QUANTITATIVE CONTENT ANALYSIS: A METHODOLOGICAL HANDBOOK WITH EXAMPLES FROM RES

Created: 3/1/1977

OCR scan of the original document, errors are possible

QUANTITATIVE CONTENTETHODOLOGICAL HANDBOOK WITH EXAMPLES FROM RESEARCH ON THE SOVIET UNION

lOCUj7

copy

^ jpa-o'ill ITT liiMiim

CENTRAL INTELLIGENCE AGENJTYjofl ond "oll'ical Anoh/iii

QUANTITATIVE CONTENTETHODOLOGICAL HANDBOOK WITH EXAMPLES FROM RESEARCH ON THE SOVIET UNION

Tlii- uutluw ii urntoful for lhc muny cummTttitl turn Mkn CIA offlorl -nd athn Rnvrmmrntal and acwlrmlc BttnrlalKi during fhf pnvmNoa nf ihh, pttdt Ownnwn'.he aulho

BLANK PAGE

PREFACE

Social scientists hnvpariety (tf systematic, methods, many of them quantitative, which are now used broadly In academic research hut have yet tolace in political research conducted hy the Intelligence community. This Is Hue in part to the differences in prolilem and purpose between |Mtlicy-oriented and theory-oriented research. In partime lag In udaptlng nnd applying unfamiliar procedures. This study of quantitative content anulysis Is part of anprogram In the Office of Regional nnd Political Analysis to examine nnd test the npptlcnhlllty of such methods to Intelligence research.

A previous study. "Measuring Support for Brezhnev: An Empirical Test of Some KremllnologlcolR,llustrated thv application of quantitative content analysis to political intelligence research. The present study deals with the methodology per se.nd II descrlhe what quantitative content analysis Is and discuss Its limitations and potential uses. They arc hroadly applicable regardless of one's country focus. Sections HI and IV discuss some special circumstances of dealing with Soviet source materials and suggest specific applications to research on the Soviet Union. Although writtenpecifically Soviet context, much of the content of these sections Is hroadly applicable to other geographicalengthy Appendixariety of specific approaches to content analysis nnd will be nf particular Interest to analysts considering uctual research applications of the technique. Technical terminology characteristic of quantitative methodology has been carefully avoided.

The mithnr will he pleased to provide, upon request,assistance to anyone In the Intelligence community Interested in using quantitative content analysisesearch tool.

ill

rm ornsiu ucc-owrr*

BLANK PAGE

*G* UTTtOM use onTy

CONTENTS

EXECUTIVE

THE

I. WHAT IS CONTENT

II. WHY QUANTIFY

Evidence t" Support Cnntrnvenlnl

SumrnnriMnR and AflKrenatlnjr. Dnlii

StuilMlcal Analyse;

SOURCE MATERIALS

Soviet Commun lent

Specific Sources

TO RESEARCH ON THE

Perception Analyus

Elite Analyst!

Political

Analysis of "Esoteric Communications"

[HycUlojtlcal

APPENDIX; APPROACHES TO QUANTITATIVE

CONTENT ANALYSIS

job UIMLIU USC QNtt

BLAMK PAGE

EXECUTIVEr

Content analysis refers to nny procedure for making Inferences hy systematically Identifying, categorizing, and recording specifiedof written or spoken communications. Although quantification is not an essential clement, most content analyses do In fact involve numerical measures of communications content. This report focuses specifically on these quantitative approaches.

The nature of the evidence obtained by quantitative content analysis is perhaps best understood by comparing it with the evidence obtained by other quantitative research techniques such ns opinion polls, questionnaires, ami certain pwclmlnglcul tests. The content analyst. In effect,e then examines speeches. Journal articles, official documents,nd checks the appropriate box on the questionnaire each time he finds nn "answer."

Content analystsamily of related techniques adaptableide var'ety of specialized purposes. Thecharacteristics of interest to the researcher might be specific themes Indicative of Micfs, attitudes, perceptions, or goals; groups of words orerhaps Indicative of psychologicalheof two or iiorc wordsof patterns ofr even the types of logic employed. And once tlie specified characteristic has been identified, the analyst may note its presence, count its frequency, measure its intensity, or. in the case of newspaper articles, assign an attention score based on amount of space, size of headline, and/or page position. The search for these textualmay he undertaken either manually or by computer. The Appendix dcscfilKrs these different approaches to content analysis and cites examples from the study of Soviet, andesser extent Chinese,and Journuls.

Content analysispecialized technique applicable only to particular smts of problems. It is essentially descriptive and helps to explain the intent nf the communicator only iusofur as uccunite descriptionfur qua non of accurate explanation. Ukc opinion polling nnd nthrt measurement techniques. It Is useful primarily for

lionM gimw

nut king precise comparisons between Individuals or groups, or between time periods. A. principal limitation is the tedious and generally time-consuming nature of tlte work required to code und tabulate all elementsext which fall Into the specified categories of Interest. This effort must be Justified by significant benefits beyond those achievable by conventional analysis.

Quantitative content analysis seems especially- appropriate for Communist urea studies, because of the Inaccessibility ot Communist elites to more conventional forms of research, and because the priority accorded lo Soviet and Chinese studies may Justify the additional investment of time and effort. But there are problems of Accessibility to political elites in all countries, and content analysis can be used to clarify analysts' Judgments aboW leaders or groups from other geographical areas as well.

Why Choose This Method?

One impetus to engage in systematic content analysis is the need to adduce more persuasive evidence in support of analytical conclusions that arc unexpected or controversial. In conventional analysis, conclusions about trends In content over time or differences in content In'tween Individuals or groups are normally sup[>ortcd by selecting nnd citing representative passages of the text. Rut the evidence on many points Is inconsistent ond contradictory, and it is often possible to cite other passages which seem to support an opposite conclusion. Quantitative content analysis Is more systematic than conventional analysis, and it Is potentially more objective and persuasive. In that It categorizes and tabulates all appearances In the text of the characteristics deemed relevant to the issue at hand.

A second reason for engaging In the more tlmc-consumlng procedures of quantitative content analysis may be the benefits of quantification per sc. Counting of frequency or measurement of Intensity transforms qualitative Information into quantitative data. Quantification Is normally unnecessary, but there are occasions when the Increased precision Is clearly beneficial. Quantification Is essential If one wishes to take advantage of statistical procedures foruantified variable to other numericalnalyze the relationship between age (political generation)uantitative measurement of the political attitudes of Soviet elites.

Content analysis Is similar to opinion polling in that It can Ih- used to aggregate themes expressed hy individuals to arriveuantitative description of the beliefs, attitudes, perceptions, values, or goals of a

group. Summing Individual statements to Infer group characteristics Is hy Its veryatter of counting the frequency of expressions, for the group Is defined as the sum of Its individual members. Thus quantitative content analysis Is well suited to measuring differences in Issue orientation between groups (distinguished by age, occupation, nationality,r to measuring change lu the.same group from orre time period to another. This aggregating or syrropslzlng of.fhcmcsVus many potential applications. Several of these, as" wctf as other applications of quantitative content analysis, are summarized below and then discussed in greatei detail In Section IV.

Elitene may measure the extent to which diverse cliti groups, such as the Soviet partyhe economic managers, and the military agree or disagreeariety of issues. Quantification of these attitudinal data then permits correlation of attitudinal information with the more readily available data on age. education, social and ethnic origin, and career background. For example, one can test hypotheses concerning the relationship between political attitudes nnd age. How significant are the differences between political generations In the USSR, and what arc the long-range implications of these differences?

Perception Analysis: In analyzing. Soviet perceptions, quantitative content analysis facili'ates going beyond such

generalizations as "the Soviets believeall

relevant stutcmcnts arc systematically recorded In thecategories. It Is easier to cope with diversity andof views. The quantitative results may beIn vurlous ways to differentiate which Individualsperceive what, when, with what frequency, and

Thl"lnlW* lhc mirrnt wlalpmmhurWUiiIlm<iinn> tlon* lit thr ini. Full ttlntww" mr IitiiiwtWt <ifal thr rral nl the

Political Indicators: We have economic Indicators to measure the state of the Soviet economy, but no comparable political Indicators to measure the state of the political system. Development of such Indicators to measure evolutionary change presents major conceptual ond methodologicalhut content analysis docs offer promise in this field.oik was an Initial attempt In this direction. He used content analysis In an effort to chart gradual changes in the Soviet systemeal" ranging from total party domination to political pluralism.

Psychologicalather different usage of content analysis is for indirect psychological assessment of political leaders. Since languageroduct of unconscious as well as conscious processes, thereumber of theories ond content anal' lie procedures for relating patterns of wordo psychological attributes, These arc described primarily in the Appendix. Psychol ingu 1st ic analysis of political leaders' public statementsew and relatively undeveloped field! The most relevant techniques require further validation, but the existing state of the art Is sufficient for experimental application.

Kremlinohgy: Kremllnologlcal analysis of the "esoteric communications" characteristic of public political discussion in the Soviet Union may be classifiedorm of content analysis whenever the rules of evidence and Inference are made explicit. The feasibility of quantitative approaches to Kremlinologicat analysis was illustrated by the recent CIA study on measuring support for Brezhnev

In sum. content analysis is an effective research tool whichimited but nonetheless useful range of application to intelligence research.

the discussion .

WHAT IS CONTENT ANALYSIS?

Contentnv of theundamental Io mml foreign policy re*earch. a* mnfl such studies draw Infercncci from on analysis nf Ihe content nf leaders* statements or articles Innewspaper! ond iprctollrcd Journals. The term has tome tn hehowever, ond L* used In this report,peciallred meaning. It refersf "making Inference* hy objectively ond systematically Identifying specified characteristics nf messages". Although acudemle definition* of content analyst* have varied over the yean, "we canontent analysis If the procedure*for processing the text arc well enough defined soecond person going through tlie same text would km! the samr pauages Into (he same Interpretive Ikucs".

Gwitent analysisystematic technique for noting whnt words ond themes are used tn selected speeches and documents. It makes explicit what Is often left Implicit In conventional modes nf analysis of the tamet specifics what communication* have been examined and precisely what characteristics nf the text ore considered to be Indicative of whnt, It then notes the presence or ahsence. or measures the frequ-ncy ond/oi the Intensity nfristic In order to make inferences about the origlnatni nf the communication or his audience. The rcults are usually presented In quantitative or at least graphic lorm to permit precise comperlsnn of two or morenf the same characteristic at different time, oruantification is not an ewntlal element of contentut most content analyse* dn generate numerical datu and this study I* concerned exclusively with that form n( analysis.

Ontent analyst*rvorch tool: like anyIt con lie used either well or badly, defiemllng upon the Imagination and skill of the craftsman.

