oqg ethos and credibility: the construct and ...jamescmccroskey.com/publications/096.pdfethos and...

11
OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ITS MEASUREMENT AFTER THREE DECADES James C. McCroskey and Thomas J. Young F E\V subjects have received as much attention from researchers in com- munication or speech as "ethos," or more commonly "source credibility." Scholar- ly attention has focused on the impact of source credibility in the communica- tion process, on the impact of a variety of variables on source credibility, and on the nature of the credibility construct and its measurement. Our attention in this paper deals with the latter concern, the nature of the ethos or source credibility construct and its measure- ment. In the course of our discussion we suggest that historical conceptualiza- tions of the ethos or source credibility construct have not been challenged seriously and that adequate measures of the construct are available. Furthermore, continued scholarly attention to the construct and its measurement may only serve to distract the field from concerns more likely to yield meaningful knowl- edge about the process of human com- munication. THE EARLY HISTORY OF THE CONSTRUCT Throughout most of the rwenty-four hundred year history of the study of rhetoric, ethos has held a central posi- tion. Aristotle's view that ethos is the most potent means of persuasion has been supported by many contemporary James C. McCroskey is Professor and Chair- person in the Department of Speech Communi- cation at West Virginia University, and Thomas ]. Young is Associate Professor of Speech Com- munication at West Virginia University. rhetorical scholars.l Similarly, other ancient rhetorical scholars (e.g., Plato, Cicero, Quintilian) expressed views comparable to Aristotle's even though their conception of the nature of ethos differed somewhat from his.::! Contemporary research generally has supported the proposition that source credibility is a very important element in the communication process, whether the goal of the communication effort be persuasion or the generation of under- standing.:l While most contemporary writers suggest that source credibility is the "attitude toward a source of com- munication held at a given time by a receiver," it is commonly noted that this "attitude" is not undimensional. but rather multidimensional. The multi- dimensionality of the construct was noted in classical times as well. Aristotle. for example. suggested that ethos had three dimensions: intelligence, char- acter. and good will.o! He believed that these were the three perceptual sources of influence on. a receiver. More recent- ly, Hovland. Janis, and Kelley in their research program investigating com- 1 Lane Cooper. The Rhetoric of Aristotle (~ew York: Appleton-Century-Crofts, 1932). :! William ~L Sattler. "Conceptions of Ethos in Ancient Rhetoric:' Speech Monographs, 14 (194;). 55-65. 3 Kenneth Andersen and Theodore Clevenger. Jr.. "A Summary of E."<perimental Research in Ethos," Speech MonograPhs, 30 (1963), 59-i8; for a more recent view. see James C. McCroskey. An lntroduction to Rhetorical Communication, 2nd ed. (Eng-lewood Cliffs. New Jersey: Prentice Hall. Inc., 1972). Ch. 4. -1Cooper. CE~TRAL STATES SPEECH JOLR~AL. Volume 32. Spring 1981

Upload: others

Post on 19-Mar-2021

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

OqG

ETHOS AND CREDIBiliTY: THE CONSTRUCTAND ITS MEASUREMENT AFTER

THREE DECADES

James C. McCroskey and Thomas J. Young

FE\V subjects have received as muchattention from researchers in com-

munication or speech as "ethos," or morecommonly "source credibility." Scholar-ly attention has focused on the impactof source credibility in the communica-tion process, on the impact of a varietyof variables on source credibility, andon the nature of the credibility constructand its measurement. Our attention in

this paper deals with the latter concern,the nature of the ethos or source

credibility construct and its measure-ment. In the course of our discussion we

suggest that historical conceptualiza-tions of the ethos or source credibilityconstruct have not been challengedseriously and that adequate measures ofthe construct are available. Furthermore,

continued scholarly attention to theconstruct and its measurement may onlyserve to distract the field from concerns

more likely to yield meaningful knowl-edge about the process of human com-munication.

THE EARLY HISTORY OF THE CONSTRUCT

Throughout most of the rwenty-fourhundred year history of the study ofrhetoric, ethos has held a central posi-tion. Aristotle's view that ethos is the

most potent means of persuasion hasbeen supported by many contemporary

James C. McCroskey is Professor and Chair-person in the Department of Speech Communi-cation at West Virginia University, and Thomas]. Young is Associate Professor of Speech Com-munication at West Virginia University.

rhetorical scholars.l Similarly, otherancient rhetorical scholars (e.g., Plato,Cicero, Quintilian) expressed viewscomparable to Aristotle's even thoughtheir conception of the nature of ethosdiffered somewhat from his.::!

