Measuring Effective Democracy Draft November, 2007 Abstract: There is no consensus in the social sciences on how to best measure democracy. Inglehart and Welzel (2005) argue that they have constructed an improved measure which they call “effective democracy” that capture not only the form of democracy but also actual democratic practises in a country. This measure has been criticized by Hadenius and Teorell (2006) on several grounds. The debate on the usefulness of the measure is still ongoing, and this article contributes to this debate. The article is sympathetic to Inglehart and Welzel’s underlying idea of capturing “substantial” rather than “formal” democracy, but is critical to the specific measure of “effective democracy” proposed by these authors. The measure has several unfortunate theoretical and distributional properties, and the empirical scores generated by the measure are often highly unintuitive. The concrete criticisms of the “effective democracy”measure proposed here, related to both the validity and reliability of the measure, should help end the ongoing debate and initiate a new search for better measures of substantial democracy. 1 1.0 Introduction. Democracy and its causes and consequences are some of the most frequently studied subjects in the social sciences. There exists no consensus however on how to define democracy, let alone on how to measure it in practice. Over the last couple of years, a debate has been going on in “Studies in Comparative International Development” between Ronald Inglehart and Chris Welzel on the one hand and Axel Hadenius and Jan Teorell on the other, on the effects of “mass attitudes” on democratization. Linked to that substantial debate is a methodological debate on how to operationalize and measure democracy. Especially, there has been a disagreement on whether Inglehart and Welzel’s measure of “effective democracy” is a valid or even meaningful measure. “Effective democracy” is constructed by multiplying the Freedom House Index with a measure of corruption, either the one provided by Transparency International (CPI) or the World Banks World Governance Indicators. According to Welzel (2007: 421) “the dispute about the usefulness of measuring effective democracy” is still unresolved. This article does not contribute directly to the substantial debate on mass attitudes and democratization, but rather looks more narrowly at the methodological debate on the meaningfulness, reliability and validity of the “effective democracy”-measure. In order to discuss how best to operationalize a concept, one needs to clarify the concept first. I argue that whether “effective democracy” is a meaningful operationalization of democracy depends on the conceptual definition chosen. Whether the operationalization then is reliable or valid are different questions. 2.0 The conceptual debate: What is democracy? Institutionally based democracy definitions Joseph Schumpeter’s critique from 1942 on what he called the “Classical Doctrine of Democracy” (Schumpeter 1976: 250-268), and his preferred alternative theory of democracy (Schumpeter 1976: 269-283) are good starting points for analyzing the concept of democracy. According to Schumpeter “[T]he eighteenth century philosophy of democracy may be couched in the following definition: the democratic method is that institutional arrangement for arriving at decisions which realizes the common good by making the people itself decide issues through the election of individuals who are to assemble in order to carry out its will” (1976: 250). Schumpeter’s critique of this democracy concept has proved to be very influential. His main point was that defining democracy in terms of institutions’ ability to 2 secure the implementation of a “general will” or a “common good” is unsatisfactory, because these constructs lack a meaningful reference. Schumpeter pointed to the relatively plausible fact that for “different individuals and groups, the common good is bound to mean different things” (1976: 251). According to Schumpeter, people in a country diverged both in their conception of means and ends, and a “common good” therefore lacked a precise reference. Some years later, theoretical arguments, presented by Kenneth Arrow (1951), sought to show the impossibility of aggregating individual (fixed and well defined) preferences to a determinate, well-defined collective preference when the number of individual and the number of issue-dimensions were allowed to increase sufficiently. To many, public choice theorists and others, one important implication of this theoretical proof was that democracy viewed as an implementation of a “general will” was nonsensical, due to the emptiness of the concept. Schumpeter’s famous alternative was defining democracy, or more precisely the democratic method, as the “institutional arrangement for arriving at political decisions in which individuals acquire the power to decide by means of a competitive struggle for the people’s vote” (1976: 269). In other words, Schumpeter leaped over the issue of popular will by referring directly to an institutional mechanism, namely competitive elections. He was thereby implicitly solving the problems of fuzziness, indeterminacy and even emptiness of the “Classical” definition-type. The heritage of Schumpeter is still looming. Adam Przeworski has been a vocal defendant of a “minimalist” democracy definition, defining democracy as a political regime that is holding contested elections. According to Przeworski et al. ““democracy” has become an altar on which everyone hangs his or her favourite ex voto. Almost all normatively desirable aspects of political life, and sometimes even of social and economic life, are credited as features of democracy” (2000: 14). The benefits of a minimalist, institutionally based definition, are analytic stringency, precision and clarity. Przeworski et al. (2000) even provide relatively clear and reliable algorithms for classifying whether an empirical system can claim the title “democratic”, or whether it should be categorized as a dictatorship. There can be formulated at least two powerful criticisms of the “Schumpeter-Przeworski” view of democracy. The first criticism takes the “institutional starting point” as given, but argues that limiting the definition to elections is too narrow. This criticism can be related to what Terry Lynn Karl termed the “fallacy of electoralism” (Diamond 1999: 9). The main line 3 of thought is that elections are by far sufficient in securing a democratic