Quantitative content analysis I* normally used to makehat Is enmpured Is. of course, determined hy the purposes nf the Individual re*enrchrr. He con measure change hy computing the value of the samexpressed attitude toward detente or Inward production of comumer goods, forame Individual (or group or country) at different point* In time. Or he can measure difference* by comparing the value nf llir same variable for tsvo or more Individuals (or group* ort thepoint In lime. One might wish to compare Individuals (Rnvhnev vs.ationalities (Great Ruwian* vs.lltr group* (parlyvs. technicalge group* (older v* youngerr countries (USSR vs.Hurope or ther the researcher may wufa tnhe rettlttondtti* lietween Iwo illffrrentow dn attitude* Inward detente correlate with attitudes concerning the way* of achieving economic growth nr wilh opinions alHiutf the party In maintaining social discipline'*

f the tubjecti of our Inquiries are to hroad-irtndrd at to rpeak of different roads lo toriallsm. toe shoulddo them by (nrtirmj on different roadt to. 0)

The procedure employed In content analysis may Iv clarified by di*ciis*lon nf the following key concent*.

Source Sampler If the quantity nf -elevnnl source muteriul I* too great tn Ik* pfocr**cd In It* entirely, n* I* commonly the case, systematic sampling procedure* are employed In much the same way dint un opinion poll Is directedmall sampleepresentative of the totul population, lodge (lOtOXmplc. sampler! the content ofoviet Journal*23 In order to Identify, compare, and

ONLY

measure ehuitge* over Hmr In 'he view* nf five Sovirl eliterample uf equal0 paragraphs) for each of thr five elite* for each of right sampled0earsata base consisting0 paragraphs) These paragraph* wrre laken from the lead articles in Issues of the relevant lournals, and the Issues from which they were drawn were spaced out to provide equal coverage of 'he months and days of the week during the sampled year.

Indlcaton? The Intelligence analyst Is familiar with the concept of early warning Indicators. Observable actions are onulyred as Indirect Indicators of something which cannot be observedIntention in Initiate hostilities, Similarly, manyvurinbles nf research Intrrcst cannot hedirectly, so the content analyst devises indlcuion to observe and measure them Indirectly..) refers to thisrocess of trai.slallon. The quantitative researcher mustheused In traditional scholarship Into terms susccn-tlhle of mathematicalhus Indicators may he used by the analyst as surrogates for variables "not otherwise accessible to systematic observation, with the obvious qualification that thry may notalkl pleasure of the full complrilty of theunder Investigation.

Indicators may he very simple. For) study nf Soviet perceptions distinguished policies motivated hy national Interest from policies motivated by Ideological Interest by noting whether references were predominantly to Soviet (or American) goals and policies or to communist (or capitalist) goals and policies. Other Indicators mjy lie quite complex and entail much theoretical )itstiflcatlnn.o-was Interested In determining which of two models of political change best describes the Soviet"oligarchic petrification" model or the "institutional pluralism" model. Hrumber of Indicators which he then measured In statements of obkom first secretaries.

CfltegoWetf Categories consist of one or more words, phrases, themes.elationships that represent vurtahles being studied by the nnulyst. They specify ami describe thoser Indicators nf concepts, that art constdpred analytically relevant. In order to analyze Soviet perception* of US disarmament goals

ONLY

und motivations, for exa-nplc. the prcvrinidynr. study Identifiedp-mibl,-dlsurniumenleneral and compi-lcSALT, hall of nuclear testing, etc. These goal* were then hroken down according to whether four possible motlvnthns do or do not apply for each goal. The poulldc motlvulIons_wrre achieve or maintain unilateral military-advantage, reducebudget or free resources for peaceful uses, promote peace ond security, or strengthen detente or peaceful coexistence. Each of these possible categories wus represented81 cell matrix, and Soviet statements about US disarmament policy were thrn "coded" on thb matrix."

Categories may be derived Inductively fromof thehe data ire. lo the extent possible, permitted to "speak for themselves"how theyategorizedrocedure permits the analyst to exerclsr maximum care that the categories reflect the actual nature and meaning of the communication liclng analyzed, bul It limits the researcher's control over definition of the research problem. AItcrnalively, categories may he derived deductively from hypotheses of particular Interestthe researcher, in which case tbe data are sorted intocategories a* necessary to lest those hypotheses. Thl* seemingly ensure* thai resull* are relevant lo thr problem us defined by the analyst, but It risks Imposing on the data Western concepts which may not be wholly applicable when dealing, for example, with Soviei documents. Some combination of ihese twosually pcefrrable. DrvWng analytically significant categories Is the most crucial aspect nf content analysis. It Is the ability to Identify and define such categories that determines whether content analysis can nr ennnnteaningful contribution tn research on any given question.

Kuimplr* nf Midi rrutrtoo atronIJ.IA

Coding: Coding Is thr process nf recording those aspects of text which fall into the specified categories Tlie coder follows detailed instnictlon* which define the boundaries of each analytical category. Ideally, these Instructions and the categories are sufficiently clear that any two coders applying the same Instructions to the sameillt Identical results, although this Ideal Is seldom achieved In practice. The degree of cooespondence lictwrrn two coders Is known as tlie measure of inlercoder agreement.

USE ONLY

FORNLY

tVnf Contents An nnulyst may work with Hllwi thet (lie latent contentoiumunU-uHon. talent content refer* to undulylng content which normallywareness and con-iciuus control by .the originator, ll Is conveyed unintentionally through pattern* nf word usage. The srnlyst Interested In latent content doesn't work with what I* sold, but with how It Is wild. From this he draw* Inferences about psychological attributes or underlying values unrelated to the manifest content. Manifest content, on the other hand, refers to the originator's Intended meaning. Manifest content Is not necessarily obvious; much may have to be Inferred to establish the "real" meunlng. either because the communicate? concealed his private beliefs or because he expressed themanner evident only to those who comprehend the nuances of the political context Inlalemcnt was made. The above-mrr'lnncd categories for coding disarmament goals and motivations arc designed to deal with manifest content.

Injerencei: Coding and then tabulating specified textual characteristics are undertaken only to provide evidence in support of Inferences. There are. therefore, two steps In any content analysis, The first Is theescription of content, the often mechanical process of tabulating Item* which fall Into specified categories. The second and final step Is drawing Inferences from this data. While the processes of description and Inference are conceptually separate, they are often Inextricably Interwoven at the Initial stage of formulating the research design, for the data base, categories. Indicators, and coding procedures are allor the purpose of drawing useful nnd valid Inferences.

An Example: The recent OPR. study.for Brezhnev:njrlcal Test of

(rules theseample of tclevnttf"tpccche* was selected. This sample consisted ofoviet regional lenders, with three speeches for each lender. The Indicators of support for Brezhnev were nine different characteristics of the lendvrs" references lo Brezhnev In iheseotal frequency of references to Brezhnev, amount of personal praise of the Soviet leader, reference to him by first name and patronymic, etc. Categories embracing these Indira-tors were defined as clearly us possible. The speeches were (ben coded byesearch assistant place each reference to Brezhnev inlo the approprlittr category or eatcghe frequency (or. In the case of personal praise, the Intensity) of references in rnch category was measured, anil Inferenceseuder's relallve support for Brezhnev were drawn hy computing these measures for- thejlf leaders. The purpose of thTs study was to, lest the validity nf these Indicators which ore', or might be. used by analysts of Soviet lenders' speeches. Inferences concerning the validity of the indicators were drown by statistical correlation of these scores on the Indicatorsanking of these sameeaders, according to the sttengtb of their support for Brezhnev, preparedanel of three CIA So-let analysts.

II. WHY QUANTIFY CONTENT?

Quantitative content analysispecialized technique applicable tn particular sorts ofot an all-purpose methodology. It seems self-evident, fc: example, that Inferences ftom quantitativeof frequency or Intensity are simply not applicable to the great majority of questions face! by the Intelligence analyst. This form of content analysis Is more applicable to Ihe needs of Ihe social scientist who seeks to document regular patterns of events und behavior, than to the needs of the Intelligence analyst whose daily concern Is (oro be) with the unique event. The Intelligence analyst generally draws Inference* by analyzing the logicingle situation. He may. for example, properlyingle journal article dissenting from established policy a* the tip of an iceberg nf behind-the-scenes agitation for change, and the Identification of this article as having unusuul significance Is notn ony quantitative measurement. Itful to note this nnd other fundamental limitation* of quantitative content'anal-ysis before considering what motivations might Impel one to conduct such an analysis despite these limitation*.

Systematic content analysisethod fotthe contentatge body of text In more precise terms than Is possible hy conventional reading nnd notclnklng. It Is only descriptive; it does not help to Illuminate the contest of the communication or to explain the intent except Insofar nshenomenon is obviously the first step toward accurate explanation. As previously noted, the description of corr.nu in lent Ions characteristics Interms Is iivfnl primarily fe' innking precise comparisons. It I* much less likely to Iw useful In

rUlntlnitll un rvnlvlng sltunlion Id determine how people und evrnt* convergeiven time and pluie lo pinducT some unique outeom".

Anotherhattematlc ending ofs often tedious and tlmc-consumlng work. In other words, the mechanic* of quantitative content analysis can be both boring and ccstly. and thi* ha* di^iour-aged its Ute. When reviewing fa large body of data to make cnmparntlveIA Soviet analysts sometimes Itnd It necessary tn engnge In ad hoc tabulation nf frequency or Intensity to chrek the valldlt' ol their own subjective lmprri*lon* alnut (lie Jatu. but they *top the systematic tabutatlon a* soon as the Impressions arc confirmed, and (hey ncrmally omit reference to these procedures In the final rrpnrt.

Clearly, time-consuming coding should he avoided tn the extent poulblc. In one of thr basic work* nn content) nntes that

"eontent iinnlysl* should lie done not us* possible hut rntlirf a* fmprrcUrly usU. a* nnighly a* ihe circumstance* of the study will allow (in order to mlnlmlre costs relative tohi* Implies lhat under nnrmnl condition* ciurful counting should not lie doneIuite nrcc**ury."

TJir burden of coding can be reduced bytuskeieorch assistant nr anImt thenf costs tn hrnrllt*unrlemmt In evaluating whetherquantitative content analysh tn anyOne cannot grneidlire about costs, a* Ihitor coding drpends upon the nature ofscheme a* well as thr volume nf tlie data.nf Indicntnrs nf support forthe nature of the coding scheme wat so

simple thaflt wu* unnecessary to actually read all theesearch assistant scrr-ned the speeches line hy line looking only for words whicheference In nrcrhnev nr to Ihe Central Committee andarginal note each time these word* appeared. The author then read only thr porugrnph* signalled hy the rrtrarch assistant ond tabulated the various types of references to Rrerhnev. This went very rapklly once thr procedure* were developed and learnednr more common, however, ft* mdlnf Inlmvequiring enrrful reading and weighing of Judgmentnls. Coding of themes.

