Bilateral relations between countries and the complexity of newspaper editorials
The archival measure of integrative complexity, developed by Suedfeld and Rank (1976), is one of several procedures for the systematic study of documentary evidence to infer various characteristics of the interactions between countries (e.g., Axelrod, 1976; Ertel, 1972; George, 1969; Hermann, 1980; Mehrabian, 1967; Winter, 1987). Integrative complexity is defined as the joint operation of two components of information processing: differentiation (the perception of 'The University of British Columbia. 601 0162-895X/92/1200-0601$06.50/1 ? 1992 International Society of Political Psychology
Suedfeld several dimensions in a stimulus situation or of several perspectives on the situation) and integration (the recognition of relations among differentiated dimensions or perspectives, e.g., as interacting, as capable of being integrated, or as being relevant to some overarching event or idea). Evidence of differentiation and integration in verbal (written or oral) statements can be reliably assessed by trained scorers (see Method). A large number of studies have shown level of complexity to vary with environmental and personal stress, political ideology, and social role, inter alia (Suedfeld, Tetlock, & Streufert, 1992). Most of these studies have used governmental documents or the pronouncements of officials as the data sources. Level of complexity is important because it indicates differences in decision-making strategies: for example, sever- al studies have confirmed the association between reduced complexity of governmental communications during international crises and imminent war (Suedfeld & Tetlock, 1977; Suedfeld, Tetlock, & Ramirez, 1977; Suedfeld & Bluck, 1988). Periods of international crisis are associated with reductions in complexity not only in official documents but also in such diverse materials as personal letters, editorials in a scientific journal, and presidential addresses to the American Psychological Association (Porter & Suedfeld, 1981; Suedfeld, 1981, 1985). These data may indicate a general state of rising tension within a society, which in turn may affect information processing and decision-making at both the private and public levels: for example, it may influence readiness for or against negotiated compromise as the solution to problems. The overwhelming majority of archival materials used in such studies have consisted of governmental communications, memoranda, policy papers, and the like, and the private or professional correspondence and speeches of various societal elites. Relatively little attention has been paid to the mass media, except as they reproduce the other kinds of materials. Systematic content analysis of media content has been used to study various aspects of national and international politics-e.g., to identify publications that served as disguised propaganda outlets for Nazi Germany, to monitor sequential changes in Comintern policy toward and away from cooperation with noncommunist countries and groups (Lasswell, Leites, & associates, 1949). Structural-i.e., complexity-analysis may also provide interesting data. The current study focuses on the association between international relations and the complexity of editorials in periodicals of broad readership and impact. Further, it looks at the concomitants of events that are more "normal" than international crises: that is, events that affect relations between pairs of countries positively or negatively but for the most part without carrying major risks such as war. Both communist bloc and Western publications were included to assess possible differences between the two types of system in this respect. Three major questions are addressed: 602
Bilateral Relations Between Countries and the Complexity of Newspaper Editorials 1. Will the complexity of newspaper editorials reflect lessening or increasing levels of international tension even when events do not reach the level of crisis but are of a more routine nature (such as trade agreements, diplomatic visits, U.N. votes, and the like)? Our hypothesis was that such changes would be found, showing the effects of "disruptive stress": i.e., changed complexity in times of increasing tension. 2. Will there be a difference between publications of communist and Western countries in this regard? One might expect so. In the former case, until very recently newspapers were instruments of the government and supported the government's current policies. While American and Canadian newspapers may be more or less subject to subtle governmental reinforcement (e.g., "leaks" of restricted information) or the lack of it, as well as needing to stay in step with public opinion, editorial philosophies are developed within the newspaper and may or may not be congruent with governmental views (cf. Charles, Shore, & Todd, 1979; Downing, 1988; Dreier, 1982). On the other hand, the effects of international tension on complexity have been shown to be quite pervasive in society-but these findings were based on times of crisis. We predicted that there would be a closer association between tension and editorial complexity in the communist than in the Western sources. 