PSYCHOLOGICAL BULLETINVOL. 55, No. 4, 1958CONTENT AND STYLE IN PERSONALITY ASSESSMENT1DOUGLAS N. JACKSONPennsylvania Stale UniversityAND SAMUEL MESSICKEducational Testing Service and Princeton UniversityIn personality theory a ubiquitousand fundamental distinction may bedrawn between the interpretation ofbehavior in terms of (a) the contentof "needs" and of cognitive structuresgenerally and in terms of (b) charac-teristic styles of response and action.The separation of these two compo-nents of personality organization hastaken a variety of forms in the handsof different theorists, as in the All-port-Vernon (2) Studies in ExpressiveMovements, in Murphy's (47) schol-arly discussion of continuity in per-sonality structure, in Klein's (40) dis-tinction between needs and controlprocesses, and in Vernon's (54) dis-tinction between adaptive and ex-pressive behavior. One may legiti-mately ask not only what a personsays or does (the particular contentof his statements and actions) buthow he acts (his characteristic modeor style of expression).What is conceptually a relativelysharp distinction is typically blurredand confounded in a particular con-crete act; the what and how are fusedin a given goal-directed response. Anobsequious person indicates his def-erence not only by the act of yield-ing, but by the tone of his voice inperforming the yielding act. Becausecontent and style are intermixed in a1 Portions of this paper were read at a sym-posium on "Experimental Approaches toPersonality Assessment" at the AmericanPsychological Association Meetings in NewYork, 1957.The authors express their thanks to LeeSechrest and Riley W. Gardner for comment-ing on the content and style of the manuscript.given behavior sequence, and be-cause there is often a theoretical pre-dilection for content components,style is often overlooked in person-ality assessment. Also, the measure-ment of content appears to be moredirect and unambiguous than the as-sessment of stylistic dimensions ofpersonality. It is possible, for exam-ple, to ask a person what his attitudeis on a given topic, or to draw infer-ences about his need patterns fromhis reported likes and dislikes (51).The obviousness of such devices,while helpful from the viewpoint oflabeling what one hopes one is meas-uring, also permits respondents todistort their scores if they so desire(32), something which is less likelyto occur in the assessment of style.In considering the general distinc-tion between content and style, thosemethods of personality and attitudeassessment which are based uponprinted questionnaires of one form oranother will be emphasized. Whilethe complementary constructs ofcontent and style have special rele-vance to questionnaire items, wherethe response-evoking properties ofthe particular item form may contrib-ute markedly to response varianceabove and beyond the contribution ofcontent, the distinction might also beapplied usefully to other areas of per-sonality assessment. For example,three possible applications are to per-ceptual and cognitive style as in thework of Thurstone (52), Witlcin (58),Klein (39, 40), Gardner (21), andothers (34); to achievement and apti-243244DOUGLAS N. JACKSON AND SAMUEL MESSICKtude testing (28, 32, 60); and to theperception of personality (2, 38, 54,59).The present discussion attempts todo two tilings: first, to present someevidence showing the important andsubtle influences upon responses ofstylistic components of item form;and, second, to illustrate how reliablemeasures of potentially useful stylis-tic dimensions may be generatedfrom characteristic responses to theform of personality and attitudeitems as distinct from measures ofcontent.PERSONALITY STYLE ANDRESPONSE SETTraditionally, responses to a par-ticular item or set of items are as-sumed to provide information aboutthe respondent in terms of the itemcontent. If, for example, a personagrees with the statement, "Under noconditions is war justified," or an-swers "true" to the item, "I havemore trouble concentrating thanothers seem to have," it is commonlyassumed that these responses, if con-sistent, will indicate respectivelysomething about the person's atti-tude toward war or his mental state.Under these conditions response de-terminants such as the subjects' gen-eralized tendency to agree are legiti-mately considered as sources of cu-mulative error, Cronbach's (13, 14)familiar "response sets." WhileCronbach's emphasis was that re-sponse sets often lead to errors of in-terpretation in the logical validity oftests, he also indicated that these re-sponse tendencies might not alwaysbe temporary and trivial, but mayhave a stable and valid componentwhich reflects a consistent individualstyle or personality trait. While rec-ognizing Cronbach's contribution indescribing the phenomenon, it is pref-erable for the present purposes tochange the label from "response set"to components of style. This changein terms emphasizes the fact that forcertain purposes in personality as-sessment opportunities for the ex-pression of personal modes for re-sponding should be enhanced andcapitalized upon, rather than consid-ered as sources of error to be avoidedor minimized. This change alsoavoids the ambiguity inherent in theconcept of "set" (22).CHARACTERISTIC STYLES INPERSONALITY AND ATTITUDEQUESTIONNAIRESAmong the more prominent re-sponse styles usually evoked by ques-tionnaire items are response acquies-cence, overgeneralization, a tendencyto respond in a socially desirable way,and the complementary tendencies torespond negativistically, critically,and in a socially undesirable or idio-syncratic manner. Some pertinent il-ilustrations will be drawn of howeach of these, operating singly and incombination, may influence the in-terpretation of responses to psycho-logical tests. Alternative proceduresfor evaluating these stylistic varia-bles will then be discussed.Response Acquiescence and Authori-tarianismIt has been long recognized that asubject who agrees with a personalityor attitude item stated in a positiveform may not necessarily disagreewith its logical opposite, but may in-stead show a fairly general tendencytoward agreement or disagreement.Studies by Rundquist and Slctto(49), by Lorge (42), and reviews byCronbach (13, 14), Berg (8), andMessick and Jackson (45), indicatethat response acquiescence is wide-spread and pervasive over a wideCONTENT AND STYLE IN PERSONALITY ASSESSMENT245variety of item content and most pro-nounced when content is highly am-biguous or