In |iiirtlcu|nr. gust iimcli slower ttuui iodine of

hullvkliiiil wank

A further limitationcontent analysis I* the uppiirent difficulty Iu applying It well. Muny nl the content analyse* revlrwrd fnr this report contain tlgnlflcatil mrlhodologlcal wraknrws. and tbe nvrr-all quality nf these studies Is unlnipretslve. Several of the problem* arr dUcuurd at-other points In thi* report: implicit assumption* that public statrmrnt* meu'urr prlvutr altitudes, and crnss-nntlnnalthat are tnvaltdutrd by nnn-coinpo nihility of categories. Other wrokneue* arr found In thr sclectliKi ofonceptualization of variables, and the definition of categories and Indicators. Theanalyst who believes contentay offer answers tn hi* research problem ond who lock* trulnlng In quantitative melhndnlogy would be wrll advised tn seek methodologicalccially

tn weigh the various options at the Inception of hh project."

What ccn br Suld almut thr benefit* nf quantitative contrnt analysis and thr circumstanceshlcli its costs In research time might br nrcrssary and ju*tlfird? Thr rrmalndrr of (hi* section Is devnted to thut question.

Evidence to Support Controversial Conclusion i

The Impetus to engage in the more time-consuming quantitative procedure* may orlglnpte withlhat the normal procedure of supportingwith selected quotation*arge Imdy of lesi Ii fnr some purposes too subjevllvr; whenrr controversial or run counter tn conventionaloncerted effort to devrlop morr objective rvldeper may br Justiflrd. '

Srlected quotations dtrd In tuppnrt of prrcrlved trend*attitudes ore less persunslve, as cvldener,ystematic clasd'lcatlon and lahulatinn of all relevant references. Selrctedons serve well tooint, but they arr Ml nrce**nrlly accepted as conclusive evidence In support nf that pnlnt. The skeptic may cite othrr passages which seem tn Indicate an opposite trend or emphasis, ond. Indeed, the rvldrnce on many ts*ue* tends to be controdlclnry. The content analyst appmachrs thi* problem hy

(IPnn) ami dimesireiplulnlnR Im* I" coodorl owlrnt

FORNLY

lrvunl expnnslnns Into some cipllrid* defined cluwlficatocy scheme. Thi*oreiil systematic description of what ho* Item sold, which In turnounder basis for Interpreting the meaning nf th" data.

The previously described study of Imllcators of support for Brezhnevimple llluttratjnn. Prior tn conducting thatanel of Soviet analysts wo* asked to comment on the pntrntlalhe nanclltt* ludged that two nf the forms of reference loof first name and patronymic, and use nfame without theprobably determined by personal factors largely unrelated to the speaker's political support fnr Brezhnev. The empirical evidence from the study, based on coding all such references to Brezhnev and comparing this data with the panelists' own Intuitive ranking of tlte leaders accmd'ng to the sttngth nf their support fnr Brezhnev, showed that these iwo forms of reference were actually the best indicator* of the Soviet party officials' support for Brezhnev It seems unlikely lhat the panelists could have been persuaded to accept this counterintuitive conclusion by another nnalyst't report ba*rd on mnre Imprcsrinnl'llc review of the speeches wllh conclusion* supported by srvernl rrlcvunthr explicit description of thed In grnrratlng the data for thi* content analysl* study, and the fact that all rrfrrrnce* to Brezhnev wrrr tabulated and that tht* duta wn* then correlated with oilier measures of support for Brezhnev, were probably essential tnersuasive case in supportonclusion which ran contrary tn Intuitive expectations.

Summarlring and Aggregating Data

Summarizing or aggregating specified clrmrnt*hkIv nf tlutii always Involve* Implicit quantification. The nmilyst studies tlte datu and draws conclusion* expressed hi terms such a* more or less. mcrca*lng or decreasing There are times wher. the volume of material lieing analyzed I* so great, and/nr the distinction* being made Ml complex orfine, that some sysleniHtlc procedure must lie employed to nllevlute the load nn the analyst's discriminatory cnpnhllltle* It i* under such circumstance* tlml more explicit quantitative procedure* may Iw rrqulrrd

A requirement for incrra*ed precision may aha prompt on analyst to turn tn quantitative methods. Meauirrmcnt* nf frequency nr intensity are wny* of transforming qualitative datu Into quantltutlvr datu. to tbut concept* such a* greater or lesver. Increasing nr decreuslng, pmvlbly nr probably con lie expressed with some degree of numerical precision. Quuntlf lent Ion I* nnl an end In Itself, hut It ran be useful undrr certainnrend over time, for example, conventional analysis may be able to determine whether the observed variahlc Is Increasing nr decreasing,eldom has the precidon hi menuire the rate of Increase or decrease, or whether that rate Is accelerating or decelerating. Content analysis may provide thi* additional Intelligence, fnr Iteasurement toolensitive to small ctungrs.

Rut coutlnn I* always Indlcalcd when drawing Infrrences from quantitative doto. Itertinent, for example, to consider the fundamental question nf what Inferences may be drawn legitimately (mm measures of frequency. Frequency Is. In mosta valid measure of preoccupatlnnopic. I* IIalid measure of Importance.concern, or Intensity of feeling? It the frequency with which Brezhnev refersroblemolicyalid measure of hnw Important he regards that problem or policy, or how strongly he feeb about It? Content analysts often do make this type of Inference.

Sometimes Information Is so well understood that it nerd not he mentioned. Thr atomicarcly mentioned In Japanese cabinet delwtrs trading lo surrender In World War II. hut Infrrqttrncy of mention In this Instance can scarcely U- equated wllh lackImportance.alid measure of Importance under many drcumslancev however. It

"heilher those who conduct thetc iludlei nor ihote icho reed ihem should he seduced hy the existence of numbers, mathematical man'piifa (torn, and tests of statistical sientft-cance Into believing that ihe results are automatically 'harder and more significant than dune produced Inj lets quantitative methods. Often the apparent precision gained hy quantification Is more lhan balanced hy the hues Incurred In the opemttons vhlch are needed lo transform ihe dataorm which can Ite)

rUlTbNIY

ONiY

bcrfl noted, for rxan.plc, that statements about gioup 1Iundientail implicit statementsthe frequency with which these belief* and attitude*held by individual member* of that go nipinference* from frequency areappropriate and likely to be valid when aggregating individual expressions to form Judgments aboutroups. Even In such -cases, however, the validity of Inference from frequency Is not self-evident. In every Instance tbe researcher must Judge the validity of the Inference, based upon hb knowledge of the context and any otber Information which might explain frequency of observedas Iwing due to something other thun their importance. There can be no question of substituting counting for Judgment: counting Is only an aid lo judgment, and in every case tbc researcher must evaluate what the numbers mean.

Statistical Analysis

Quantification may be required In order to makeide variety of formal statistical procedures for calculating probah'litle* or the significanceiven pattern of results, and for measuring correlation between variables or the degree of Influence ol one variable us compared with others. This would be highly relevant, for example, in determining the impact of generational differences within the Soviet leadership. Age Isuantitative variable, so If attitudesample of leaders were quantified through content analysis It would be possible to use statistical procedures to analyze the relationship between age and attitudes and to yeofect changes In dominant elite attitudes over time.

III. SOVIET SOURCE MATERIALS

In one sense, all available written end sptirn common feat Ions represent potential jouhc materials for content analysis. Tlie quantity of such material Is potentially unlimited, for written andervasive aspect of all political activity.owever, the selection of appropriate and useful source material for content analysis Is con-stiiilurd by the nature nf thr Soviet political system, the requirements of the methodology, and the purpose* for which the methodology I* best used

Soviet Communications

It ha* lieen suggested that the nature of political communication In the Soviet Union preclude*content analysis. Three paints are sulKumesi under thbfi' I. that cenvonhipullhat public statements areruitful area to search for dtlfvring opinions: second, that when political dclutr is :onductcd through indirect attack, obscure historical allusions, and hy the omission of any reference al all,oo complex to Ik* pigeonholed In. the necessarily simplified categories of ihe-cont jit analyst: and third, that Soviet media content and official statements cannni be taken a: face value in any event, so inferences drawn from content analysis are Inevitably ofvalidity.

That Soviet media differ greatly from Western media Is beyond dispute, but it dors not follow thai effective content analysis is precluded. On the contrary, there are marked udvantoget to working wilh Soviet as compared with Western media. The Soviet press Is linked directly to policy-makers; Its content reflects the views of policy-ma ken. whereas Western media normally reflect the opinion of those nongovernmental elites who seek lo influence the policy process. And discussion of political topics in the Sovieto highly structured that consistent patterns of relatively small differences are likely to he more signifkont in Soviet communications than patterns of considerably larger differences in Western communications. Because the changes or differences one Is seeking In the pattern of Soviet communication* may be so subtle orystematic method of observation, such as content analysis, becomes all the more necessary to identify them succes*fully.

There arc three forms of political debate In the Soviet Union, First, some polky question* are officially sanctioned for more or less open discussion, especially in the specialited journals. Problems of economic priorities and how best lo manage the economy frequently fall Into this category, and some aspects of defense policy hove, occasionally, been the subject of publicecond form of deflate is carried out through discussion of how policies are Implemented. Evenolicyecreed a* accepted by unanimous consent, so that direct challenge to the pnlicy Is not permissible. ImpVrnirn-ration of that policy Is fair game for discussion. Most political commentary in the Soviet Union is aimed at mobilizing support for. improving efficiency inout. or criticizing shortenming* in the in ptemen-tation of approved polities Discussion takes place

Fgfl nrtiis.ro-Tiiiit

FOR OFFICI

a content of ostensibly full support for established policy, but divergent emphases reIn fact exprcsi divergent policyA third form In which dissenting views arche "esoteric communication" studied bynew twistraditionalformulation, omission of referenceeader, etc Content analysisotentially fruitful tool for examining all three forms of political debate.

The extent lo which one can take at face valueoviet (or any other) Official says or writes b. of course, another question. But thr problem ut the content analyst Is little different from that of the conventional analyst. For bettei or worse, both are obliged to depend upon published sources, and both must take great care to justify Inferences drawn from such material.

Specific Sourcat

A cardinal rule of content analysis which is not easy to observe in practice Is that like must be compared with like. Statements by Individuals may be compared reliably only If the situational stimuli which prompted those statements are roughly eomporablc with respect to such variables as time andne can control the influence of many contextual variables in either of two ways: first, compare only speeches given in an Identical context, such as speeches givenPSU Party Congress; or se.-ond.epresentative sample with enough speeches for each individual that differrnccs due to different contexts cancel each other out.

There are several rvrnts In the political life of the Soviet Union which regularly prompt many leaders to make pubjlc statements at approximately the same time under more or less Identical circumstances. These Include the central und rrglonul Party Congresses, Central Committer Plenums, elections to the Supreme

Soviet, the anniversary of the October devolution, and Red Army Day. Comparability, and hence validity of conclusion. Is best served If the statement selected to represent the views of Individuals or groups of iudivlduals to be compared with each other an? dnwn. to the maximum extent possible, from statements mode In response ttriuch comparable stimuli..

