENTAL DESIGN: DIDICATING MESSAGE COUNTERBALANCING PROCEHJRE AND DISPOSITION OF TIME-ONE AND TIME—TWO MESSAGES Engage EON/00mg I oommomm Engage hex/comp I 00mm garage coapompgm EOpmmfim I OMEN Am>flpogmcoo #6308me :EOpmmfium I 0mm IOH Amso>mwcsv QOHEmuIQHm vsoadmw I 0433mm ImHQmQQ/MMGB 8303.95 “Boise I 0.2m Ramona Emu/008 I oommomm madman hurt/00mg I ovum 330.38 coabomubue 6.653% I 815E $35538 83030an 150533 I imam ImHanQ/mmv GOEIOOUIOSQ 950.33 I minim ImHAmHO>mmv coapomvofim ”Boise I 3% "mHOQEAm OP >va omuz HHmU . ownz HHmo . omnz Harv . omuz Sumo . omnz :00 . omuz HHmO _ ONHHZ . . Amqu . . . 2”..qu . .d H 21: ”I I Iommmomm- H H are ”I I I wows- I” are wow . Wmammmm . an: . . 8510mm . ATS . Seesaw 1% m . . SEE . . . umE . we a. _ Aomncv . IIIIIII . IIIIIII . Sang . IIIIIII 4 IIIIII ... I I I I 1.1 m . . ANHCV . . . 2..qu . m 0 a “commune U SE: ”I I IommwommI H umohfim H :ch ”I I I we mm I I” $115 a w W. SEE . Bayard . Sufi . . commie _ SHE . mommies w om unease . Side . SEE . . . 6am . a W m . ....... . ....... _ . ....... . ...... .---- we . . Amnfiv . . . Amqu . 1+ . AmnCv _ OMEN—H . . Amufiv _ 0E . AHJHCV . REE .I I I Nmmcw I . . emerge .I I IAMHmVI I. Sammie _ Sfiqfi . . . . . . . Human L . . 012m . mac: . 8mm . «83mm . Ono: . 95mm L 53mm HM. pm odomoamsoo metamonm . HE Em ownmow mcoo memhwonm Op mEmoOflBmHom Hadrofim . on. ownmcoapmémm Hugogm goapm Eppsnmm . gpoém m>mpopfimcoo mummmmmzdspImfih mmwmwmmz OEImeflB new OQOIOEIE. mo :Oafimommflm "a, manure Admtsnwn anemone 301608 I0ommomm A0>HE§mco0 firewood NEE/000,» I 00mm I0HEQO>0IH§V coflbmubhm 890.3083 I Ban—mm Imanmfio>mmc3 #8300989 Eooohmafl I 0mm cowbomubcd 933.3% I 0.23% £0.30mean page...“ I 0.2m Ramona heme/0009 I mommomm Endonnm Ewe/000% I 8mm Admtonwu 63.08.93 60508.6 I arm—ME AgHEfimcg 8.310303 Esooamwuu I dem Imaompozmmv 183000.93 Bogota I mailman Imanggmmv coaubmpoao ”503.6% I aim "3096mm 8. Knox 95 HE E omeoncoo 910%me 8. aEmcoflmHmm flgfim 88.08% SEER as a. Emeficoo assent Op marzmcompmamm EQB0apm 093.03% $303me00 ong .300 . omuz .300 . omuz :00 . omnz .300 . omnz .300 . ong .300 . omHnZ — . AMHCV — u — Amncv — I... _ SHE . mommomm . . Annoy . 3mm . Ammucv mm . mommomm. IIIIIII . . . IIIIIII . pm Tm. . SEE . ATE _ . Samara . :15 . Seaman rm. w . . 915E . . . 9E . mm a Aomnflv . IIIIIII . lllllll . AONHCV . IIIIIII . IIIIIII . IIIII m m. w . . AmnCv . . . AmnCv . «cw _ . mogmm . 3mm . . momma . m. WWI. QOMMIUMM . AwHCV . lllllll . mmmfléh _ Awnfiv . lllllll . Amjnflv .Me may mamas . mommomm . ole _ . Sammie . $ch . 831 a m0 om égm . Alma/E . . . . . s . ....... ride- _ _ ....... Izmir--- 1. _ . A1HCV - . — AmHCv . . . @mHmmHm . . . 9.3M _ . 2.qu . IIIIIII . . Aquv . IIIIIII _ Abmucv . QMEHM . AmHCv . . . AJHGV . . Side . SEE _ . RES/m . SE . omESE . . . . . go: Qtdm P Lame/emu” P @902 _ gm . ,Hmzmmn . . . www.mmmz OEIQfiH 8,5089% 3an HI HIGRN IIIIIIIZI STATE 9310 U NIV. LIBRRRIES MHIIIIHIIHHIIIUWWI 0719750