Contemporary research generally hassupported the proposition that sourcecredibility is a very important elementin the communication process, whetherthe goal of the communication effort bepersuasion or the generation of under-standing.:l While most contemporarywriters suggest that source credibility isthe "attitude toward a source of com-

munication held at a given time by areceiver," it is commonly noted that this"attitude" is not undimensional. butrather multidimensional. The multi-

dimensionality of the construct wasnoted in classical times as well. Aristotle.

for example. suggested that ethos hadthree dimensions: intelligence, char-acter. and good will.o! He believed thatthese were the three perceptual sourcesof influence on. a receiver. More recent-

ly, Hovland. Janis, and Kelley in theirresearch program investigating com-

1 Lane Cooper. The Rhetoric of Aristotle(~ew York: Appleton-Century-Crofts, 1932).

:!William ~L Sattler. "Conceptions of Ethosin Ancient Rhetoric:' Speech Monographs, 14(194;). 55-65.

3 Kenneth Andersen and Theodore Clevenger.Jr.. "A Summary of E."<perimental Research inEthos," Speech MonograPhs, 30 (1963), 59-i8;for a more recent view. see James C. McCroskey.An lntroduction to Rhetorical Communication,2nd ed. (Eng-lewood Cliffs. New Jersey: PrenticeHall. Inc., 1972). Ch. 4.

-1Cooper.

CE~TRAL STATES SPEECH JOLR~AL. Volume 32. Spring 1981

Page 2: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

ETHOS AND CREDIBILITY 25

munication and persuasion during andafter World War II, suggested threedimensions for the credibility construct:expertness, trustworthiness, and inten-tion toward the receiver.5 Althoughthese authors do not acknowledgefamiliarity with Aristotle's work, thedimensions which they suggest closelyparallel those that have been discussedthroughout the centuries. Historically,then, the construct of ethos or sourcecredibility has long been thought to in-volve a source's knowledge of the subjectthat he or she discusses, his or herveracity, and his or her attitude towardthe well-being of the receiver.

As one reviews early discussions of thecredibility construct, three points standout. First, the theoretical or definitionalperspectives regarding the dimensionsof the credibility construct were derivedwithout the aid of psychometric tech-nology. Although much has since beenwritten about the source credibility con-struct and its dimensionality, as we willnote later, the basic construct as

originally and subsequently defined,has not been challenged seriously bysophisticated psychometric technology.Second, early authors were addressingthe issue of evaluations of a source dis-

seminating some message in a monologi-cal setting. This body of scholarship didnot direct its attention to dyadic orsmall group communication settings.Finally, there was little or no discussionof the "personality" of the source. Forexample, Hovland et al. when workingwith personality and persuasion wereconcerned with personality characteris-tics of receivers not sources.

RECE:-lT HISTORY OF THE CONSTRr.:c:r

Over the last three decades, there has

5 Carl 1. Hovland. Irving L. Janis. andHarold H. Kelley, Communication and Persua-sion (~ew Haven. Connecticut: Yale UniversityPress. 1953).

been extensive attention directed to-ward the measurement of source

credibility. Prior to that time, measure-ment was crude at best, and often noteven attempted. In 1948, Haiman in hisclassic study of the effect of credibilityon persuasion viewed source credibilityas being composed of the dimensions ofreputation and competence, which areanalogous to Aristotle's character andintellige.nce.6 Haiman's measurement ofthe two dimensions was crude, butlValter, at the same time, attempted todevelop a precise measure of one ofthem, character.'i Subsequent to "Walter'seffort, attention to both the constructand its measurement lagged for over adecacIe. During this time, Osgood's workon the measurement of meaning ap-peared and was widel y disseminated.8His studies made it popular to conceiveof receiver perceptions as being multi-dimensional. Since source credibilitywas conceived as a perception held bya receiver, and since theory and defini-tion had suggested multidimensionality,it was natural to extend Osgood's workvia factor analysis in empirically deter-mining the components of sourcecredibility.

In 1961, two studies were conductedthat had a major impact on subsequentwork. In his doctoral dissertation,

Andersen reported the development ofsemantic differential scales designed tomeasure ethos. These scales purportedlytapped the perceived "authoritative-ness" and "dynamism" of a speaker.9

6 Franklyn Haiman, "An Experimental Studyof the Effects of Ethos in Public Speaking,"Diss. Northwestern UniversitY, 1948.

