Figure 1-Democracy Rating
May 8, 1995
Christopher R. Kedzie, kedzie@rand.org
This inquiry is intended to begin substantive examination, both empirical and theoretical, into the relationship between new information and communications technologies and democracy. Visual evidence of this relationship is provocative. Figure 1 shows Freedom House democracy ratings for all the countries of the world. Darker shading indicate higher levels of democracy.
Figure 1-Democracy Rating
Figure 2 is a comparable world projection denoting prevalence of major worldwide email exchanging computer networks.
Figure 2-Interconnectivity Scores
The metric used in the second chart is termed "interconnectivity" [10]. Darker shading indicates greater level of interconnectivity. The similarity of patterns between the two maps inspires more rigorous analyses.
The two variables portrayed are also fundamental to the statistical analyses that follow. Fuller explanations of these measures follow. Generally, the concept of democracy is open to various interpretations. For this study, democracy is defined broadly as the combination of representative government and individual freedom. Numerical ratings are adapted from Freedom House's Comparative Survey of Freedom for 1993-1994 [11]. The survey produces two quantitative measures for each country: "Political Rights," the extent to which people freely participate in the selection of policy makers and in the formulation of policy and "Civil Liberties," the extent to which people are able to develop and express ideas independent of the state's. Since the correlation between these two measures is very high, the independent "Democracy" variable used here is the normalized average. 2 This use of Freedom House data follows an evolving academic practice for the evaluation of correlates to democracy [2], [4], [12]-[16]. There are difficulties inherent to quantifying a subjective multi-dimensional democratic quality across widely varying governments with a single scalar [17], [18]. Despite these problems, a practical consensus for relative rankings prevails quite broadly. Differing attempts to provide numerical ratings show substantial agreement. This conformity in the ordinal rankings suggests that, although the concept of democracy may be difficult to describe explicitly, it is well understood intuitively (at least by Western analysts). Alex Inkeles summarized a conference on the topic of measuring democracy with the following observation:
"[D]emocracy is a distinctive and highly coherent syndrome of characteristics such that anyone measuring only a few of the salient characteristics will classify nations in much the same way as will another analyst who also measured only a few qualities but uses a different set of characteristics, so long as both have selected their indicators from the same larger pool of valid measures. Far from being like the elephant confronting the blind sages, democracy is more like a ball of wax." [17].
Prevalence of information revolution technologies would seem to be more easily countable since tangible equipment is essential. Here too, however, quantification difficulties exist. Some are definitional. As the capabilities of communication technologies increasingly overlap, recalling Ithiel de Sola Pool's "convergence of modes," [19] the question of what is properly included becomes decidedly non-trivial. Computers can send faxes and radio waves and television cables can transmit email messages. Electronic mail is the specific capability of interest in this study because it enables people to discourse across the borders in ways that have never been possible since the evolution of the nation state. Internationally, email is also the most mature, most widespread, and most commonly used element of what John Quarterman calls the "Matrix."[20] Quantifying email is difficult. Of the numerous email networks, four are globally dominant: Internet, BITNET, UUCP and FidoNet. Record keeping has not been regular and accurate on all of these networks. The best available and most comprehensive data is for the numbers of nodes which therefore constitute the basic unit of measure for this variable.
Nodes themselves, however, are not all equivalent, even within the same network. A node may consist of a single computer and user or an entire organization with many of both. The Matrix Information Directory Service (MIDS) tracks and maintains historic data on the size of these networks aggregated by country [21]. The "Interconnectivity" metric used here is a combined measure of MIDS data on nodes per capita per country for each the four major computer systems that can exchange electronic mail. Within each network, countries are ranked and scored with a number from 0 to 4. The 0 is assigned to all countries which do not have any nodes of a particular network. The numbers 1 through 4 are assigned by quartile. The lowest quartile of countries with one or more nodes for a network receive a score of 1. The highest quartile countries receive a score of 4. The sum of the four scores determines the level of interconnectivity on a scale from 0 to 16.
The combined scores weight each of the four networks equally because the ability to exchange email is a relatively generic capability. Nevertheless, the equal weightings introduce some theoretic difficulties. Although each of the networks support email, they are not necessarily comparable in other respects. For instance, the Internet, with specialized services such as the World Wide Web and remote log on, has much more functional capacity than the others. The discrete interconnectivity variable, therefore is neither uniform nor strictly monotonic. It is arithmetically possible, for instance, that a country with a low interconnectivity score and Internet may actually have more communications capability than a country with a higher score but no Internet. In practice this is not likely to occur and in the analysis none of the potential degradation of this variable was observed. There are several reasons why this theoretic possibility was not a practical problem. First, email, not necessarily the other services, embodies specific characteristics that are hypothesized to have dynamic implications for democratization: the ability to conduct multi-directional discourse3 across borders in a timely and inexpensive manner, unbounded by geographic and institutional constraints. Second, practical aspects of evolutionary development in interconnectivity avoid the theoretical problems with monotonicity. Less capable systems are also less expensive and easier to implement, so initially they are more prevalent. Improvements on the micro-level tend to be cumulative via successive upgrades. These often incorporate switching to or adding on Internet capabilities. In this way, a general progression emerges in the enhancement of interconnectivity that this scale approximates. Furthermore, to the extent that interconnectivity as a predictor for democracy is measured imprecisely, the effect is reduced statistical significance of the predictor. Thus the conclusions would still be valid from this a fortiori analysis.