The content of Journals Is another common source of text for content analysis.ournal ts formally affiliatedecognized grot. It may be regarded as the vehicle for expression ofiews of thil group. Thus Pravda officially expresses 'he views of the Party, while Itwlia reflects governmental policy. The analyst may wish tu generalize from views expressedournal to views of the parent group. But analyzing specialized Journals on the assumption that they represent the views o* specialized sub-elites poses certain methodological problems.pecialut Journals as Indicators of the attitudes ofoproty ekonomlkt and Ekonomtchcikaya gatcla to represent the economic elite, but this has been criticized (lloreflck, Johnson andJ.n grounds that whether or not specialist Journals are indeed representative of identifiableups" isesearch issue rather lhan an appropriate assumptiontudy of Interest groups. As noted by.. the problem concerns the proper conceptualization of an elite. Is an elite best rWined in terms of Its performance of specified functions ih lis affiliation with an Institution? For ihr; military elite, the functional and Institutional definitions encompass the same body nf Individuals. Butn economic elite? Lodge conceptualized It functionally, then measured It via the output of two institutions Omirnuls) which do not necessarily represent the broad spectrum of those who perform economic functions.

t> l< preferable under some circumstances to definen terms of Identified individuals who play key functionalrather tion In termspecializedample of statements by these Individuals would serve to measure Intro-group cohesion as well as differrnccs with othrr group*.

IV. APPLICATION TO RESEARCH ON THE USSR

Taking Into account the particular costs-and benefit* of quantitative content analysis, plus the type

Al USE ONLY

data known lo be available. It Is pnsriblr to Identify1b|ect area* or lypci of studies for which tuch analysis scrrm to offer the greatest potential fnraluable contribution. These ore described below. It Is certainly not Intended to limit potential applications tn (bete fields, but only to sugjrrtl that these are areat In which utefulmljtht be found.

Perception Ar.alyi'i

The Vietnam experience made Americanmore icnsltlve to tbe Importanceectslon-maklng.on the basis of what they believe to be true, quite Independent of what may appear to others to be the objective factsituation. To understand and predict the actionsoreign leader (orne must know how that leader defines the situation In which he finds himself. One must understand that leader's perceptions of thr International system, aad af the position of his own state and Hi potential adversaries In that system. Penetrating the minds of foreign leaden is as difficult as It Is Important. In the absence of Interviews, questlonnalrrs and opinion polls, which are out of thr question In Ihe Soviet Union, content analysis has been suggrstrdeam for measuring perceptions.

A particular advantagr of quantitative content analysis In perceptionhat It facilitates going beyond such generalizations as "the Soviets believe* There iv of course, no such thingSoviet" perception. Aggregation a' ihe national level Is often desirable, but It does obscure the fact that different Soviet leaden perceive different things at different times, and thatven ambivalence and Incontittency In thr peereptlons of any single leader. Content anelysb provide* procedures for aggregating dataay which retalnt this dlvrnlty. so that nne may analyze who perceive* what, when, with what frequency and how consistently.

Of coo nr. neither thr content analyst nor the analyst using conventional techniques It drallng with actual perceptlnns.ealing with publicand media content which moy reflect rlthrr true perceptlnns and belief* or what thr tpraker or writer wanlt the target audience to bcllrvr. Contentrscrlptlvr technique. It dors not explain why something hat been said ot written.

although tbe mitre precise drscripllont ha* hern tuld buseful to 'he analyst who must rxplaln the full meaning and significance of the data.

Quantitative content analysis of perccptloni must proceedwo-stage process. Thr flntimited to describing perceptions atIn written or spoken communlcotlons, Jhe second stage draws Inferences concerning" to vefiflt degree. In what respect, and under what circumstances thrcontent reflects actual perceptlnns rather than communication! ttratcglcs. These Inferences arejudgmmti madethe area specialht. In other woedt. quantitative content analysis provides the raw material fnr subsequent qualitative evaluation of perceptions. Such qualitativeequired for each Individual conclusion drawn from thedata.#

If thb limitation Is observed, quantitative content analysis can. fot example, help lo define how the Soviet Union perceives and evaluates the US both as an adversary in political competition andorlnrr In arms control negotiations and economic relations, and It can help to measure changes In these perceptions over time.

Elite Analysis

* ll tn Oiirl villignmlIt iwililr

vahrHhIn dravlni Inirrram annulirmn ii-iWie statementi. and Ihen tn pwcredtwin given,nting mild helK*i) aod CACI.all pre* tn this iem(Maib*saajset*hetf Martm Itneutii thai Ihe redntrepresent generallynt true allliudn and perceptions

CJLCJor eiample. claim. In Hi tllle tnudv n( prtrrptMo. UnIn faetrtpftnnhJk ilulemrnls NHlher anllMr deals nnllrttlv mil. llie secmdnnnlxlsiiilanr.

Interest In political rlltrs has brrn stimulated hy recognition that nationalften thr product of bargaining and compromise among diverse Interest groups and bureaucratic organizations with competing and contradictory goals, rather than thr resultatlnnal processingle decision-maker. In order to undrntand the dynamics ofone must know the strengths and tbe positions of the elite groups which Influence the decision-making process. Contentore precise than Iradltlnnal onalysbethod to aggregate data pertaining tn beliefs and attitudes of

fOR OrFIClAl-tlSfoNlY

groups, so tlml one imiy incusine (III* clnu In which different group* agreeariety

Itf If V

Quuntiflcutltin of iittitudlnul (lato also makes Itn cnerclutt utlltttdlnul data with tin*readily obtainable dutn un agr. education, sociiil nnd rthnlc origin, und career background. One can. test nwnythe rrLtlnnslilp Imtwrcn political attitudes nnd agr. How significant an? the difference* Itrtvveen political generations In the USSR? On what point* ore the difference* greatest? Are these difference* Increasing? Do ynungrr leaders rank their political valor* differently than their rlden, and If so what are the long-ranee political Implications oi thb fact?

Kry topic* nn which ogr. career or nationality groups might lie expected to hold somewhat differing nttltudrs Include detente. iHidgetury prlorltlri, mean* for Improving productivity and efficiency of the economy, policy toward Intellectual and national dissent, and the role of the putty. Explanation or predlctlnn of Soviet decision-making In ihese field* requires knowledge of the diverse policy Inputs by such interesi group* a* the central putty organ! rat Inns, regional party opftaralchAi. economic managers, ihe scientific establishment, and the military, and knowl-rdgr of trends of altitude change within these group* would heful fnr long-range forecasting.'

The problem* Inherent In drawingrom media cm lent are less serious for elite analyst* than for perception analysis. Since Interest focuses nnifferences between elites,nr some purpmcv und In some degree, sufficient lo demon-strute that different elites have different cornmunlco-llon*t may not be nrcrssary tn show that media content reflect* "true" attitude*.

Political Indicators

ighly developed set of rconomlc Indlcutiws to measure thr statr of the Soviet economy. Init no cttmparabte set of Indicators tn measure the state of the Soviet politicalet mooltoring fuiulmnt'iiiul political trends Is as Important us nioultnrlitg economic trends. Analysis of fundamental chuugf In the Soviet political system promts niujor conceptual and mcthtxlologlcal challenges.ecessary to determine which aspects of the political system nre to be recognized as Indicators of the fundamental stute of the system and then In devise methods ttrmensure change In these Indicators.

Contentlhmtslng method forpolitical Indicators. Some measure of the extent to which Identifiable Interest groups articulatepolicy position* would surelyan appropriate part of any program for measurement of political Indicators. Other Indicators might deal with tlie Impact of Ideology, attitudes toward dissent, attention paid to consumer Interests, or the amount of expressed or perceived hostilityis the foreignThere Is no dearth of candidates for the mlc of key Indicators of the state of the political system, but discussion of specific indicators and how thry might be mrnsurrd Is hrynnd the scope of thb study. It suffices hrrr to point out the need for such Indicators und tu observe that content analyst*orkable tool for measuring some of them.

The strength of conventionalecognition of nnumiilles In the flow of officio! statements and publications, anomalies that indicate new policy formulation* or small but significant differences In the way the same policy Is formulated by different Irudrrs. The strrngth of contrnt analysis is In measuring changes In the background against which dolly events are ultserved. It run measure slow changi-over tlmr In factors which nre sornrtlmr* regarded n* constants rather than variables In the Soviei system. When chungc Is very gradual and thr vnlumr of relevant, primary source material so enormous, trrntls may easily he nverlooked tw misinterpreted. Gradual evolution in the political system i* best Identified nud measured by applying explicit criteria to aselected sample of the data.

Anolyili of "Eiottrle Communlcatloni"

The Soviet systemremium on public iinuiilmlty. but this Is oturfuce harmony. Tlte ways In which divergent opinions ure rxprrssed nre radically different tluin In the West, but thry nrr expressed.

Foj^MtcrxrusE only*

(kUftHan stutcmrnt* on contentious siihjccl* cuniml br made explicitly, they lake the (mm of esotericwhoie deepest meanings can he grasped hyart nf Iheir audience"B,

p. m

Krrmllnologlstsariety of technique*-to analyze esoteric communication! and. when the data and rules for making Inferences from the data are specified, these techniques fall within the general definition of systematic content analysis. In most cases, however, these techniques are not quantitative and. therefore, not within the purview of this study.

Quantitative approaches to Kremllnology arehowever, ai dernonstrated by the recentsupport for

Brezhnev,tiaiyzTng vethal formulations which tend to be used repeatedly, such ai references to Brezhnev or to key policies, one can quantify by tabulating ibe frequency and form ofeferencesumber of speeches or articles. These data may then tie usedenchmurk (or comparing leaders with each other or for mnnttnrlng changes In attitudes over time. Another approach to Kremllnologlcnl onnlysls. Involving ending nfot snld, Is din uoril In the Appendix.

Psyc ^logical Aisaiimant

Content analysis offers promiseool for Indirect psychological assessment of political leaders, fur the latent content of messages provides dues topsychological charactrrlsllcs of the originators nf the messages. The apparent fact that psychological cho wet eristics Influence patterns of languagehe foundation nf the study nf psychnllngulstlcs. The spinoff to political psychology has been limited, a* political communications obviously contain less latent content than the many lest structured fnrms nf communication nf Interest In the psychnloglst or psychoanalyst. But avullable research Indicates that political communication* do eontuln lulent contentlutprcM, The best source of suchwould be Interviews or debute* In whichIs tpcukliig cxtcmporancomly. butpresumed to be sub*tuntlully writtenedited hy the leadi himself are

.As reviewed Ir).the Appendix under the rubric Analysis of Wntxt Usage, currently availablefnr exploiting Intent content In political speeches nr writing* Include method* (or measuring the following characteristics: need for power, need (or achievement, need for affiliation, dngmnllsm.complexity, belief In ability to Influence one's own destiny, nationalism, use nf ideologicalamount of ttres* helng (citeader, and the comparative ranking nf variousariety of hypntheses have been advanced linking theseto predlspnsltlons tnward various types of foreign policy behavior. But tbe field nf Indirect psychnlnglcal assessment nf political leaden Is nnt well developed, and the empirical study of linkages between psychnlnglcal characteristics and fnrelgn policy brbavlnr has Just begun. Additionalequired to advance the state of the art and to validate the technique* and proposition* already developed."