3. The third question involves the absolute level of complexity rather than changes in complexity. Because of previous findings, we predicted that Pravda will generally show lower complexity-more dogmatism, less flexibility and recognition of alternate points of view-than the Western publications. The relevant findings showed such differences between American and Soviet govern- mental statements (Suedfeld et al., 1977; Tetlock, 1988; Wallace & Suedfeld, 1988), although the current reformist leaders of the U.S.S.R. may break the pattern (Tetlock & Boettger, 1989). METHOD The two types of data used in this study were 1. Editorials from national newspapers, scored for integrative complexity. 2. Events involving both members of each pair of countries, scored on a scale of extremely positive to extremely negative. Source Documents The documents analyzed were editorials that appeared in the newspaper of a particular country (the "source country") and had as their topic relations with, or the actions of, a specific other country (the "subject country"). The source 603
Suedfeld countries were the Soviet Union, Canada, and the United States; the subject countries were these three plus the People's Republic of China. The choice of newspapers was based on the national and international prominence of the publications. This was no problem in the case of the U.S.S.R. Pravda was an official national newspaper with wide domestic and foreign circulation. Its 11 million subscribers and up to 40 million readers included the elite of the Soviet Union: members of the intelligentsia (22% of the readership), engineers and technicians (18%), and 90% of the members of the Communist Party. More than any other Soviet newspaper, Pravda had the "task of presenting Party policy, in the most unambiguous and authoritative form, both for domestic consumption and for the world outside" (Roxburgh, 1987, p. 79). The New York Times is certainly one of the most prestigious and widely distributed American dailies. In a 1982 poll of newspaper publishers, editors, and professors of journalism, it was a clear first choice as the best newspaper in the country, a tradition it has long held by "telling the news with completeness and integrity" (Emery & Emery, 1984, p. 653). It has long been considered the newspaper of record, and has been described as "a key part of every Washington journalist's and every legislator's morning. The Times is an indispensable source for writers, editors and embassies. It's sober, steady and hardly ever rocks the establishment boat; it's been viewed by foreign governments as a reliable chronicle of American positions" (Tataryn, 1985, p. 71). Although Canada has no real equivalent of either of the above two publications, the Toronto Globe and Mail calls itself, and is generally considered to be the closest approximation of, Canada's national (rather than local or regional) newspaper (Tataryn, 1985). Like the other two, it is distributed nationwide and appeals primarily to a relatively well-educated readership. Relevant editorials (those dealing with any of the four subject countries) were collected from these three newspapers (for Pravda, using the English translations in The Current Digest of the Soviet Press) for every year between 1947 and 1982. There were some problems in selecting material to be scored. Most editorials in all of the newspapers dealt with domestic issues, and in some years few editorials appeared dealing with some subject countries (both the Times and Pravda had relatively few editorials related to Canada). Furthermore, these editorials varied greatly in length. We decided to use databases of equal size as far as possible. For each source newspaper during each year, all editorials dealing with the subject country were identified. The complete texts of up to five of these (all, if there were only five or fewer in the sample; otherwise, five were selected randomly) were copied into the computer, which then used a random numbers program to select five scorable paragraphs. The selected paragraphs were pre- pared for scoring by the removal of any names, dates, or other specific material that would identify either the source or the subject country. 604