.. Otis Walter, Jr., "The Measurement ofEthos," Diss. Northwestern University, 1948.

8 Charles E. Osgood, George J. Suci, andPercy H. Tannenbaum, The Measurement ofA[eaning (Urbana. Illinois: University of IllinoisPress. 195i).

9 Kcnneth E. Andersen, "An ExperimentalStudv of the Interaction of .-'-rustic and Non-Artis'tic Ethos in Persuasion," Diss. Universitvof Wisconsin, 1961. .

Page 3: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

26 CE:-lTRAL STATES SPEECH JOURNAL

Later that same year, Berlo and Lemertpresented a convention paper in whichthey reported on semantic differential-type scales used to measure the dimen-sions of competence, trustworthiness,and dynamism.1°

As a result of these two studies, theconcept of dynamism surfaced for thefirst time as a component of sburcecredibility. While we discuss the sig-nificance of this in more detail at a laterpoint, it is important to note that theSGtles which defined this factor (foreX:lmple, bold-timid, aggressive-meek,energetic-tired, active-passive) were notdesigned to measure the evaluative di-mension of meaning as reported byOsgood et aI., but rather were associatedwith the activity and potency dimen-sions. Such items were unrelated to

previous conceptualizations of ethos orcredibility. Source credibility had beenconceived as an evaluation or attitudetoward a source; thus, items includedfrom the activity and potency dimen-sions served only to obscure the con-struct. They probably represented anindependent factor because they werenot associated with the credibility con-struct but were associated with eachother.

In 1966, on the basis of a series of

studies begun in 1963, McCroskey, fol-lowing in the footsteps of Haiman and'Walter, reported both Likert-type andsemantic differential-type measures fordimensions of credibility which helabelled "authoritativeness" and "char-

acter".11 In I969, after further replica-tion of their earlier work, BerIo et aI.published a report providing scales tomeasure three dimensions of credibility.

10 David K. BerIo and James B. Lemert, "AFactor AnalYtic Study of the Dimensions ofSource Credibility." 'Paper presented at the1961 convention of the Speech Association ofAmerica, XeIV York.

11James C. ~(cCroskey, "Scales for theMeasurement of Ethos:' Speech t'l,lonographs,33 (1966), 65-72.

Again, the measure developed was ofthe semantic differential type. Overthis eight year period, two new labelsappeared for dimensions developed,"safety" and "qualification." These twoare essentially equivalent to the "compe-tence" and "trustworthiness" dimen-

sions they originally reported. The"dynamism" label remained the same.1Z

About this time, a substantial numberof papers appeared in speech and com-munication journals concerning thedimensionality of the source credibilityconstruct.13 All employed factor analysesas their primary research tool. Thenumber of dimensions reported variedwidely, and the labels applied to thosedimensions varied even more widely.Many of these studies employed suchsmall samples that their internalvalidity was highly questionable. Othersemployed so few sources that their ex-ternal validity was highly questionable.As a response to this confused situation,McCroskey and his associates. reported aseries of studies conducted between 1971

and 1975, in which they investigated awide variety of types of communicationsources and types of communication re-ceivers with scales drawn from most of

the previous factor analytic studies thathad been reported prior to that time.a

lZ David K. Berlo, James B. Lemert, andRobert J. Mertz, "Dimensions for Evaluating theAcceptability of ~(essage 50urces," PublicOpinion Quarterly, 33 (1969),563.76.

13 See, for example, Jack L. Whitehead, Jr.,"Factors of Source Credibility," QuarterlyJournal of Speech, 54 (1968), 59.63; E. ScottBandhuim and Margaret K. Davis. "Scales forthe ~(easurement of Ethos: Another Attempt,"Speech Monographs, 39 (1972), 296.301; RonaldL. Applbaum and Karl W. E. Anatol, "Dimen-sions of Source CredibilitY: A Test for Re.producibility," Speech Mo~?graphs, 40 (1973),231-37; Raymond L. Falaone, "The FactorStructUre of Source Credibility Scales for Im-mediate Supervisors in an OrganizationalContext," Central States Speech Journal, 25(1974), 63.66; and Christopher J. S. Tuppen,"Dimensions of Communicator Credibility: AnOblique Solution:' Speech Monographs, 41(1974), 253-60.

