University of Bergen

Any analysis of the conditions for the internationalization of the social sciences must
start out from descriptions of changes in the communication flows at three distinct
levels of the research process:

- at the level of the commercially and publicly circulated products, the reports,
the papers, the volumes;

- at the level of direct scholar-to-scholar interaction, whether in quick international encounters at conferences and congresses, through joint research ventures, or through long-term immersion in several cultures in the course of cross-national research careers;

- and at the level of the raw data for research, whether accessible through shortterm
ad hoc arrangements between scholars or centres, or through organized exchanges
within archival networks.

Increases in the flow of communications at any one of these levels do not necessarily lead to increases in the volume of actual attempts at cross-cultural or crossnational comparisons: a great deal of technical, methodological and substantive information can be spread across cultural or national borders v/ithout any direct impact on efforts of cross-societal comparison. But the mix of the three components clearly affects the quantity and the quality of comparative research: you need a maximum of documentary evidence and a maximum of research reports for all the units you want to compare, but you also need raw data to control your comparisons and you need direct contacts with contextually immersed scholars to check your evaluations of the data and your interpretations of your analytical findings.

It is easy enough to warm against one-sidednesses in the approach to comparisons:

- you can only get so far in the direction of systematic comparisons on the basis
of "process-produced" documentary evidence and on the basis of such "frozen" outputs
of single-culture or single-nation studies as papers and monographs;

— it is even easier to get stuck in single-society perspectives through direct immersion

in context after context without any concern for the development of overarching
models for the explanation of similarities and differences;

- and you may also get yourself into frustrating impasses through overreliance on
abstract manipulation of "instant data" in ignorance of earlier research on each unit
and in isolation from contextual expertise.

It is vastly more difficult to establish any rules about optimal mixes. I have toyed
with the idea of analyzing every documented case of comparative research along
three basic dimensions:

(1) theory-directedness -at the one pole purely empirical comparisons on similar variables without any consideration of hypothesis testing or model construction ("more car accident casualties per million car-miles in country X than in country Y"), at the other pole explicitly theory-directed comparisons designed to test a set of alternative models against a set of alternative empirical indicators;

(2) contextuality — at the one pole single-variable or single-correlation comparisons without any consideration of conditioning contexts, at the other pole comparisons of configurational complexes with only minimum losses of contextual information;

(3) globality — at the one pole paired comparisons of cultural and/or geographical neighbours, at the other pole comparisons of all culturally or politically defined societies across all regions, whatever their historical fate or their location in the networks of communication and influence.

In my checkered career as a comparative political sociologist I have been travelling
up and down each of these three dimensions:

- I have engaged in highly mundane empirical comparisons of the percentages of samples for each of the Nordic populations acquainted with citizens in one of the other countries,¹ but I have also attempted elaborate tests of complex macro-theories of political development;²

- I have compared teachers in seven European countries on such simple correlations at those between party preference and "F"-scores,³ but I have also been heavily involved in an effort of multi-cultural contextual immersion in a project on the politics of the smaller European democracies;⁴

- and I have engaged in simple two-country comparisons,₅ and in fact believe that this is, at least for the present, the most fniitful of all styles of comparative research on macro-micro configurations, but I have also helped to push forward the movement toward wo rid-wide comparisons in the Deutsch-Russet style.6

The ideal mixes between theory-directedness, contextuality and globality of perspective are all very expensive, even for the richest of our research centres: direct tests of ambitious theoretical models can rarely be achieved without costly investments in co-ordinated field operations, and detailed controls against contextual information cannot be achieved without extensive interaction among scholars fully familiar with conditions within each of the societies compared. It is easy to point to deficiencies and imbalances in all the projects of comparative inquiry so far completed : weaknesses in the fit between the theoretical structure and the actual empirical indicators used in the tests, distortions and misunderstandings in the interpretation

of contextual conditions, difficulties in the sampling of units for comparison or in the rationale for a focus on particular regions of cultural homegeneity. It is easy enough to criticize, it is much more difficult to devise workable compromises between the strictures of research economics and the ideal requirements of theory-directedness, contextual depth, and rigour in the choice of units for comparison.