political regime. One needs additional institutional guarantees, and therefore the democracy definition needs to be broadened. One paradigmatic list of such “institutional guarantees” can be found in Robert Dahl’s “Polyarchy” from 1971. The list includes freedom to form and join organizations, freedom of expression, right to vote, eligibility for public office, right of political leaders to compete for support, alternative sources of information, free and fair elections and particular institutions for making government policies depend on votes and other expressions of preference, like elected parliament and (directly or indirectly) elected executive government (Dahl 1971: 3). Substantial definitions of democracy The second criticism of the electorally based definition disagrees with the underlying premise of defining democracy in terms of institutions, and, despite the strong criticisms presented above from Schumpeter and Arrow, takes the role of the populace as point of departure. If one claims that democracy merely consists of a matrix of certain rights, liberties and institutions, one misses the mark, according to Beetham (1999). The question can still be raised about why particular institutions and rights are to be considered democratic. Any answer risks ending in a tautologous argument. According to Beetham “[T]he only way to avoid circularity is by specifying the underlying principles which these institutions embody or help to realize, and in terms of which they can plausibly be characterized as democratic” (Beetham 1999: 90). In modern day political, academic and daily life, the usage of the term democracy often refer to the role of the population in political life, and this is in my view also the most fruitful starting point when thinking and theorizing about democracy. Even authors using institutional definitions are often at least implicitly interested in how the role of the population in political processes are affected or are affecting certain other variables. They are for example interested in “contested elections” because they give broader population segments influence over political alternatives, and further policy formulation. David Beetham exclaims relatively straightforward that “[T]he core idea of democracy is that of popular rule or popular control over collective decision making” (Beetham 1999: 90), and Beetham adds that the degree of political equality among citizens is an important criteria backing up this general definition. Institutions are to be considered democratic only if they contribute to realizing this underlying idea. One can still be interested in particular institutions, their existence and also their functioning, when studying democracy, but these are only instruments that underpin democracy. Take notice that this definition does not refer to a “general will” or to aggregated 4 preferences of the collective, but only to the populace’s opportunities for actual influence in public decision making, and further it implies something about the distribution of these opportunities for influence among the citizens. By not referring to a “general will” the definition therefore evades some of the direct criticisms of the “Classical Doctrine of Democracy” discussed above, but the “vagueness” and “indeterminacy” criticisms can nevertheless to a certain extent still be applied to a Beethamian democracy concept: What does actually “popular control” mean, and who is the relevant demos? Economists often stress that there exists no free lunch. No one definition of democracy is optimal on all the normative dimensions scientists cherish. As I see it, when it comes to choosing between an institutional and substantial democracy definition, there is a trade off between precision and stringency on one hand, and the want to validly capture the nature of an underlying concept on the other. Defining democracy institutionally will have the first advantages, evading problems of vagueness and indeterminacy, whereas the “substantial democracy definition” actually takes the intuitive and common notion of democracy as something related to people’s power over politics seriously. My proposal is that if you want to study for example the effects of particular institutions, there is an opportunity for doing so in a relatively more stringent fashion, but one is then not studying the “effects of democracy”, but rather the effects of for example “elections” or “freedom of speech”. “Popular control over politics” is a different concept analytically from “contested elections”, even though they are empirically connected. Even though we can not make precise scales for “democracy” in the Beethamian sense that are perfectly valid and reliable this does not imply we should refrain from trying measure it. “Claiming that a concept is hard to measure (epistemologically), does not make the concept disappear (ontologically)” (XXXX 2006: 25). The implications of the conceptual debate for the “effective democracy”-measure. Note that the position taken in the debate on the right conceptual definition of democracy in many ways preconditions the discussion that is to follow on the “effective democracy”measure. If one takes an institutionalist definition of democracy seriously, it is hard to make extra claims on what constitutes “effective democracy” within the limits of the definition. Effectiveness can of course be defined as an independent concept, and one could measure whether democratic institutions were empirically effective in bringing for example economic growth, equity or improve schooling. The mere existence, or possibly the degree of existence, of one or more institutions is the sole criteria looked at when defining democracy. However, if 5 one uses the definition of democracy as “popular control over collective decision making”, the question of defining “effective democracy”, not only becomes a sensible question, but also a very vital one. A first approximation to operationalizing “popular control over politics” can be made by looking at the empirical existence of certain formal institutions like elections, constitutional guarantees of freedom of speech and association, which have historically proved to be efficient in fulfilling this criterion to a larger degree than other institutional arrangements. However, measuring “effective democracy”, can now plausibly be interpreted as a way of incorporating also how these institutions in a particular country actually functions in translating the voice and demands of broader segments of the population