Bilateral Relations Between Countries and the Complexity of Newspaper Editorials Complexity Scoring Each paragraph was then scored for integrative complexity. The two scorers had participated in a one-week training workshop and had reached acceptable levels of reliability (r = .85 or higher) with the trainer on an extensive set of training passages. In addition, they had performed complexity scoring on previous research materials, and had maintained an interrater reliability of at least 80% with each other. This is a more rigorous criterion than the reliability coefficient, as it requires agreement on specific scores rather than merely a correlation. The scorers rated each paragraph independently, using a detailed scoring manual as an additional resource (Baker-Brown, Ballard, Bluck, de Vries, Suedfeld, & Tetlock, 1992). The scoring system uses a scale of 1-7, with each paragraph scored separately. The bottom half of the scale (1-3) identifies increasing levels of differentiation: the recognition and utilization of different stimulus dimensions, and/or of different perspectives, in judging the stimulus (which may be another person, an event, an idea, a country, etc.). Scores of 5-7 recognize increased integration (syntheses, interactions, trade-offs) among the differentiated dimensions or perspectives, with 4 as a transition score (see Table I for examples). Paragraphs that consist entirely of factual statements (e.g., statistics, historical events) or quota- Table I. Complexity of Editorials: Examples Score Year Countries Paragraph 1 1951 USSR-Canada "Ordinary Canadians are meeting 1952 in a situation of continually deteriorating living conditions. The number of unemployed, which has reached almost 200,000, is increasing in the country. Queues of unemployed and homeless stand at the doors of charitable institutions waiting for Christmas alms." EXPLANATION OF SCORE: Only one dimension or perspective is presented, i.e., the plight of unemployed and homeless Canadians. 3 1951 Canada-USA "Events have justified these assertions all along the line. The facts in the UN survey, together with what we see happening on this continent, completely discount the views of those Pollyannas who say inflation has just about run its course. There is, in fact, no limit to its possible extension so long as the factors causing it are still operative. Stockpiling has not denuded the world of essential materials. But it has boosted prices to a new high table-land undreamed of a few years ago. Higher military production at home has created new volumes of consumer demand, so that countries abroad are virtually unable to purchase with their new flood of dollars the full range of American products which would be useful in stabilizing their respective economies." EXPLANATION OF SCORE: Several causes of inflation are recognized: stockpiling, military production, greater consumer demand, and the inability of other countries to buy American products, even though they have the money. 5 1954 USA-PRC "The British delegation also found out that Mao Tse-tung wants Britain to spearhead a drive against US foreign policy, and wants trade with Britain and other Western countries. The latter, presumably, is the bait for the former." EXPLANATION OF SCORE: Not only are two dimensions recognized (trade and foreign policy), but a causal inference integrating the two (the former is being used as a tool to obtain goals related to the latter).
Suedfeld tions, without evaluation or commentary, are deemed unscorable and are replaced. Interrater disagreements of one scale point are resolved by discussion; paragraphs where disagreement reaches two or more points are discarded and replaced. Not all of these paragraphs were used in the final data analysis, because there were years during the sampling period when bilateral events involving any pair of countries could not be characterized as having positive or negative valence (see Events, below). Table II shows the number of years for which both an acceptable sample of editorials and valenced events were available. For the total of 133 country/years, the number of paragraphs used from relevant source editorials (those dealing with the particular subject country) were five paragraphs for 50 years each from the Globe and Mail and the Times (total of 250 for each of these two), and from Pravda, five paragraphs for 26 years (= 130) plus seven years with fewer than five relevant paragraphs each (total = 24), for a total of 654 paragraphs included in the data analysis. Mean complexity ratings were calculated for each newspaper, for each year, for each subject country. These means were used in the subsequent data analyses. Events A total of 315 relevant historical events were identified from the Encyclopedia Britannica Book of the Year in the years 1947-1982. This encyclopedia was used because its yearbooks note significant events involving major nations soon after the events occur. Relevant events were operationally defined as those involving one of the source and one of the subject countries. In preparation for further scoring, a description of each event was transferred to an index card. All possible precautions were taken to remove identifying information such as the names of the countries, well-known individuals, and dates. All cards were then scored independently on a scale of -3 to +3 by five research assistants who were uninformed about the topic of this particular study. The raters were instructed to assign negative numbers when in their judgment the Table II. Scorable "Country/Years"a Subject Source Canada USSR China USA Total Canada -14 9 27 50 U.S.S.R. 4 - 13 16 33 U.S.A. 13 27 10 -50 Total 17 41 32 43 133 a Number of years for which both relevant editorials and valenced events were available. 606