14James C. :\(cCroskev, Michael D. Scott. andThomas J. Young, "The Dimensions of Source

Page 4: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

ETHOS AND CREDIBILITY 27

The conclusion from most of McCroskeyet aI.'s studies was that there were five

dimensions of credibility: competence,character, sociability, extroversion, andcomposure. Specific scales were recom-mended to measure each of these dimen-

sions on the basis of the type of sourcewhich was involved.

After all of this research effort, itwould be reasonable to expect that thefield would have obtained closure on thenumber of dimensions in the source

credibility construct and the most ap-propriate measures of those dimensions.Unfortunately, quite the opposite ap-pears to be true. New studies are con-stantly appe:lring and almost alwayssuggest some new dimensional structure.Instead of conceptual or operationalclosure, we have calls for additionalstudies to investigate dimensionality, aswell as new methodologies for analysis.15

It is our position that the dimen-sionality of the source credibility con-struct has been sufficiently demonstratedthrough many studies. In addition,satisfactory measures of those dimen-

Credibility for Spouses and Peers." Paper pre-sented at the 1971 convention of the 'WesternSpeech Communication Association; James C.;l.IcCroskey. Thomas Jensen. Cynthia Todd, andJ. Kevin Toomb, "Measurement of the Credibil.ity of Organization Sources." Paper presentedat the 1972 convention of the Western SpeechCommunication Association; James C. Mc-Croskey. Thomas Jensen. and Cynthia Todd."The Generalizability of Source CredibilityScales for Public Figures.'" Paper presented atthe 197~ convention of the Speech Communica.tion .-\ssociation; James C. McCroskey. ThomasJensen. and Cynthia Valencia. ":'vIeasurement ofthe Credibility of Peers and Spouses." Paperpresented at the 1973 convention of the Inter-national Communication Association; James C.~fcCroskey. Thomas Jensen. and CynthiaValencia. ":Yfeasurement of the Credibility ofMass ~fedia Sources." Paper presented at the1973 'Western Speech Communication Asso-ciation; and James C. McCroskev, WilliamHoldridge. and J. Kevin Toomb. ';An Instru-ment for ~Ieasuring the Source Credibility ofBasic Speech Communication Instructors,"Speech Teacher, ~3 (1974). ~6-33.

I;) Gary Cronkhite and Jo Liska. ".-\ Critiqueof Factor Analytic Approaches to the Studyof Credibility," Communication Monographs,43 (1976),91-107.

sions have been available for manyyears. We believe the cause of the cur-rent confusion is a distortion of the

source credibility construct which oc-curred in the early 1960's and has beenperpetuated to the present.

THE DrSTORTION OF THE CONSTRUCT

Prior to 1960, it was clearly recognizedthat the construct of ethos or sourcecredibility involved the evaluation ofthe source on the part of the receiver.Both theoretical and methodologicalpapers held to this tenet. The work ofBerlo was singularly influential in tak-ing later researchers beyond the originallimitations of the construct. Probablybecause of the popularity of the work ofOsgood and the semantic differential atthe time of his early research in 1961,Berlo, as previously noted, includedscales in his research which had no

apparent isomorphic relationship to theconstruct under investigation. In par-ticular, these scales were designed tomeasure the activity and potency di-mensions of meaning that had beenisobted by Osgood. Inclusion of suchscales, and the use of factor analytictechnique, resulted in the "discovery"of a factor subsequently labeled dy-namism. While noting that his primaryconcern was with the me:lsurement of

the "evaluation of message sources,"Berlo persisted in maintaining a dy-namism dimension of source credibilityeven in his 1969 report.

This "dynamism" dimension ap-parently had a direct bearing on the de-velopments in subsequent rese:lrch. Itshould be pointed out that Berlo et aI.,were hesitant when discussing this di-mension. They stated that". . . the rela-tive instability of dynamism suggeststhat it may not be psychologic:llly in-dependent of the other two factors."16

16 Berlo. Lemert. and :'v[crtz.

Page 5: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

28 CE~TRAL STATES SPEECH JOURNAL

McCroskey in his 1966 report, whendiscussing dynamism, made two crucialpoints. First, he suggested that adynamism dimension did not appear inhis work simply because there were noscales included designed to test thatdimension. Second, and perhaps moreimportant, he pointed out that theearlier definitional and theoretical work

suggested that "ethos" or "sourcecredibility" was held to be an attitudeof a receiver toward a source. And since

Osgood's work suggested that the evalua-tive dimension of meaning was represen-tative of attitude, their work withcongruity would have been confoundedby a "dynamism" dimension.!' It is im-portant to note that these two earlyresearch reports did not differ sub-stantially in their findings. There weretwo clear and stable dimensions of

credibility reported, but dynamism,while being statistically independent,was psychologically unclear and relative-ly unstable.