Figures 1 and 2 suggest a specific conjecture that univariate analyses support. A strong correlation between democracy and interconnectivity does, indeed, exist.
Figure 3-Regression and Scatterplot
The question of causality will be addressed in detail later but as the matrix attests, the correlation coefficient for interconnectivity is not only large, it is substantially larger than that of any other traditional predictors of democracy. Even the coefficient on economic development is smaller by 0.16.
Table 1. Correlation Matrix
Economic development, reported here as a per capita GDP (and abreviated simply as GDP), is quantified terms of purchasing power parity, as is traditional. Education is commonly paired with economic development as a predictor of democracy [1], [2], [15]. Direct causality is easy to imagine. An educated public is likely to be both more aware of political events and more capable of intervening to influence them. Indirectly, education conceivably enhances democracy by contributing to economic growth. The average number of years of schooling across the entire population is considered to be the best measure of education for analyses such as these [2].
Human development and health indicators are also often correlated with democracy. Most prevalent in the literature are infant mortality rates and life expectancies [13], [16]. causal argument could be posed that as citizens become more assured of their own well-being they have more incentive and wherewithal to demand civil rights and political liberties. In the opposite causal direction, an empowered public could be influential in government spending and could impel the provision of better basic services. A democratic government is also inclined to value more highly an individual human life. These two measures are highly correlated, but forward causality seems more plausible in terms of life expectancy. Individuals whose lives are extended would seem more likely to recognize their increased personal stake in democracy than babies who would not otherwise have survived infancy. United Nations Development Program sources provided all the economic, education and health data which are used in these analyses.
Cultural and ethnic factors may also have certain roles in democratization. "Homogeneous national entities may be more likely to evolve into peaceable democracies than states rent by harsh linguistic and cultural antagonisms" [25]. A variable for the percentage of the population which is comprised by the largest ethnic group in each nation is included to account for these effects of ethnic homogeneity. These data are published in the CIA World Fact Book [24]. In a few cases mostly in Northern Europe and Africa these data were not available. Where applicable, the percentage of largest religious affiliation substituted for the missing data.
In multi-variate analyses, cultural differences across countries are potentially more important than the internal mix. Debates continue as to whether certain cultures or civilizations are favorably disposed or fundamentally disinclined to embrace democratic principles [26], [27]. In either case, it is not difficult to believe that cultural aspects influence the characterization of the political regimes and the appreciation of personal liberties. Binary categorical variables on region are generated to account for these effects. Demarcation between cultures can never be exact. Inexorably, the classification of some countries into any of the regional categories is susceptible to quibbling. Six regional categories were defined that incorporated elements of geography, history and religion. These six labeled Africa, Asia, Eurasia, Latin America, Middle East, and Western Europe, map reasonably well onto the eight civilizations identified by Sam Huntington [26]. Western Europe also includes countries that are not on the continent but which have a dominant Western European heritage: United States, Canada, Australia and New Zealand. Israel is also included in the Western European category. The Middle East category is predominantly Muslim, includes the Islamic North African states and extends from and extends from Egypt to Pakistan. Africa is defined in fairly obvious geographic terms including South Africa, minus the northern states grouped into the Middle East. Asia includes the Confucian countries and the Pacific Islands, plus India and Japan (both non-Confucian), minus North Korea. Latin America stretches from Mexico through Argentina including all the Caribbean except Cuba. Cuba and North Korea, plus Albania and the splinter states of Yugoslavia, in addition to the members of the former Warsaw Pact countries and are all grouped in the Eurasian category. The effects of culture on role of communication in society may be as profound as that on democracy. Therefore, some of the regression models that follow include interaction terms which are the products of the regional dummy variables and the interconnectivity scores.
Population completes this list of independent variables. Presumably, the size of a country could influence the type and effectiveness of governance. Very small countries may be anomalous. Therefore, only countries whose populations exceed 1,000,000 (and for whom data is available) are included in this study.4 Above this threshold minimum, country populations span more than three orders of magnitude. Population, therefore, is best included here as an exogenous variable in a log form.