The existing state of the art ofe**ment I* adequate for experimental application and merits further testingroject to measure relevant psychntoglcal characteristics of Soviet leaden and rank them onsychological dimension*.leaden' speeches to tbe CPSU Party Congress or their Supreme Soviet election speeches would satisfy the requirement that commiinicalions beinghe taken from Ihe same sltnnllonal context. Analysl* of the latent content In these speeches would lap dimension* of available data nnl presently reached by conventional analyst*.

' Winter androvide an eslenilwImm an acudemlrnf view, nf matenlrtiinbpte fctrl-iui pxOllcat leaden

SFVK ly '

IAI USE ONIT

jy

approaches to quantitative content analysis

most inoK. content analysis romn In various type* and Thl* Appendix Is hosedeview ol* Ihc academic literature to Identify various content analytic techniques and how they have been applied to research on the USSR, ondesser extent China. It conveys an Impression of the variety of research problems that cun be uddrcsscd through quantitative content analysis and describes some of the indicators developed for Indirect measurement of variables which do not lend themselves to direct observation. An overall objective- Is to stimulate thinking by Intelligence analysts concerning ways In which such techniques might contribute to their research needs. Analysts considering employing quantitative content analysis will findseful guide to the various paths open to them.

There are several ways In which approaches to quantitative content analyst- night be classified for discussion andasis for classification Is the unit ofhe key Items the researcher looks for might be Individual words or symbols, themes, the length and position of newspaper articles, or other characteristic* such as types of logic or patterns of co-occurrence of two or more words orecond bosb forhe kind nf measurement that Is made; one may note an Item's presence or iiluence. count Its frequency, measure Its Intensity, or score the amount of Attention devoted to It.

ombines these two classification scheme* In matrix form The lohlr shows which units of analysis pnd methods of measurement are lodged most applicable to lntrlligrnce research and ore given most attention here. Types of content analysisre considered most relevant and ore therefore discussed In some detail. Thosere mentioned briefly so the reader will be aware of the existence of such techniques, while typesre not mentioned at all. because these combinations are either not feasible or only marginally applicable.

logic would seem to dictate organizing this section either according to units of analysts or kinds of measurement. In practice, however, brevity Is liest servrc!ixture of the two. The flnt approach discussed hereind ofpresence or absenceheme or group of words of analytical Interest Discussion then shafts tn the four units of analysis and Ihe various quantitative measure* most appropriate to each. Regardless o( the unit of analysis or kind of measurement, coding may be handled manually or by computer, andanalysis is the sixth and final approach to content analysis discussed below.

or Abwrvce

Inferences may be drown from the singleor nonappearance of some characteristic as well a* from Its frequency. This Is. Inimple form III quantitative ut^lysls In which the numbers used are limitednd I.1 coined thr trrm nnnfreqtiency analysis for ihis variant of the grnrr.

If onr wlshrd to apply content analysis In study agreement or disagreement between Soviet lr.iden on Ihr policy of detente, one approacho revlrw all dhcusslnn of detenteample of speeches and to list ond categorize nil arguments

Air-OStToNtY

results of such analysis may be presented in graphic form,atrix with all possible arguments along Ihr horizontal axis nnd names of. say. Politburo memliers nn the vertical axis.belwcen Politburo members would be apparentlance.

It Ishort step to mnvc from coding presence or nhsence lo other forms nf measurement. For example, nne mighlimple hlgh-medlum-lnw scale tn take Intn account not Just the presence or absence nf an argument, hut also tbe amosint of emphasis given to it. Or one mightumlvei of speeches by each lender and aggregate the results tn arriverequency score based on the niimlter nf speeches Inpecifiedappear*.

Anolyili of Themai

A theme Isubject-mutter"1 Thematic analysis may lie used to describe the appearance or nonappeuranee, the frequency, ond/nr the Intensity nf expressed liellefs. attitudes, perceptions,alues or goal* of either Individual* or gmopx. Itieclally useful for processing Urge bodies nf data In ttrslrf lo descrllie In quantitative term* the orientation of elite

Figure I

which reflect the perceived consequences nf detente fnr the Soviet Union. Given the structured nature of political discourse In the USSR, thereimitedf *uch arguments and they tend to repeat many of the same formulations. Soviet statements reflecting the perceived cc^rscqucnces of detente can prnbahly be encompassed within about nine oreduce risk of nuclear war. speed economic growih by acquiring Western technology. Improve csmdltlons for struggle of revolutionary forces abroad, etc. Concern about adverse consequences might lie reflected hy reference to the need for strengthened Internal political controls, rejecting Ideological coexistence, etc.

A sumplc nf speeches representing the views nf selected leadersiven point In time could then he analyzed tn determine the presence or absence nf eoch argument. One purposeomprehensive set nf categories including all passible argument! Is to permit Identification ofeader chooses noi tn say In his speech, as this may be more revealing than whal he does say. Although certain klndi nfulmui detente ate virtually mandatnry. one might expect dlffcrnt patterns In what different leuden emphasize or Ignore.

groups pf nation*hole. Till* Iv became thr

pXWrUUff n( aggregating themes liy

Individual* to arrive al group view* I* by lit vrryatter of rr iting frrquenclrv

Thr focus onore akin to traditional rr-seurch than are .the othrr form* of quantitative contrnt anatysls. in the area specialist may feel morr at home In this medium. On the nther hand, ending li more difficult,heme I* Irss susceptible to unambiguous definition and rapid identification than ore categories composed of Individual wordsrases,

In setting up the categorieshematic analysis, the researcher. In effect,uestionnaire. He thenommunication to seek answen tn the questions he has posed, so that the originator of the communication becomes an unwitting respondent to the questionnaire. Various techniques may be Identl-fled according tn the type of question being aslrd. These art designated here as single dimensional quesllnn, mullidlmenslonal question, scaling question and open-ended question.

Single Dimemtmalingle multiple-clinlce question Is posed, For example: How can Soviet agricultural production he Increased? The researcher Identifies possible answen to the question nnd places these Into categories that are mutually cxclu-ivc nnd which exhaust nr virtually exhaust all possible anssven to the question. Soviet speeches and/or writings ore analyzed tn determine which answen thry glvr to the question. Two or more individuals or groups ore compared on the hosts nf the frequency nr Intensity of their answen lo the questionlvrn time period, or answen for thr snmr Individual or group an? compared for different time periods.

. for example, comparrd Sovirl and American fnrrign policy altitudes. One of the many quesllnn* he asked was: What Is the nature of the present International struggle? He set up the following three categories to encompass posslhlr answer* to this question: between two belief systems, between Iwn social systems, nr between two centers nf power. Snvirt elite attltndrs nn thb ond othrr question* wrre compared with American elite attitudes during the same time period,ata base drawnnmplr of uitlclex In Soviet and Americanhe frequency of responses falling Into each cotrgory

was tnbulatrd. Since dlffrrrul quunllllrs ufweir extimluei* for each country,counts were rendered comparablethe frequency (or each category nsof the total answers to that questioncountry. Schematically, this takes the

I. What fs-tlie oihire nf the present Internutlunul struggle?

Between two belief systems

Between two sociot systems

Betsvecn two ccnten of power

2

Multldtmentlonaluestion Iswhen ft includes subordinate questionsbetween more than one aspect ofThis technique was usedto analyze Soviet perception* anddevelopedatrices, roch of whichspecific political, economic or military topicon Soviet self-percept looovietof the US. or Soviet perception of how theIhe USSR. Three of theeproduced asond the variety ofbe

selected.

The reader should refer at thb point lo Figurehichwo-dimenslnnal matrix fnr thr analysis of Sovirl military goals and force structureIn effect, two questions are being asked simultaneously, hence the multidimensional coding. What are Sovirl military goals? And what orr thr means rnvhngrd for achieving each goal? Answen to the former question nre coded Into the goal categories listed along the vertical axis of the matrix, while answers lo the latter question ace anted Into the mean* categories listed along the horizontal axis.

*lnHaVdf thr srkrtnermplmeJ hyrry UmlLn nrnrrslii"by AnerU (lOWlnf Snvirt anitsiil tiiiurs.

Soviei Coal, (Military) Forte Structure rVfl.ulrrrnrnti

llu SovietIn attain/maintain ihtcapohtlllUi liy Ihr mean? Indicated:

(Service)*

(Type Warfare)'

<

7.

<

az

o

G

a

Warfare

5

j

-

ll

a 3

n

1

n c

p

1

t2

ihr capability In defend ajealmt or deier an altaclIhr USSR. Eatt European Communlil cnun-(riri and Cuba

Ihr capability to defeat or drift US ImpwlalUtand lupport of counter-revoluUoo to the Third World

the capability (nr the military defeat of China

Ihr capability It- the mililary defeat nf NATO

(he capability to lupport Soviet foralfn poller (ipedllcally Mated ai >uch)

military copaWlltlei (ornwm (tn Include Inferred tiippnrt o( Soviet Wetfn polky) or for umprrlfled rrowwi

fMpect to the Unitedapital lit, or potential aumaort Inther or both type meant will be coded ai appropriate.

Flflurt 3

USE ONLY

lo

If

P ll

1 tn ^

- | Cn o

8

-cT 8-

9 utupc, jo

irMy AJRIIIIM

G

U

mi jouit>v

liy

e o

!

s

1

E 3

ill

8

pecifleA US (capltanit) capabilities relative to Soviet (Sorielbt bloc) capabilities are:

FORUSE ONLY

icial use only

f 1X1 Cnah (Military) OlMumamrnl Cab and Mnti.elhmi

-

ovkt prrcrfriton thaiS goal to:

pptk-biUlv

a

1

a

li

<

h

Si Ii

ll

i

I

i

ii

irnrial dhar moment (Indixlri compl'tr dl<arnwmrnt and irtrrrotrt In "dnarmamcnl"

$tma

achieve (nrr*(MBFRi

achieve arm* control (SALT. ASSAY, halt arm race

hill nuclear letting

i mi r- or mnlmlnd uie of chemical, hlnlndcal

prevent nuclear pralifcrallnn/ritabltih nticlrar lie* nav

Figura *

If sufficientvailableiner hirakdown. uny number nf additional question* or dimensionsdded, although II bcci*nrs somes* hat complicated In plciure thrse graphicallyinglen lhc CA.CI. study, 'heatrix wai lienken down.nlong four supplementaryEach statement coded Into any cell of-the matrix, for example each reference tothe capability for military defeat of NATO) being achieved by meunsconventional warfare) was nko coded for answer* to the following four questions: Is thr slated goal In using ennventlnoal forcesh NATO to deter, defend, retallntr. or defeat? Is Ihe goid to achlrvr superiority or parity? I* the goal toapaldtlty ne to maintain an already existing capability? Is the goat qualitative or quantitative superiority (or parity)?