At about this time studies which werenot concerned with .source credibilityper se, but which employed factoranalytic procedures, also appeared.Most notable were ones by Markhamand Norman,18 Both of these were con-

cerned with personality perception, notsource credibility, and the Normanstudy, in particular, stressed that theconcern was not with evaluation. Their

effect, however, appeared to be furtherexpansion of the source credibility con-struct.

17 McCroskey, "Scales for the :\feasurement ofEthos." .

18 Dayid Markham. "The Dimensions ofSource CredibilitY of Television Newscasters,"Journal of Communication, 18 (1968), 57.64 and

.Warren T. Norman, "Toward an AdequateTaxonomy of Personality Attributes: ReplicatedFactor Structure in Peer Nomination PersonalityRatings," Journal of Abnormal and Socia'lPsychology, 66 (1963), 574-83.

IDRaymond K. Tucker, "On the :\rcCroskeyScales for the :\fe:J.surement of Ethos." Centra'lStates Speech Journal, 22 (1971), 127-29.

In 1970, Tucker published an articlewhich called for caution in the use ofthe scales that were available at that

time to measure source credibility.19 Heproperly noted that semantic differen-tial-type scales may take on differentmeaning as a function of the conceptwhich they are designed to measure.Unfortunately, the recommended solu-tion for this problem was for the re-searcher to factor analyze his or her datain each study to be sure that thedimensionality held stable. Such anapproach, however, will guarantee a lackof generalizability from study to study,since presumably the dimensionalitywould change from study to study.

In a well-intentioned, but we now be-

lieve misguided, attempt to respond tothe problem pointed out by Tucker,McCroskey and his associates set out totest generalizability of scales for themeasurement of source credibilitv across

types of sources and types of subjects.While in his earlier report, McCroskeyhad challenged the validity of thedynamism dimension and argued forthe existence of only authoritativenessand character dimensions, in this series

of studies he ignored his own advice andinclucIed scales to measure that dimen-

sion. In addition, he drew upon scalesthat had been reported by Markhamand Norman which were never intended

to measure source crecIibility. As a re-sult, the McCroskey series produceddimensions which were fairly generaliz-able across source types and acrosssubject types, but their isomorphismwith the credibility construct wastenuous at best.

Recently, Cronkhite and Liska havepresented a critique of the sourcecredibility research.2O While we will notelaborate their position here at length,the basic thrust of their argument was

20 Cronkhite and Liska.

Page 6: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

ETHOS AND CREDIBILITY 29

that if we were to do our factor analyticresearch correctly. we would find thetrue dimensions of source credibility.We take issue 'with the point. Much ofthe factor analytic work has been doneproperly. The major problem with theresearch is the distortion of the con-

struct prior to collection of the data. Aswe have noted elsewhere. the work onsource credibility possibly presents thebest example of the abuse of factoranalysis extant in the communicationliterature.~l To illustrate. let us reportwhat several years ago might have beenconsidered the "ultimate factor analyticstudy of source credibility." So that weare not misunderstood. let us stress thatthis is a real study conducted with realsubjects, and not a fictitious examplethat we have created here.

A.'1 ILLt:STRATIVE STUDY

Scale Selection

Scales were generated for use in thisstudy by three methods: I) a sample of726 college students enrolled in basiccommunication classes were asked to

provide adjectives to describe "theperson you would be most likely to be-lieve" and "the person you would beleast likely to believe;" 2) the theoreticalliterature concerning ethos and sourcecredibility was surveyed to find adjec-tives most frequently employed to de-scribe credible and non-credible sources;

and 3) previous research concerning themeasurement of source credibility andethos was reviewed to determine the

adjectives most commonly found to ap-pear in the various studies.

This procedure resulted in selectionof thirty pairs of bipolar adjectives.' Tothis list we added six pairs of bipolar

~l James C. ~IcCroskev and Thomas J. Young,"The Cse and Abuse of Factor :\nalvsis in Com.munication Rese:lrch," Human COTnmzmicationResearch, 5 (19i9),3i5-82.

adjectives thought to measure the per-ception of size. since it has beensuggested in previous research that sizehas an impact on credibility.22 In ad-dition, we added five pairs of bipolaradjectives that we believed wouldmeasure nothing at all related to sourcecredib ili ty.