One possible compromise solution is the Data Confrontation Seminar: this paper
is a plea for continued exploration of the potentialities of this new style of international
research interaction.

The DCS: The Ideal-type Construct

The Data Confrontation Seminar is an organizational novelty: it was not possible before the arrival of the large-scale computer and it did not make sense as a device in cross-national research before the establishment in country after country of archives of computer-readable data.

The essence of the Data Confrontation Seminar is intensified interaction in joint analysis: interaction between scholar and computer and between scholar and scholar. This cannot be achieved through the traditional presentation and discussion of papers: papers represent so many "freezings" of analysis options and allow very little leeway for recalculations during discussions of divergencies in the findings. The Data Confrontation Seminar seeks to increase the options for reanalysis and to intensify the interaction over the interpretation of divergencies through regular interaction with a computer holding the original data. Participants in this novel type of seminar do not present papers in advance but machine-readable data files: these are submitted, with the appropriate explanations of codings and definitions, to the organizing centre some months in advance of the actual event, are "reformatted" for the established computing system, and are subjected to a series of standard analyses. On arrival at the seminar the participants are presented with the initial output of analysis results for each set of data and a series of questions about the interpretation of similarities and discrepancies formulated by the rapporteur general: these constitute the agenda of the two-to-three week course of intensified interaction.

The advantages of this procedure are obvious:

Accelerated internationalization — participation in such seminars wil increase the familiarity with the handling of data for other countries vastly faster than any alternative method such as lecture courses, exchanges of papers, or short-term visits at foreign centres.

Spread of analytical skills - analysis procedures tried out in one country will be much more rapidly applied to parallel bodies of data in other countries and this will accelerate the diffusion of new techniques, whether of statistical dissection and inference or of computer programming for social science uses.

Stimulation of national efforts in the social sciences — participation in Data Confrontation
Seminars will help to "jack up" the levels of aspiration in the les advanced

countries and to stimulate investments of time and resources in the development
of broader data files for further analysis in each country.

Encouragement of collaborative efforts of dat a-gathering - Data Confrontation Seminars will help to pinpoint lacunae and to establish priorities for fresh jobs of data assembly (e.g., sampling of parish registers to stretch a demographic time series) or data gathering (sample surveys) and may provide the basis for programmes of co-operation among a number of research centres.

Such "multiplier-accelerator" effects will not necessarily be generated automatically: to ensure a maximum of cumulative linkage per currency unit invested it would appear important to bring in an internationally legitimate sponsoring organization to act as a liaison between the organizers of Data Confrontation Seminars and the national funding agencies likely to finance follow-up efforts of further comparison. A body such as the International Social Science Council might serve this function if adequately financed: it has established close contacts with national academies, research councils and other funding agencies throughout Europe and the West and should be in a position to link up efforts pursued within Data Confrontation Seminars with activities already under way within the different countries.

A DCS will require a minimum of one year of intensive preparation. The timetable of each of the Seminars will obviously vary as a function of the amount of material already coded and computerized in each country. A typical sequence of preparations might be:

Month 0: Decision on basic structure, appointment of director and establishment
of basic budget.
Months 1-3: Establishment of inventories of available data files for each potential
country.
Month 4: Preparatory meeting with 3-4 core experts for decision on choice of
countries and specification of core variables.
Month 5: Contracts with participant scholars/centres.
Month 6: Delivery of files for reformatting and initial analysis.
Month 10: Distribution of initial output to participants.
Month 11: Preparation of seminar "agenda" based on output.
Month 12: 2-3 week Seminar.

For fields where the data files have not yet been computerized for more than one or
two countries the timetable will obviously be much longer: the delay between the initial
meeting and the delivery of data files may have to be stretched to up a year or more.

The First DCS: Lessons for the Future

This is perhaps the one most important lesson of the initial exploratory DCS organized by the Inter-University Consortium and UNESCO in April, 1969: the data sent in for reformatting and analysis were new to most of the participants and much of the time set apart for discussions of strategies of comparison was in fact devoted to

elementary checking of the records and to initial single-country tabulations and calculations. The Ann Arbor Seminar was a remarkable success at the technical level: it taught the participants a great deal about the potentialities of the OSIRIS programming system and it whetted their appetites for new styles of statistical analysis. But there was very little in the way of direct data confrontation: a number of parallel analyses were carried out nation by nation but there was no time for the next step, the actual formulation and discussion of conclusions at the cross-societal level. The Seminar was highly successful as an exploratory step» but it did not achieve what we hoped at the level of direct comparisons: the data, had simply not been sufficiently digested to allow anyone to move on to that level.