A multidimensional qursllnn enables the researcher to deal with relatively fine distinctions on complex problems. Putting resultsomputer program together with data on date and source permit* cross-tabulation of data to plot trends over time or measure dlffetrncTS between source* on any combination ofhen dealing with large Imdlcs of data, so that one ha* sufficient frrquenclcs lu ull relevant cell* of the matrix, thisowerful analytical tool.

Seeling Qiiriflon; Th- third schema for thematicnvolves asking questions about where some characteristic ranksontinuum running from high to low or from more to less of some quality.) employed this schema to measure, inter alia, the extent to which specialist elites have come to participate In the political process In the Soviet Union, as compared with complete domination of the political process hy the party flppamfcrtlti, Me asked five questions concerninguch as: Whoesponsible forand. Who should be responsible forThe dc'a base from which answen were

hn*tblril dimension oddest along the

* nn Ihrhore dimensions nnd extensivel

the use nt ta-trke. lu ennlenl umtlytU.

The ilnto Use used In the CA.CI. sl.idy wus selected In. It' nwlluhlllty rulher limn Its adequacy. Thenf Ihe study was tn test nnd demnnMnilemethodology, nnd Itnf Interest from lhl> point nt vtevv rulhe. thon fm Its -ohsKotlve cnm-ltislonsSovietund perceptions.

ohtalned consisted ofoviet Journals selected as representative of five eliteystematic sampling technique was used In select ftom thesemimnk GOO paragraph* per elite for each of right sampled yeani'ach pani-gruoh which dealt with the ctIb.'vc rrspomlbility for policy-making of the parly upp-.rut and thr specialist elites was coded on the flvrj>olnt scab shown In Figurelie following lllu*trates_a':possace codedhe midpoint on tbe scale, indicating Joint party-spectalbt participation: -

"You say 'Comrade Khrushchev said thus andhe highest authority In agricultural science? You are President of tbe Ukraine Republic Academy of Science*m the Secretary of the Party Central Cnmmltlre. You must help me In these matten. andight be wrong, andm. ynu, ns an honest scientist, should say; 'Comrade Khrushchev, you do not quite undentand thef sou explain things to mesillou for it. Ixt usas wrong. But you will say. 'Comrade Khrushchev said thl*upportedhut sort of scientist Is litis comrades? Thb Is toadyism and(From Pravdo.iteddge. p. II )

Beeniise this scaleontinuum running from morr to less party participation, (Ik- coding results can Ik- uvrrnged and presentedingle quantity. This gives the schema shown In Figuren which thr numbers reprrsent the average scores fnr rach elite for each year.

Lodge compared thr answen to the question who fi responsible for policy-making with those fotoijM lie responsible In order to answer qurstkms such as thr following- Arr the specialist rlltrs pressuring thr party for grratrr Influrnce In thrarena? Answer:s rrflrcled hy the Increasing discrepancy between how much the party elite Iselleve* the specialists should participate and what theem selves believe their role should he. a* well as an Increasing discrepancy betwren what Ihr specialist* bellrve their role actually I* uml what they brllrvr II should be. A* specialist elite purtlclpotlnnoe* pnrly-rlltr conflict Increase or decrease? Answer: It Increases Score* on all five questions for all elite* were combined to4 percent Increase In specialbt elite participation ineriod a* compared with

FOfl OfilClAL USE ONIY

be responsible foe policy-making t

1,0 Party participation Party participation Joint party-specialist Specialist participation Spcciallit participation solely.--

while overall party-elite conflictbetween party and specialist elite views on participation) increased threefold during the same period.*

Open Ended Question: This differs from the previously described procedures In that the questions posed by the researcher are oprn-rnded rather than multiple choice. The answers are recorded verbatim, without (at least Initially) forcing tbem into mutually exclusiveet nl open-ended questionsonceptual fromesvnrk for systematicof all relevant Items of Informationiven group uf documents. The explicit formulation of thr set of rrlalrd qurstlons permits different researchers to apply ihr same conceptual framework to different bodies of text, with the result that their work will be' cumulative and/or comparable. The answers obtained in rrspousr to these open-ended questions tire then subjected to other forms of analysis, either qualltai'vr or quantitative.

The procedures recently developed by) to codify ihe concept of an "operational code"

Illustrate this technique. The oprratlonal codeto .inalysii of foreign policy decisionmakersdeveloped bynd) and was first used In study the underlying beliefs and assumptions which influenced thebehavior of Soviet Iradrn. Thrf an operational code refrrsecision-makers" system of beliefs about the fundamental issues of politics and social life. Analysiseaders* code involves asking questlcns concerning his beliefs about the nature of politics, nature of the contemporary International system, how muchradrr has over historical developments, how one selects goals or objectives, attitudes toward Ihe problem of risk. etc. This approach has been used to analyseozen leaden, but each researcher has conceptualiwd the questions somewhat differently.

Drawing on the experience of previous operational code studies. Holstl, with assistance from Geott-e. has now defined these questions for more explicitly ond drafted comprehensive guidelines for coding on operational code.wo-step procedure In which the researcher first codes answers to open-ended qurstlons. These answers then form the data base for subsequent slnfclc-dlnirnskvnol or multi-dimensional coding. For example. In the first step thr researcher asks, what are the un'ret's beliefs about the roleeader In shaping history? The second strp then sorts each answer to this question into one of rieht possible

foronly

In research which is area-oriented rather than theory-nricnted. the second step Is optional.

Analysis of Word Usage

When words rather than themes arc taken as the unit of analysis, the researcher is most commonly Interested In studying psychological attributes-of political leaden or their transient uttiludes such as hostility or friendship. Discussion here Is limited to techniques for measuring relatively stableattributes and personal values.*

Analysis of word usage often, but not always, deals with latent content which escapes the conscious awareness and control of the communicator. Analys's of the choice of words used toiven thought seeks to identify characteristics of the author which may have little or no direct relationship to the manifest content of the th' ght Itself. It Is based on the assumption that underlying psychologicaland personal valuesignificant impact on patterns of language usage. When It deals with this latent content, analysis of word usage is relatively invulnerable to being distorted by deliberatestrategies, although tbe results may be strongly influenced by variations in situational

M.dentified patterns of word usage to serve as Indicators of six personal characteristics: need for power, need for affiliation, conceptual complexity, distrust of others, belief In one's ability to control events In which one is involved, and nationalism. These traits were measured foreads of state or prime ministers. Including Khrushchev and Kosygin. through content analysis of transcripts of their press conferences. Hermanntwo general personality types among the foreign leaders, those whose personality traits predisposed them to aggressive responses to their environment, and those whose personality traits predisposed them toward more conciliatory responses. The aggressive leader is more Inclined to try to manipulate and control othen. more suspicious of others" motives, mere Inclined to initiate action, less Jikcly toide rangeof olternaltvcs^arid mote'concerned with maintaining notion ill 'Identity and sovereignty.then ton-elated the measures of personal characteristics wiih quantitative measures of foreign policy output in otder to test the validity.of the indicators and of the hypothesis that personal characteristics do have predictable effects on foreign policy behavior.

Need for power refers to an Individual's concern for exercising control or influence. Hermann measured it by counting the percentage of verbs which Indicate forceful action, giving unsolicited advice, attempting to influence or pcrjiicde or impress, etc. Conceptual complexity refers to the degree toerson observing his environment differentiates between mony shades of gray, rather thon perceiving his world In black ond white terms. It we* measured by determining the ratio between low conceptualwordsbove all. certainly, definitely) and high conceptual complexity wordsmay, possibly,imilar indicators were used to develop measures nf the other personal characteristics.

) coded verbal Imagery used by all Americans who campaigned In6 presidential primaries. He coded Imagery Indicative of three fundamentalfor power, need for affiliation, and needata base the speeches in which these Individualstheir candidacy. Donley and) had previously coded need for power and need for achievement Imagery In the Inaugural addresses of American presidents5

The coding proceduresodification of the procedures for scoring Thematic Apperception Tests. They differ from the procedures used by Hermann to code some of the samehile Hermann focused on singleorder to permit computer-assisted coding. Winter looked ot the Imagery in an entire sentence. Examples of sentences coded by Winter are the following: "Government has become more intrusive, more coercive, more meddlesome and

FOR OFFIpanfTjSE ONLY

FOR OFFICII

lew effectiveot the work of one person, it Is the work of all of us workingatterns of motive Imagery are not related to political philosophy, but they are related to leadership style. Winter's work Indicates that, at least In an American context, such analysis can predict how actively andeader will govern,ertain extent, how effective he will be. whether be will pick advisors who arc like himself or who are technical experts in their fields, and how much he will rely on his advisors. Winter's procedures have not yet been applied to Soviet or other foreign leaders.

ighly experimental vein.) has identifiedategories ofof use of simile andpsychoanalytical theory relates to specified personality types. Examples of aggressive/striking Imagery taken from Stalin's writings and cited by Luck include the following:

"One of our tasks Is to break down this blank"

"Wc shall continue to throw out such

Lenin drove themave already severed whole sections

of the British working class from the

reactionary leaders."

Luck found that Stalin's imagery was primarily of this type, while Hitler's writings were characterized by imagery relating to food and mouth, such as:

. the big city which avidly tucked men In"

his wretched licking of France's"

rguments which gave me food for thought."

is supporters will choke down this oppressive feeling."

Other classes of Images of psychoanalytical Interest include those relating to birth, sex. seeing, taking In. height, motion, dirt and cleanliness. Luck analyzed the writings of Stalin. Hitler. Mao and Liu Shao-ch'I and found that conclusions from imagery analysis correspond with otber psychoanalytic evaluations of these leaders- His ongoing research Is directed toward including analyses of Ixnln and Trotsky. In order to further confirm that differences In Imagery Indeed reflect personality variables rather than different cultures; tlmc-scrlcs analysis lo check changes In an Individual's Imagery patterns over time and correlate such changes with known biographic or situational tjSTONlY

factors; and determining the extent to which different subject matters or contexts evoke different kinds nf imagery. Luck hopes to validate the technique using well-known historical personages so that it may be opplied with some confidence to analysis of new leaders (or whom the only available dataeir public speeches and writings,

) hoi" employed imagery analysis to measure stress in political speeches. His technique might be used lo determine which topicsengthy speech induce Ihr greatest stress in the speaker. Imagery analysis In general appears to offer rich potential for further work to develop and validate techniques for the indirect assessment of political leaden.