A total of 2,057 students enrolled inundergraduate classes in communicationserved as subjects in this study. None ofthe subjects who provided adjectivesinitially were included. Each subjectresponded on the 41 scales with regardto a particular source. Six different typesof sources were employed: 1) peers,operationalized as "the student youtalked with most recently," N = 372; 2)spouse, operationalized as "the personyou date most often or if married, yourhusband or wife," N = 352; 3) a major

organization, operationalized as twelvedifferent organizations within theUnited States (the particular organiza-tion to which a student responded wasdetermined randomly), N = 344; 4)a media source, operationalized astwelve different radio, television, and

print sources (the particular source towhich a student responded was de-termined randomly). N = 331; 5) po-litical figure, operationalized as twelvedifferent political figures, currentlyprominent in the election being held atthe time the data were collected (thesource to which the student respondedagain was determined randomly), N =323; 6) teacher, operationalized as theteacher of the class in which the student

was currently enrolled (a total of 15different teachers was used). N = 335.

Data Analysis

The data were subjected to principal

~~Eldon E. Baker and W. Charles Redding."The Effects of Percei\'ed Tallness in Persuasi\'eSpeaking: An Experiment," Journal of Com-munication, 12 (1962).51-53.

Page 7: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

30 CENTRAL STATES SPEECH JOURNAL

components factor analysis and varima."{rotation. An eigenvalue equal 1.0 wasestablished as the CUt-off for rotation.

For a solution to be acceptable, werequired that each factor have aminimum of two scales with their pri-mary loading on that factor. In addition,at least two items had to have a loadingon that factor of .60 with no other load-

ing above .40 on other factors. Thesecriteria were selected because they werehighly representative of decision criteriaemployed in previous studies. Moreliberal criteria would not have alteredthe obtained results.

Results

An eight-factor solution was obtainedwhich accounted for sixty-four percentof the total variance. Table I reportsthe items and their factor loadings. Asnoted in Table 1, the factors werelabeled as follows: sociability, size,extroversion, composure, competence,time, weight, and character. Subsequentanalysis of the data for each source typeproduced results extremely comparableto those reported in Table I and thuswill not be reported here.

Conclusions

There are clearly eight dimensions of"source credibility" reflected from theanalysis reported above. In the traditionof prior factor analytic studies withsource credibility, it is clear that wehave confirmed the presence of the fivedimensions that the McCroskey studieshave demonstrated previously: sociabil-ity, competence, extroversion, com-posure, and character. In addition, wehave "discovered" three new dimensions

of source credibility: size, weight, andtime.

While the preceding interpretationmay seem ludicrous' on its face, it isquite comparable to that provided by

many previous factor analytic researchersconcerned with source credibility. Inpoint of fact, we have confirmed nodimensions of credibility at all, and wecertainly have not discovered any newones. We have as strong an empiricalbase for arguing a dimension called"general size" or "time," however, as wedo for one we call "extroversion" or

"composure".Why have we gone to such lengths to

collect data that we now say really donot demonstrate much of anything? vVehave done so to illustrate that if a con-

struct is not carefully defined, and aresearcher does not closely adhere tothat definition, massive amounts of dataand sophisticated computer resourcescannot lead to the conclusion that onedimension or another exists. That is not

to say that if ethos or source credibilityresearch had begun simply with a ques-tion, such as "How do people perceiveother people?", structures such as theone we found might not have emerged.However, the ethos or source credibilityconstruct has been carefully defined inthe theoretical literature, and early re-search attempted to verify empiricallyits existence and to provide reliablescales for its measurement. When we

begin adding other scales, scales whichbear no resemblance to the originallydefined construct, it doesn't somehowchange the construct. At best it suggeststhe credibility construct is but a subsetof all that which is perceivable. We mustremember that methodological researchon a construct is designed to improvemeasurement, not to alter the nature ofthe construct.