Like the adjacent maps presented earlier, the correlation matrix exhibits a surprisingly powerful correlation between interconnectivity and democracy. Multiple linear regressions provide further convincing evidence that one cannot dismiss this correlation as spurious. Regression results are shown below in Table 2. Models I and II show the resulting statistical output of ordinary least squares (OLS) regressions. Model I is an inclusive model which involves all six predictors.
Immediately apparent is that, again, interconnectivity emerges as the dominant predictor. More startling, perhaps, is that interconnectivity is the only statistically significant predictor over which policy may have any influence. The coefficient on population is also significant, but the size of a country's population, largely inaccessible to foreign intervention, offers scarcely few policy recommendations (except perhaps to shine a glimmer of hope on the fractious states of Yugoslavia and the former Soviet Union which potentially may have a more democratic future than their larger predecessors.)
None of the other predictors show a statistically significant correlation, including economic development. Thus, the very high statistical significance on interconnectivity is all the more striking. With greater than 99.9% certainty, one can reject the null hypothesis that there is no relationship between democracy and interconnectivity. Furthermore, the coefficient on interconnectivity is substantial in practical terms. A single point increase on the interconnectivity scale corresponds to an increase of 5% in the democracy ratings.
Model II contains a more parsimonious model retaining only those predictors which showed statistical significance or have a substantial theoretic causal basis (GDP). Not surprisingly, outputs are similar. The drop in adjusted-R2 is minimal (0.025) for having excluded three predictors, underlining the relative importance of those variables retained. These few variables explain more than 50% of the variation in democracy for 140 countries. Models III and IV, with the addition of the regional interaction terms, are analogous to I and II, respectively. These next two models show that the positive correlation of interconnectivity with democracy is consistent across and within regional boundaries. In all the regions the coefficient is positive. In half of the regions, the coefficient is substantial and statistically significant. The correlation is most pronounced in those regions undergoing dramatic political transformation. This fact is important when considering causality. If the correlation were positive only where democracy preceded the information revolution, one might be able to argue that the latter strengthened the former but certainly not that the latter caused the former. The evidence, however is that the relationship is weakest in the regions characterized by established democracies and strongest in regions which are cultivating nascent democracies. In Eurasia, the t-statistic indicates a level of significance at better than the 0.1% level. In Africa, the coefficient on the interaction term is the highest and the t-statistic corresponds to a 1% level of significance. The coefficient is also substantial for Latin America with a 10% significance level. The regression lines which accompany the six scatterplots in Figure 4 approximate these multi-variate regression results for visual comparison. Western Europe shows the most paltry correlation. In this region, the high interconnectivity levels do not vary much and the high democracy ratings move even less.
Figure 4-Regional Regressions
Table 2-Regression Models
It is tempting to infer causality from these impressive correlations and conclude that interconnectivity influences democratization. However, to do so might be premature. Causality could, in fact, flow in the opposite direction. Democracies rely on an informed public and uninhibited communication and may therefore seek interconnectivity. One way to test this possibility analytically is via a system of simultaneous equations to be resolved by two-stage least squares (2SLS) estimation. Both democracy and interconnectivity are endogenous in Model V. Instrumental variables for interconnectivity are determinable from the characteristics of email. Electronic mail is text based and travels over telephones lines. Appropriate instruments, therefore, are percent literacy and the number of telephone lines per capita. Exogenous variables in the democracy equation are, as before, related to economic growth and human development.
The resulting regression coefficients are also listed above on Table 2. Interconnectivity is shown to be an even better predictor of democracy than before. The magnitude of the coefficient for interconnectivity on democracy is greater than in the OLS models. The level of significance remains above the 0.1% level. Democracy, however, does not prove to have any significant effect on interconnectivity. Thus, the suggestion that democracy leads to interconnectivity is not supported and the hypothesis that there is no positive effect cannot be rejected. The coefficient on population is still significant and negative. In this model, GDP is also negatively correlated to democracy and statistically significant. This supports the notion of some scholars that democracy is not costless [28], [29]. All else being equal, such as interconnectivity and population, greater economic development might be available only at the expense of democratization. Proponents of Pinochet and Lee Kuan Yu models for development have made similar arguments.
The other alternative explanation for the strong correlation between interconnectivity and democracy is that a third variable may influence both simultaneously. The obvious candidate is economic development which many contend is an important prerequisite for democratization. The correlation between interconnectivity and GDP, at 0.84, is also very high, suggesting that the third variable hypothesis deserves further examination. In practical terms, the equipment necessary to communicate electronically is expensive, especially for citizens of the Third World regions which Western democratization policy would be most eager to influence. The same economic resources which can finance participation in the communications revolution could conceivably fuel demands for personal rights and freedoms. Again, a system of simultaneous equations can help unravel complex reciprocal effects. Model VI includes all three variables, GDP, democracy and interconnectivity as endogenous. Each is included in the forcing function for the other two. The interconnectivity equation utilizes the same two instrumental variables. The exogenous variables in the democracy equation are the same as before except that schooling, which has never been a statistically significant predictor for democracy in this set of models, switches over to serve as an instrument for economic growth. As noted earlier, scholars have surmised that education can influence democracy by increasing personal and national wealth. The 2SLS estimation results, shown in Table 2, are consistent with all those that preceded and do not support the hypothesis of economic development as the confounding third variable. Strongly to the contrary, the regression coefficients for interconnectivity on democracy and GDP are both substantial and statistically significant, again above the 0.1% level. Neither democracy nor GDP proves to strongly influence interconnectivity. GDP again shows a negative correlation with democracy but at a weaker significance level, 20%.