Triska and) and) both analyzed word usage to measure the Influence of ideology, the former to measure the influence of ideological stereotypes In speeches to the XXII CPSU Congress, the latter to measure the ideological content of Soviet school textbooks. Finleyist of words and phrases with high Ideological content. Terminnlngy was included or rejected depending upon "whether or not ithorthand symboloncept or relationship or characteristic property clearly derived fromninistformulation"he number ofloaded words or phrasesext, divided by the total number of words, resultsraction which Finley colls the Doctrinal Stereotype Quotient (DSQ) for each text. Politburo members were rankedlo their DSQ. and DSQ was then correloted with other variables. It was found, for example, that older Politburo membersigher DSQ than younger members, those whose elite slalus as* achieved mainly in party workigher DSQ than those who had punned primarily gm-ernment careen, and Politburo mrmben directly ond extrn-slvply Involved In foreign affalnigher overall DSQ than those engaged primarily In domestic affoln.

Valueechnique developed by) and used by Eckhardl and1 to compare Ihc values of Kennedy andalue. In thbefined as "any goal or standard of judgment whichiven culture is nrdinorlly referred to as if ll were self-rvidrnlly desirable"hirty-onr self-rvidrnl valurs wrre Idrntlflrd and defined fnr ready rrcognitinn by

coders. Those Include freedom, miliiary strength, .sovereignty, ilemocracy. morality, truth, friendship, economic welfare, peace, knowledge, security, and culture. Tor subsequent analysis, these vulues may be aggregated in varioustrength values, moral values, and economic values, or according to whether the values are attributed to one's own country or to others. The number of Judgments invclvlng-the denunciation of others for nol wanting or respect inr certain values, divided by the total number of value judgmentsest, provides an Indicator of "confllct-mlndedness."

Erkhardt and White upplied value analysis to comublic speeches by President Kennedy and Premier Khrushchev to test the hypothesis that Soviet perceptions of the USirror-Image of American perceptions of the Soviet Union. Such cross-national comparison seems highly questionable, as words like freedom, democracy, and morality hovr entirely different meanings for Soviets and Americans. But the technique could be used to rank order values expressed by Soviet leaders or elite groups In order to study differences between them or to measure changes over time."

The above examples In which individual words or phrase! form the unit of analysis all Involve studying the attributes of Individual political leaders Analysis of word usage is most applicable to the study of Individuals, but it need not be limited to the individual level. In his study of participation of Soviet economic, military, legal and literary elites in the political) sought to measure the group consciousness of these elites. He hypothesised that groupnherently inimical to total party dominance. In order to qualifyistinct group, hr reasoned, an elite must perceive Itselfeparate group and he perceived by othrrs In thb way. Lodgr measured these perceptions by tallying the frequency with which the elites referred to themselves and to each other by meansepresentative collective noun. Results showed that expressions nf group consciousness, as reflected by use of collective nouns, wrre virtually nonexistent under Stalin but then increased ropidly In the immediate post-Stalin years. Scores nn thb measure of "grnupbm"pmthehlna) correlated well with lodge's of political participation by the specialist

' hr an or.IIo.IW,Chinese

see, different v.. .re Cummin, (IBM)

eneral upward trendhy temporary declines fnllnwjng confrontationspecialist elite and ihe party.

The word culegories med to analyze latent content seem simplistic to some intelligence analysts. There are so many factors whichneaker's usage of the collective noun.rgued, thai this hardly seems a valid measure of group-consciousness. Yet tbe validity of ihi*rounded In extensive sociological irsearch. Like any other statisticalcontent analysis disregards the uniqueness of the Individual case In order to focus on the uniformities in Ihe mass of cases. The many factors which Influence the usageollective noun in any Individual case are not relevant for statbtical analysis; whenarge number of cases, these other factors cither remain constant or are presumed to varynd hence to cancel each other out. When the :ame standard of measurement Is applied to two sets of otherwise comparableystematicallysamples of aitlcles from the same journal for two different years, it docs seem reasonable to attribute significant differences In frequency of certain collective nouns to differences in grmip-consciousnes* from one year to the othrr.

Analysis of Article Length and Placement

The newspaper article (or broadcast item) is the unit of analysis in many studies of massmedia. Items of information such a* column inches, headline size, position on page, and page on whichocated are combined in various ways to arrive at nn "attentionhe amount of attention devoted to various topics serves,asis for Inferences almut editorial bins, propaganda goals, social trends, or presumed audience Irnpact. Journalism and sociology students have produced many master's these* and doctoralof this type, but application to interesting political research on the Soviet Union has been limited.

An attention score canus. Jul in conjunction wilh other meusiirrs.) examined thr relationship between factional stniggles in tbe Soviet leadership and Soviet policy toward Ihecounlrc*2luctuations in Sovirl Interest In these counlrirs were measuredonthly count of article* and words devotes! to tbe undeulevelnpeil nreas by Fraoda and /iwsrfrt. This

FOR OFFJCIAVUSE ONLY

supplemented by tlin.tatlc analysis to measure the direction ol' policy change. The procedure permitted Pendlll to establish with considerable precision the dates when policy change* werewhich dates were then correlated with qualitative analysis of the factional struggle.*

Analysis of ihe amount and type of attention devoted lo various countries and general subject area, has loneainstay of propaganda analysis The Department of Defenseajor project (PAMIS/FMAP* lo content analyze Soviet. Chinese. North Vietnamese, and North Korean media. Its purpose Is lo support contingency planning for psychological warfare operations against thesebut it aboata base of some Interest for political analysis. Articles In selected journals arc coded according tn subjecteme, country dealt with, tone (favorable, critical,nd length. Pravda has been coded4 and the Peking Peoples Dally has been codedut unfortunately there are gaps when coding was nnl done regularly. Other journals and radio broadcasts have been coded for specific purposes for shorter periods of lime. Computer processing permits retrieval of data on number of Items and amount of space by country, subject, tone, time period, source, and audience.

The utility ofata base to political analysts fa heavily dependent upon the length of lime covered and lhc nature of the subject/theme hecakdown. Length of llnirf critical Importance because the analysl working with national-level aggregate data Is usually Interestn trends over lime. The PAMIS/ FMA dataeak from this point of view, buttrong In subject categories. Current procedures permit coding ofierarchically structured subjects nr theri.es. An average ofof these subjects or themes Is found In each six-page Issue of Praoda. This goes well beyond thr gross calegntles normally found In measures of medio

Other Units of Anolysis

Although themrs. words and newspaper articlesmost common units ufariety ofstrategics Is possible. These includerelationships, types of logicalof reasoning, and patlcrri_jyf co-occurrence

Axrlrndiasrocedure for coding causal relationships tocognitiveognitive map has two basic elements. One element consists of all the conceptserson (or group nf persons) uses Inven situation. The second element consists of all the assertions of causal relationships between these concepts, The concepts are represented by pnlnls. end the causal relationships by arrows between these points. This pictorial representation of the relationships between concepts fa what Isognitive map. Policy alternatives, ca.vcs and effects, goals ond values can all be reprcs.-ntcd as concepts in the cognitivearticular benefit of the graphic technique is that it permits one to see the overall structure of an argument, rather than separate nnd oftenbe techniquee! of niles and procedures for Identifying concepts ond causalfor aggregating data from two or more documents or persons, and for displayinglso Includes mathematicalfor deriving Inferences from the analysis of paths Isetsveen points on the map. Tills technique has not yet been applied to Soviet documents.

Types of premises, 0nd types of conclusions drown from tlie premises, were analyzed hy Triska ondn alternative measure of the Influence of Ideological doctrine. Statements In the text lo Ite analyzed were examined

or the presence of explicit, related premises und concltnlnns. Where the text Indicated that Its authoronsul or contingent relationship between two propositions, the pair (premise undwere extracted. These pairs were then characterized according to the nature of the premise nnd Its Interpretation us reflected in lhc conclusion"

Premises nnd conclusions were classified In several different ways. Including svhrtlirr Ihe premises ss*ere furnished Its'doctrine or by empirical

FOR

OFnpAf-ijsi only

nbscrvjtiim. If fumbhrd by rmpltlcal observation, premises were coded as In whetherere based on generalized perception* uf conditions, itccurrcncrs or altitudes, or specific perceptions of same, and according to whether or not conclusions drawn frnm them were given identifiable doctrinal IntcrpeetatIon. The results were used to test several propositions concerning the rote ofnlnbt doctrine In Soviei foreign

eveloped an Interesting and provocative procedure for logic analysis.ased on the assumption that "the stylean's thinking is directly related to other aspects of his personality and that meaningful Inferences toor psychological traits can be made from an analysis of an Individual's thinking.e Identifiedypes of Idlosyncroeles of reasoning such as suppressed premise or conclusion, appeal to force nfear or pltv when these concepts are Irrelevant toonclusion, conflicting assertions. lnp.Lc-.il fallacies etc. He also identifiedifferent "cognitive maneuvers" relating to the flow ofwitching from descriptive to normative mode, summarizing, digressing, using un-sulsst ant luted allegation, enlarging upon preceding statements, deduction from preceding statements, agreeing with the whole but taking issueart, etc. These characteristics are then used to deduce an individual's ranking on oboutifferentdimensions such as flexible vs. rigid, spontaneous vs. Inhibited, npen-mlnded vs. hlusrd. consistrnt vs. changeable, organized vs. unorganized, purposeful vs. lacking In direction, fact-oriented vs tbeory-orlrnlrd. etc.

Shnrldman hns applied the technique to such diverse communications as suktdr notes and the Kennesly-Nixon debatesn unpublished manuscriptlsn Includes analysis of speeches by Khrushchev. Unfortunately, coding for this inrtlcular approach appears to require specialized (ruining und Is so difficult as lo discourage use nf iheurther limitation Is that Shneldmon hns not defined explicit rules for correlating the thought patterns with the |HyclHiloglcid chnructrrbtic*.

Contingency analysis, described hyeasures poltcfns of association of words nr themes. A

" Tttr rod in*iplalnrd more fully In UnlrVs St oof-"si slnlserslls oWtxral dlKrrtalira.inJVitam JfiialQB,than la the Trtska aid4

word or theme is counted in the coding process onlv if it cst-occtiri wilh one or more otherhemes specified hy the researcher. Contingency analysis is bused on psychological principles of association and xhnws what is related to what In tbe mind of the communicator. It Is especially relevant to problems of psychology and psychoenolysb btrt-olso has political applications. Pool eichanges In political symbols In'newspaper editorials In five major countries Including the Soviet Union0 year period. Heorm of contingency analysis to trace changes In tbe meaning of the concept of democracy In different countries In different time periods. Thb entailed counting the frequency with which the word democracy co-occurred In the same context as symbols signifying one or the otber of the three principal conceptual components of democraticgovernment, the common people, or freedom.) used contingency analysis, looking for co-occurrence of themes rather than words, to study the role of the Chinese People's Liberation Army during the Cultural Revolution.