THE CONSTRCCT THEN .-\.'..-0 Now

Since we have gone to considerableexpense in terms of time and effort toillustrate the futility of massive factoranalytic studies to determine the dimen-sionality of source credibility, it should

Page 8: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

TABLE I

FACTORLOAI>INGSFOR 41 ITEMS(N = 2057)-----------

Item Sociability Size Extro- Compo- Competence Time Weight Characterversion sure

in telligen t.uni ntelligent . . . . -.73 . . .nntrained-trained . . . . .74 . . .expert-inexpert . . . . -.78 . . .nlliuformed-illformed * * * * .73 * . .('om peten t -i ncompetell t * . . * -.67 . * .stupid.hright -.30 . . . .71 * . .timid-hoh! . . -.74 * . . . *verhal-quiet . . .8.1 * . . . .ta II: a ti vc.silent . . .81 . . . . . :I:mcek-aggressh'e . . -.76 . . . . . 0ex trovertcd.in troverted . . .67 . . * . . ennot dyuamic-dynamic -.30 . -.50 . AI . * . >Zsinful-virtuous . . . . * . . .69dishonest-houest -.33 . . . .33 . * .67 C")IlilSeifish.selfish .43 . . * * . . -.50 asympa tlletic- ullsympa thetic .59 . * . * * . -.40 ....high character-low character .47 . . . -.13 . . -.47 tdUlltHlstworthy. trllstworthy -.42 . . . .35 . . .58 ....

t""....Ilervolis-poised . . . .74 . * . . 0<relaxed-tellse . . . -.80 . . . .calm-allxious . . . -.81 * * . .cxcita ble.composed . * . .68 . . . .uptight-cool -.33 . . .69 . . . .C(IIItrolled-fearful . . . -.62 * . * .sociable-unsociable .74 . . . . * . *uufriendly. frielHlly -.78 * . . * * * .cheerful-gloomy .77 . . . . * * *goodnaturcd-irritable .80 . . . * . * .cold.wann -.81 . . . . * . *pleasaut-unpleasant .79 . . . . . . .large-small . -.83 . . . * . *lillle.hig * .84 * . . * . .

I.>CI...

Page 9: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

lall-shOl'lskinny-fatlight.heavyhnge-tiny

black.whiteday-nightearly-latehard-softfew-many

-.42..Loadings< :!:.30 are omitted.

.......

TABLE 1- (Continucd)

-.80. ...31

-.66..

.. ........

---'--

....

....... .37....

---------'--

l>Ot-o

...(")t>1Z..,

~l""

en

~..,

~en'1;jt>1t>1("):I:

....0~~Z>t"

. .. .. .. .. .. .83. .83. .. .30--""""'._-_.

. ..88 .$1 .

-.16 .

Page 10: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

ETHOS AND CREDIBILITY 33

be apparent that we are prepared toargue that there is a better method.That indeed is the case. Although webelieve that factor analysis is a veryuseful technique, it cannot be a sub-stitute for the essential ingredient of thehuman critical capacity. 'We believe theesablishment of the dimensionality ofthe construct, and in this case that ofsource credibility, must follow a two-step process. The first step is to definecarefully, through critical evaluation,the nature of the construct to be

measured. "\Ve believe this step wassatisfactorily completed by Aristotle,and subsequently by Hovland and hisassociates. The second step in thisprocess is carefully designing instru-ments to measure the theoretical di-mensionsof the construct. In this case,these theoretical dimensions can be

labeled competence, character, and goodwill or intention. Once such measures

are developed, data can be collectedfrom subjects and submitted to factoranalysis to verify that theoreticallyuniq ue dimensions are indeed at leastpartially independent from one another.

This procedure was originally fol-lowed by McCroskey in a series ofstudies reported in 1966. At that timehe found that the dimensions of

character and good will were not in-dependent perceptions. In the currentinvestigation, there were scales includedwhich were designed to measure thethree theoretical dimensions of compe-tence, character, and intention. Onceagain, the scales designed to measurecharacter and intention would not formseparate clusters. Rather, two dimen-sions appeared from this group of scalesand can be labeled "competence" and"character". The related scales and their

appropriate factor loadings from a two-factor solution are reported in Table 2.

It is our conclusion, therefore, thatwhile theoretically there are three di-mensions in the source credibility orethos construct, in terms of empiricallybased perceptions, these three collapseto two. This is not to suggest that in-tention is unimportant, but only that aperception of the source's intention isdependent on perceptions of thatsource's character.