There were good reasons for this: the organizers had decided to focus the Seminar on ecological data for localities within each of ten countries but only in three or four of the countries had data of the type sent to the Seminar been extensively used in earlier researches.7 The seminar has clearly helped to accelerate the development of ecological data files in countries so far untouched by this style of research but this represents only a first step in the process of internationalization. In the future, it will clearly be essential to focus such Seminars on data already extensively explored within each of the countries to be covered. For most types of data, this will mean that the initial Seminars will have to be restricted to fewer countries: there is bound to be greater analytical pay-off from concerted work on a few well-explored data sets than from parallel work on a large number of little-known files. This is the strategy chosen for the next DCS to be financed by UNESCO: the "junior" Seminar planned for August-September, 1970, in Bergen. This will also focus on ecological data but for three countries only: France, Norway and Poland. And the theoretical focus will be sharper: the purpose will be to test hypotheses about the sources of local inequalities in educational achievements. This does not necessarily guarantee success but at least it is a strategy well worth exploring. A similar scheme is under consideration within Scandinavia. The group of political ecologists from Denmark, Norway and Sweden who met at Ann Arbor propose to continue their co-operation through joint explorations of the potentialities of the same analytical scheme: Gösta Carlsson's model for the spread of political innovations through the evening out of differences in the socio-economic basis for party support throughout each national territory.8 Through such follow-up Seminars it should be pos:sible to fill in the gaps that were left open at the first stint at Ann Arbor: the more the data have been worked through nationally in advance and the more marked the theory-directedness of the analysis design, the greater the chances for effective international confrontation over issues of comparison.

The Future: Varieties of Possible DCS Themes

The first DCS's focus on ecological data: typically aggregated statistics from censuses,
enumerations and elections. Clearly, a DCS can be based on any set of potentially

comparable data sets for any set of countries found worthy of comparison: not just data from the official bookkeeping machineries but also data from documentary records, folk literature, mass media, biographical files, sample surveys, psychological tests, anthropological field observations.

What is important in each case is the formulation of a framework of hypotheses for testing and the specification of data requirements. The less structured the framework and the more diffuse the data specification, the less the likelihood of fruitful confrontations: the DCS will degenerate into just another exchange of single-society tidbits.

I hold no brief for any single set of collective priorities and will be happy to encourage attempts in a variety of directions but I do see some very exciting possibilities for the next three to five years. Let me end this brief note by spelling these out very schematically by way of examples:

One set of Data Confrontation Seminars might focus on the analysis of indicators of social, cultural and political development: they would be designed to ensure comparisons of time series data for changes in stratification and settlement structure, in the reach of the central and "counter-central" communications networks, in the diffusion of literacy and educational facilities, in the levels of participation and partisan mobilization, and in the channels of recruitment to elite positions. These are all fields attracting the active interest of scholars in a great number of advanced countries but still virgin fields of comparative research: the few attempts at cross-national analysis recorded in the literature hardly amount to more than initial explorations and exemplifications.

Examples of such Seminars would be these:

1. Changes in Political Oppositions in the Course of Democratization

Units of analysis: smallest administrative units, communes in most of Europe, counties
and incorporated towns in U.S. (difficulties in the U.K., must use parliamentary constituencies).

Theme variables: strength of each party in national elections before and after major legislative
or autonomous (immigration, increase in numbers meeting franchise criteria) expansions
of electorate, roughly from the 1890s through the 1950s .
Background theory: S. M. Lipset and S. Rokkan, eds., Party Systems and Voter Alignments
(New York: Free Press, 1967); S. Rokkan, "The Structuring of Mass Politics in the
Smaller European Democracies: a Developmental Typology", Comparative Studies in
Society and History, vol. X no.: 2, 1968, pp. 173-210.
National data and studies: extensively inventorized in S. Rokkan, "Electoral Mobilization,
Party Competition and National Integretation", J. LaPalombara and M. Weiner, eds.
Political Parties and Political Development (Princeton: Princeton University Press, 1966)
and in S. Rokkan and J. Meyriat, eds., International Guide to Electoral Statistics, Vol. I:
National Elections in W. Europe (Paris: Mouton, 1969).