Computerized Anolysis

Computerized contentharacterized hy Its coding procedures rother thon by the unit of analysis. Computerized processing of text offers many advantages such as: avoidance of thr tedious and sometimes ormr-ladcn process of manual aiding; prncedurrs for coding Intenslly as well as frequency ofhe ability to re-onalyze the same data as often as desired to examine new hypotheses not envisaged at the start of the project; and capability to perform advanced statistical procedures such as duster anil fuclor analysis.

* The Stanton) Pullltcol Dictionary, used with the Oorosfri ofexnlent anatssh. rnnVs Ihe Inlensllt- nf apronsOTI vprds on ihe< ii .inc. slntnt-ncni. nnd atllie-paxlve.

Much of the academic svork with computeriVed content analysis of foreign political documents has Involved measuring the intensity of attitudes or actions of some actor toward some target, suchoreign country. Hostility Is the nllltudr mostanalyzed In thb monner. Such measures have hren used siiccessfnlly fnr empirical testing of thmrrti-cul hypotheses, but tbe only likely application to Intelligence requirements would lie for measuring changes In tlie level of expressed hostility between Inn

countries such as the USSR and China. The present start- nf Ihr arl uf computerized coding dors nut prrmil thr degree of discrimination Isrtssrrn themes ami issue orientations which is required foranalysis and which is obtainable hy manual coding.*

Putng mnrtV In thiit*ail nativeol computed >ed coding In Chlncn- urrn

Because the computer Identifies words much rosier than themes, the techniques for analyzing word usage are far more adaptable to computerized coding lhan are the various forms of thematic analysis. Some nf the measures of psychologicalome of those developed hy. arc clearly codablc by computer.

M-rious limitations tosl content unalysji have been ihr time and effort required to transform political tests into machine-tradable form, and to "pre-code" tlir lest so the computer can recognize pronoun referents and syntacticalThe former limitation svill be purl fullysvben tbe FBIS' computerized Information handling _systcm goes into operation and starts producing rnacbinrcrcadsshlc tapes of the FBIS Dallyhe latter llmllullon may eventually lie overcome by progress In ibr computer processing of natural language, or by research designs which olrviate thr need for pre-coding "

fOJVG

USE ONIY

references

) "Social Values ot Sovietof Canfllei flroofiirion. VIII,M

solierl.) Tht Sintlwe of Otetdon. Prtmxtoo.rlisrrtim University hru

) Socio/ faaflraron. Cornlvldfr.

Th*ress

Bf.BKLSON.) Conrrnr AaVyu la Coenmuniro. (ion Unearth. Clrncne, III: TV Free Pmi

ond) "Mlllcat Aliunde ed ihr CPSU HUei Wlrrt on Pailvaper presented In ihc Conference on Change and Adoplallon In the CommutiM Pnrly of ihr Soviei Union, iponsnrrd

ihr Aeoertron AMoriollon fot ihe Advonceenrnl of Slo vie Stud lev4

,1 TV AppHnrum of A'eor Mrihodnlaetn tohr Sa-ofel Ffcrpltoni of I'S PrihrsVi. ond Further Mrihrxinlnriftl fJeref-sjmirnf of Sorter Pr nvp'dmi Crmlrntul flepoet Adlit|loo..

CABNKY. Thomas) Cnnrrnr Ann.uiii. Inndon. BT.l

C.ARY.Unn. of Fjeipliosl* hjhwi Morstst-IretnhlA Omputrr Cnnlrnl Anolxlt of SovirlrOon. Gn^'aphy. oral Social Science Tetlhmas- Papr*ol4f iV Midwntonferred*

CUMMINS.) .Xiao. Ilnoo. CaWtM ondFrnonulrmd JVciitoa-Mnlfo/ Safe Prof"-tkxiol Papers In Inlemotlonol Studies,1 Rrvrrly tlllhinuon;nhllroilon<

lnlrreasuring, iheuMlr OfficialsMoncetudy of Aenertcon Pnrsldmlv" nVhactonaf Sclener. IS. pp

KOillABDT. WAom tod rUlpk r.Af Iheypothesis: Kennrtk ondnlnffmano ondlemn.rfi. Thr ConifWi nf StatlH Fneeicnhicago. New York: Aldlne-Atlicrlon.

FLFRON.r.W1 "So-let Aero Studies ond the Social Sciences: Some MelhorWoglrul Pmhlrmt In Commimlslarin Studiei. SIX (JanuaryI.

FRANK.Jnpdinr Ane.tu.tt of PniWiraf FJun.of VenW KtnrMn Sage Profr*4onol Poprrs tn lolerno-IWwl Si2 RrvedyBs oodftrisltcalmrn

CATI.rea SadtoJ ond lalemotlonol Befollons Introductorytudlet In Cnnipomllor (nmoiunlim. VIII. Spectnl Issue. "Symposium nn the Com-porollve Study of Communist Foreltn Pnllrles"

CFOflCF.IffflflJ/iwiopinofanstonetersonCo _V

Tfce 'Opr.Hanoie^leeted Ap-

pmorh la the Stud* of Political Leaden andosortW Sf-Aer (Wrrrt, XIII.tt

MKftMANN. Maraaret)nd FonHiinhiv InWnou.ornra PotkUt. Urn York: Johnon.

(Hi" I-Pnllcy Behavior Uslno

aras1etlUlc. nl PoliticalKF.ON. Pope* delivered ot Annual Meetlnf of (hr American Political Sclencr Assodotlon. San Fionclsco. Calif.

Support for

BrTiTinrv. Anlcol twof Somr

lloan" Central inirlUfroes- Ajsmcy. tXflce ol Polean

IIOI.STI. Ole tlWM)nnh-I. far thr Snrtof Vlenee. andBeodlor..

(lOTD'^Oprtotlonol Code' Beliefod'-honk '* Unite Unl*enlty. ml.oeoll. Oclohrr IB7S

IIOnFl.ICK.. Boss JnhliUHltm I)) Thr Slu.f* ofHrlttt-vftecleu nf OrelitnnT/imrv IXrlatrd /Ippnwnri Brpnn* Sanlo Monleo Banal Oirp

JFBVIS.MI "Thr Cr-ts nf tKr Seienllflr Slod.llles Aa IjomlMllon ol ihrnnlenln klous Knorr and |orars li (tmcaaa.nntrma'tntn lotrrMlVMitl PooWrt. Mncelor. Mncton llots-n-

(Is Prrss

)Htonvtlrn1o rem forrttn Alrrffi AaMB^atl IFfHoof

Amedcan Institutes fnt Research

KB INCn A.n Fsplorolloo o| the'

In Clitiso Tliroufli Cnnlrot AnoKsk.' .lilonp

lASSWFIJ.olhoaondIlftM"/MIT poperkort rJlahai (aml-hicT Pnm.

IIOVll

Prni

I.FITIS.1 TV (Ipemrioeof Codr nf ihre* VonVII!

A iiudg nfencir.in- lire

FOR OFFiaftt^USe-ONlY

rOR OFftCHt USF. ONLY

Nathan. Elia nernaul and flavmnnd I.nlllhum Irnaies iJ'cWrf Puriiln.p..

UAO. Kuan<.dvrn* eivdMtlnr,Pre-IVMpllnniia'Threat: The US awl India.tv.O-flT

l-GDCF.ei fair Alltrudrl SinceMerrill Fuhl*oln|"

LUCK.) "AApproachcadet Prnooai.lv Miller. Stalin. Ma* andShao-chV SnWlei la fwan l'i Convnunlrm.p 4.

M ITCH ELL nobrrt) "TW Ute ofpUnaiorywWlc.

NORTH. Robert C. Ok fl Holitl. Ccorie M. ZaiUnovteh.a A.) Contentandbook irfrh Application! for ike Study of tatrmaltonel OtJU. Evans ton: Northwestern University Prru

OSGOOD.) "TheMode! and PHevani IWarchn Ithlrl de Sola Pool.rend. In Conlral Anafutti. Urhana' University Ii Frew.

ranl. JlT) "In SovietnP. Hoffmann and Frederic J. FTeron. Jr. rdi.Conduct af Sortrr Forrijn Mica. Chlcaan. New York. Aldlne-Alheltnn.

POOL. Ilhlel de SoU.) Trerwi fx Comtrxtrli. UnSana: L'r.lwnJty ot Illinois.

Cnolenl Analysis for Intelligence Purposes.'

World fohtin.

ne Pmrlre Preiiumpumlice Slndu "/

robtuel SaBRatffc. Cambridge:re**.

RUSH. Myronhe RUe of KhmArhrv. Apprndli Two.r nf Esoteric Com mimical Inn In Soviet FallWlc Allalri Preti.

SHNEIOMAN. Edwin S.A hyehornetcal Analysts of PorWlcal Thinking: The Kennedy-Nl.nn 'Great Debates* and the KennedyKhnrihrhev film Deleter.'" Harvard University, ml-

ilWl) "The Ij-iIc nfnmni and m. a

May. rdi. Trl'iUlon aad Human "enactor. New Ynrl Apptrtim-Ceniuev-Cna'ti

(lOK) "Lrsflcal Contenl Anil.tR An Kspllcallmt .4

Slyln ofn Cmrge Cerbnri cf af.aIvtU of Comnma^-crionejr. New York John Wiley and Son*

avldanB^mertcan Foreign Policy AWIihIm, -Conl-nt Annlvsli nf Ulir Aiik-ululi-,'o/ Omflm nepjinUoiiViyi. ppj

STONE. Philip) "Report nn the Workshop on (Violent AnaKrl. In the Sodal Sciences." Social .science lafommtlam. XIV. L

STONE. fWhp J. ttS) The CenereJomnMc-Appmacn lo Conienl AnolurU. Camlirldj' in<

TRISKA. fan F. and) Snilei ForrtfnNew York: The Macmlllan Co.

1 "The Political Rolct otple't Uhrratmn Army ai Perceived hy the Chinese Communist Prni In Ihe Cultural RevoluUon."end Srurflei. X.

WELSH. WltUra A.)nlllple-Straieiy Apprnach Inn Cnmparallsr CnmmimKt Polllleetmpirical and Osianlilatlren Frederic J. Fleron.d, Cumniii'iiK Sniiflei aad ihr Social Sciencea. mRnndn.

Introduction The Cnmparallsr Stud, ol

Polillcal Leadership in Communis!n Carl Peck el al.munuf PoArleWktp. New York: OartfJ MelavCn

WHITE. Ralph) MaV Aaofufb: Saint and IVn/ihe Mrthod. Ann Arbor Society (nr (he hycholofilcal Slitdy nl Social Issues

WINTER. Davidl)"Whal Makes thetuehrjoiv T'idav.9

WINTER. Duvtd C. andtewartRI "Consent Analystsechniquearsal an; PoliticalnwarMaftral EwrMnariMtden. Nrw Vork- The Free Peesv

WONC. Paul ildfiT)and Analysli nf Documentary Malerlah Inun Communldtlaa Sarvrn. VII, Jl.

Original document.

Comment about this article or add new information about this topic:

CAPTCHA