In each model presented here, without exception, interconnectivity positively correlates with democracy at the highest levels of significance. In each model, at lower but still high significance, the correlation with population on democracy is negative. Stories to explain both the country size and the interconnectivity phenomena may share a common plot. Smaller size and greater interconnectivity may similarly be conducive to democracy by facilitating coordinated civic action. Although it might by now be on its way towards becoming cliché, the often repeated analogy that information revolution technologies are shrinking the world offers appropriate insight. Interestingly, the most populous country which Freedom House labels as completely "free" became a democracy in 1776 when its population was only a fraction of its current size. At that time and at that size, available communication technologies, like pamphleteering, were sufficient to gel public support into popular action.
It is the globe, however, not just countries which are "shrinking" in the wash of information flows. The worldwide expansion of democracy may have less to do with the way in which these technologies favor democratic processes domestically than the way they spread democratic ideals internationally. In an inward direction from the perspective of the citizens of a prospective democracy, information revolution technologies enable them to learn more about how other societies operate. If they discover that others living elsewhere are living better thanks to the democratic governance "over there," they are likely to demand some of that democratization for themselves. For this reason the Soviets jammed broadcasts from the West. Several of my Russian associates have expressed that they felt fortunate to have been born in a communist state until learned what life was really like in America. In an outward direction, information revolution technologies empower citizens anywhere to broadcast to the world infractions against their "inalienable rights" by their own government. Thus world pressure can be brought to bear against repressive regimes which can no longer hide their misdeeds as successfully as before. That demonstrators in Tian An Men Square displayed signs written in English was not a coincidence. Cross-border communication in the defense of democracy and human rights is the activity on which citizen diplomacy groups like Amnesty International stake their success. The new technologies enhance these capabilities.
Governments that try to squelch the new information technologies in order to protect their monopoly on power, do so essentially at the peril of economic growth. This is the inference from Model VI and is precisely what leading analysts have been predicting, "For nations to be economically competitive, they must allow individual citizens access to information networks and computer technology. In doing so, they cede significant control over economic, cultural, and eventually political events in their countries" [30].
Despite the inherent limitations of statistical analysis, every different analytic perspective of this study coherently and repeatedly emphasizes two important observations. First, interconnectivity consistently emerges as a powerful predictor of democracy. In univariate analysis, the correlation coefficient for interconnectivity on democracy is larger than that of any other variable. As a variable in an ordinary least squares multiple linear regression, interconnectivity is the dominant predictor in all models. As an interaction term in conjunction with regional categorical variables, the correlation of interconnectivity with democracy is everywhere positive, and has both the largest substantive value and greatest statistical significance in regions characterized by dynamic political transformations. As an endogenous variable in systems of simultaneous equations, interconnectivity always proves to be a significant predictor of democracy and economic development, but never was the reverse true. These analytic results overlay a background of mounting anecdotal evidence that new information and communication technologies are facilitating democratic change worldwide.
Second, none of the traditional variables which measure and guide foreign assistance policies with respect to promoting democracy and peace seem to produce the desired effect. The only statistically significant policy variable that appears in any of the models is economic development and its regression coefficient on democracy in those models is negative.
The minimum policy implication, when considering these conclusions together, is that the effects of revolutionary information and communication technologies on the objectives of foreign aid and national security must be better understood, and the search for appropriate policy instruments is a critical aspect of further study. At a maximum, the priority of policies regarding international communication should be at least as high as the priority for foreign economic development and perhaps as high as that of some national security programs. References
Christopher R. Kedzie has earned degrees in Aeronautical Engineering at the U.S Air Force Academy (B.S. 1982) and Massachusetts Institute of Technology, 1984 (S.M. 1984), and in Public Policy at Harvard University (M.P.P. 1992). He is currently a doctoral fellow at the RAND Graduate School where his primary research efforts focus on the effects of information revolution technologies in the arena of international affairs. Before coming to RAND, he was a founder and director of organizations in both Ukraine and Uzbekistan which exploited information revolution technologies to support economic and political reform in the republics of the former Soviet Union.
Chris Kedzie
can be contacted at:
2101 Ocean Avenue, #3
Santa Monica, CA 90405
310-393-0411 x6885
kedzie@rand.org