2. The Democratizations of the Recruitment to Elite Positions

Units: all members (alternates) of national legislatures since the early periods of partisan
conflict; samples of top administrators for selected time points since the 1850s.
Theme variables: place of birth; father's occupation; completed education; place of residence
at time of election (appointment); ethnic origin; religious affiliation. For elected
officials: occupation before entering electoral politics, early political career, occupation, if
any, after period of elective office.

National data and studies: extensive files of biographical statistics established or in preparation
for parliamentarians in Austria, Finland, France, Italy, the Netherlands, Norway,
Sweden, Switzerland.

Another set of Seminars might focus on the analysis of recent and current data on the transformation of social structures and cultural-political oppositions in Europe and the West. Most of these data would come from sample surveys, but these would, whenever feasible, be supplemented with aggregate statistics and institutional data to allow multi-level analyses. A major effort should be made to inventorize the available data for differences in social, cultural and political orientations between generations and between educational strata. Two topics might be given top priority:

1. The Uses of Literacy in the Age of Television: data on reading, listening and viewing
habits by age, education and exposure to urban-modern information stimuli, data on
acceptance-rejection of different styles of consumption and leisure use, data on the
"generation gap" and on reactions to the "revolt of the young".
2. The Fate of National Identities under the Impact of International Communication:
data on the strength of national ties (as expressed in reactions to sports events, consumer
choices, choice of holiday sites, reading matter, etc.), by age, education and level of
international exposure (linguistic skills, travel, acquaintances, organizational ties, media
information), attitudes to specific efforts of integration and to potential international
partners.

Data for most of these variables should be easy to assemble for 4—5 countries in Western Europe, quite a lot of such data can be retrieved from such repositories as the Zentralarchiv, the Steinmetz Stichting, the Roper Center and the Inter-University Consortium. What is needed to move forward toward serious comparisons through direct confrontations are initiative and motivation, theoretical imagination and organizational skill.

There is no way of ensuring foolproof success but the cause is worthy of quite a
few failures.

---

NOTES

1 K. D. Jacobsen, S. Rokkan og N. Vetti, Kontakt og samarbeid mellom de nordiske folk (Oslo, Institute for Social Research, 1957) mimeo.

2 S. Rokkan, "Models and Methods in the Comparative Study of Nation-Building". Acta Sociologica, 12 (2) 1969: 53-73.

3 S. Rokkan, "Party Preferences and Opinion Patterns in Western Europe" Int. Soc. Sci. 8.7 (4) 1955: 575-596 and "Ideological consistency and party preference", a paper from 1956 to be printed in S. Rokkan Citizens, Elections, Parties (Oslo: the University Press, and New York: D. McKay, 1969). For a fuller discussion of macro-controls in micro-comparisons see S. Rokkan, S. Verba, J. Viet and E. Almasy, Comparative Survey Analysis (Paris and the Hague: Mouton, 1969).

4 S. Rokkan, "The Structuring of Mass Politics in the Smaller European Democracies". Comp. Stud. Soc. Hist. 10(2) 1968: 173-210.

5 S. Rokkan and A. Campbell, "Citizen Participation in Political Life: Norway and the United States of America". Int. Soc. Sci. J. 12(1) 1960: 69-99.

6 R. L. Merritt & S. Rokkan, eds., Comparing Nations (New Haven: Yale University Press, 1966).

---

7 For information on efforts of ecological data archiving, see M. Dogan and S. Rokkan, eds., Quantitative Ecological Analysis in the Social Sciences (Cambridge: M. I. T. Press, 1969).

8 G. Carlsson, "Partiförskjutningar som tillväkstprocesser" Statsvet. ts. 66(2-3) 1963: 172-213, "Time and Continuity in Mass Attitude Change". Publ. Opin. Quart. 29(1) 1965: 1-15.