The important point is that sourcecredibility must be distinguished fromother perceptions that people have ofother people. We are not arguing thatpeople do not perceive other peo-ple's extroversion, composure, sociabil-ity, general size, or for that matter theirattractiveness, their similarity, or any of

TABLE 2

FACTOR LO.-\DI:SCSFOR CO~PETE:SCE AND CHARACTER ITE:>rs, Two-FACTOR SOLUTIO:>;--

(N = 2057)

Item

in telligen t-uninteIligen tuntrained-trainedexpert-inexpertuninformed-informedcompetent-incompetentstupid-brightsinful.virtuousdishonest-honestunselfish-selfishsvmpa the tic- unsympa thetichigh character-low characterun trustn-orth v-trusnvorth v

'Loadin~ below .30 are omitted."Correlation between dimensions = .58 based on unweighted scores.

FACTOR L1BEL

Competence Character Communality

.74 . .61-.77 . .62

.81 . .68-.77 . .63

.68 -.33 .57-.73 .37 .67. .57 ;35

. .76 .67- -.7-f .57- -.75 .58

Al " .69-./--.32 .76 .68

Page 11: OqG ETHOS AND CREDIBiliTY: THE CONSTRUCT AND ...jamescmccroskey.com/publications/096.pdfETHOS AND CREDIBILITY 25 munication and persuasion during and after World War II, suggested

M CEXTRAL STATES SPEECH JOURNAL

a wide variety of other perceptions.Rather, we are arguing that we mustseparate those perceptions which arerelated to source credibility from thosewhich are not. If we are interested in

perceptions of attraction, we shouldmeasure that directly and work with theattraction construct. Such scales areavailable.23 If we are concerned. with

perceived interpersonal similarity, weshould measure that direcrly, and again,there are scales available for that pur-pose.24 Similarly, if we are interested inperceived extroversion, composure, orsociability, the scales reported in Table

, I that are related to those perceptionswould be appropriate. But sourcecredibility and person perception arenot the same thing. Source credibility ismerely a subset of a much larger con-struct of person perception. For otherfacets of the person perception phe-nomenon, the reader may consult theextensive body of literature in thatarea.25

?fEASL'Rl1':G THE CREDIBILITY CONSTRUCT

As we noted previously, we believethat the construct of ethos and source

credibility has been amply defined inthe past. We also believe that there areadequate measures of that constructreadily available. The Likert-type scalesfor measuring authoritativeness (orcompetence) and character which Mc-Croskey developed have proved useful,particularly in assessments of public

2:\James C. :'.IcCroskey and Thomas A. Mc-Cain, "The ~[easurement of InterpersonalAttraction," Speech Monographs, 41 (1974),261-66.

2-1James C. McCroskev. Virginia P. Richmond.and John A. Daly. "The Development of aMeasure of Perceived Homophily in Inter-personal Communication," Human Communica.tion Research, 1 (1975), 323-32.

2~ See. for example. David J. Schneider. "Im-plicit Personality TheOI"V: A Review," Psy-chological Bulletin, 79 (1973). ::!94-309.

figures.26 These scales have had veryhigh internal reliability, and their usein a wide variety of studies over the past15 years indicates their predictive andconstruct validity. Minor modificationsin the wording of these scales would benecessary for use with other types ofsources. For researchers who wish tohave more concise measures of the two

dimensions of source credibility, thebipo'lar scales reported by Berlo orMcCroskey for these two dimensions orthose reported in Table 2 are adequate.In all three instances, reliability in theneighborhood of .80 should be expected.

FUTURE DIRECTIONS

It is obvious from the preceding dis-cussion that we believe it is time, onceand for all, to call a hal t to theproliferation of factor analytic studiesof source credibility; that is, those in-tended to "discover" its dimensions. We

believe further studies of this type willserve no useful purpose, and in fact willcontinue to direct our attention awayfrom more worthwhile pursuits.21 Hope-fully, as a field we will learn from ourexperiences and misdirected efforts.Sophisticated statistical analyses andhigh speed computers cannot substitutefor the critical capacity of humanbeings. Factor analysis must be precededby careful conceptualization and con-struct delineation, or the product ofeven the most massive research effort, as

we hope we have illustrated above, willlead to nothing, or worse, to inappro-priate knowledge claims.

26 McCroskey, "Scales for the Measurement ofEthos."

2; Although the rationale set forth in thisessay was developed prior to the publication ofa paper by Jesse D. Delia ("A ConstructivistAnalysis of the Concept of Credibility," Quar.terly Journal of Speech. 62 (1976), 361.375). thereader should consult this article for a similarview. 'Ve concur with Delia's analvsis of theproblem. but take exception with 'his recom-mended solUtion.