Bilateral Relations Between Countries and the Complexity of Newspaper Editorials Table III. Positive and Negative Events: Examples Score Year Countries Event Positive +2.6 1970 Canada-P.R.C. Diplomatic ties established. +2.4 1981 U.S.A.-U.S.S.R. Trade increases nearly 50% over previous year Neutral +0.4 1978 U.S.A.-Canada U.S. considering natural-gas pipeline through Canada; mixed economic results anticipated for Canada. 0.0 1971 U.S.A.-P.R.C. U.S. president says forthcoming visit to the U.S.S.R. has no connection with visit to P.R.C. Negative -2.4 1980 U.S.A.-U.S.S.R. U.S. boycott of Moscow Olympic Games. -2.8 1960 U.S.S.R.-U.S.A. Soviets shoot down U.S. aircraft over Bering Sea event was characterized by or led to international tension, hostility, conflict, strained or reduced interactions, etc.; and positive numbers when the event reflected or led to friendship, friendly trade and exchanges, alliances, agree- ments, treaties and the like (Table III). Interrater correlations across all events were .82-.90, indicating a high level of agreement. Six events, which received ambivalent ratings, were dropped. Although the raters had experience in research on international relations, they were not professional experts. However, the high interrater agreement indicates that the stimuli were sufficiently clear-cut that no great degree of expertise was required to categorize them. Event ratings were then averaged within each year for use in further analyses. RESULTS As the relations between two countries became more positive, there was a significantly higher level of complexity in the editorials of each country referring to the other, r = .26, p < .001. This pattern did not differ significantly among the countries. Table IV shows mean complexity scores. The overall means were 1.83 for years with positive relations and 1.62 for years in which relations were negative, F(1, 132) = 3.67, p < .06. A 2 x 2 ANOVA showed significant differences in integrative complexity among the countries in which the editorials appeared: Ms = 1.92 for Canada, 1.81 for the U.S.A., and 1.35 for the U.S.S.R., F(2, 132) = 16.10, p < .0001 (see Table IV). The difference between Canada and the United States was not statistically significant, but both countries differed significantly from the Soviet 607
Suedfeld Table IV. Mean Complexity Scores Positive Years Negative Years Country (No. of Yrs.) (No. of Yrs.) Source: Toronto Globe and Mail (Canada) Subject U.S.A. 1.99 (18) 1.68 (19) U.S.S.R. 1.93 (9) 1.90 (5) P.R.C. 2.04 (7) 2.00 (2) Source: New York Times (USA) Subject Canada 2.09 (7) 1.98 (6) U.S.S.R. 1.88 (12) 1.58 (15) P.R.C. 1.97 (6) 1.50 (4) Source: Pravda (U.S.S.R.) Subject Canada 1.53 (3) 1.00 (1) U.S.A. 1.32 (8) 1.59 (8) P.R.C. 1.18 (5) 1.24 (8) III* -II - III.- - I,~ Union by the Tukey test for pairwise comparisons. There were no significant differences as a function of which country was being written about. The three countries differed in the prevalence of positive and negative interactions. The means were +0.64 for Canada, +0.15 for the U.S.A., and -0.08 for the Soviet Union, F(2, 132) = 3.13, p < .05. None of the pairwise comparisons was statistically significant. DISCUSSION As predicted, integrative complexity was negatively correlated with international tension. This finding extends previous data indicating that international stress is associated with reduced complexity. Unlike the situation in previous studies, the events included here consisted primarily of minor or routine episodes such as international trade transactions and political negotiations. Event-related changes in complexity, even under such mundane conditions, may be an indication of how useful the measure is in tracking the impact of world events. Our data do not indicate to what extent there is an actual change in how editorial writers process information as opposed to deliberately attempting to manipulate the tone of public discussion (Tetlock & Manstead, 1985). One important implication of this finding is the unexpected pervasiveness of tension-related complexity decrease even when the precipitating events are with- in the normal range of international relations. The fact that societal opinion leaders are affected in this way by impending or actual war may have been less 608