into legislation and implemented policies. Viewed in this light, “effective democracy” is not a combination of two analytical concepts, but it is rather a better empirical approximation to the theoretical definition, than the original “formal” measure. As Robert Dahl (1971) suggested, democracy should be viewed as an ideal concept. Egalitarian distribution of power over collective decision making between all citizens provides the proper measure stick. Empirical regimes will to a larger or lesser degree fulfil the ideal. Ceteris paribus, to have broader parts of the population involved in politics makes a system more democratic. Ceteris paribus, a more equal influence between the participants in legislation and policy formulation also makes it more democratic i . The possibility of equal political influence hinges not only upon the existence of formal institutions, but also on how these institutions actually function in practice when transforming demands and other “inputs” into political outcomes. “Effective democracy” therefore becomes somewhat of a misnomer, since it is actually here intended as a more comprehensive and better measure of democracy. As several authors have recognized, the need to take the actual functioning and not only the existence of formal rights and institutions seriously has become even more urgent in the post Cold War world. Przeworski et al. partly legitimizes their usage of their minimalist theoretical definition and further operationalization by the fact that “[D]ifferent views of democracy, including those that entail highly subjective judgements, yield a robust classification” (2000: 57). They for example find that their measure correlate 93,7% with FHI. However their analyses end in 1990, and as Diamond (1999: 286) notes, after the end of the cold war with the change in international political and ideological context, there is reason to believe that there might be larger divergence between the formal properties of institutions and the actual democratic character of political regimes. As Welzel and Inglehart notes “[E]lectoral democracy can easily be abused to hide severe deficiencies in the actual practice of civil and 6 political liberties” (2005: 175), and this is probably an option for several regimes in an era when everything from EU-membership to World Bank loans, as well as generally being recognized as a “civilized” nation, requires democratic institutions and practises. Summing up this general empirical trait Inglehart and Welzel notes that the political development of the new democracies of the “third wave” have been more disappointing than initial optimism among some suggested, since “many of the new democracies show severe deficiencies in their actual practice of civil and political liberties” (2005: 149). Elections is no longer (if it ever were) a sufficient proxy for what many view as “true” democracy. The construction of a broad democracy indicator is therefore of utmost importance. 3.0 Operationalizing democracy The Freedom House Index The Freedom House Index (FHI) is one of the most utilized democracy indicators in contemporary academic research. Larry Diamond calls it the “best available empirical indicator of liberal democracy” (1999: 12). The FHI scores countries after a scale ranging from 1 to 7 (where 1 is most democratic) on two dimensions: political rights and civil liberties. Political rights “enable people to participate freely in the political process, including through the right to vote and stand for public office, and elect representatives who have a decisive impact on policies and are accountable to the electorate. Civil liberties allow for the freedom of expression and belief, associational and organizational rights, rule of law, and personal autonomy without interference from the state” (Freedom House 2005: 775). Both categories are indexes constructed from large subsets of indicators, which are formulated as “check questions”. Freedom House scores its indexes (primarily) on the basis of a list of 25 such questions, which each has several sub-questions. Ten of the 25 questions relate to political rights, and fifteen to civil liberties. The FHI is to a large degree based on the presence or absence of different institutions, but it also seeks to account for the actual functioning of these institutions. Freedom House itself claims to measure “the real-world rights and freedoms enjoyed by individuals” (Freedom House 2005a: 1) Therefore this operationalization is not only valid for institutional definitions of democracy, but, I would claim, also to a certain degree relatively valid for more substantial democracy definitions. However, the presence or absence of certain institutions remains core even to the Freedom House Index; dominating the check list of questions on especially political rights. 7 “Effective democracy” If one is using a substantial definition of democracy, one needs to look beyond the existence formal institutions and constitutional rights, when measuring. One important quest is therefore to look for systematic components affecting democracy other than the formal institutions and rights. Power dispersion and the functioning of institutions in transforming popular demands to collective decisions are in their very nature hard to observe and evaluate. Inglehart and Welzel (2005) make a strong case for one such component affecting the actual functioning of institutions: They highlight the absence of corruption and the practice of rule of law more generally. The idea is that in order to make them effective “civil and political rights also require honest, law-abiding elites. Corruption, broadly understood, reflects nepotism, favouritism, and other illegal mechanisms used by elites to circumvent the rule of law and use their power for their own benefit, depriving ordinary people of their legal rights” (Inglehart and Welzel” (2005: 192). By this Inglehart and Welzel drive a clear wedge between formal democracy and what they call “effective democracy”. In order to make democratic rights and institutions effective in practice, there is a clear requirement related to lack of elite corruption: “In short, effective democracy measures not only the extent to which a society has liberties on paper but the extent to which these liberties are actually practised by the state and its officials. This variable spans a continuum ranging from little or no real democracy to fully effective democracy” (Inglehart and Welzel 2005: 194). The authors operationalize effective democracy by multiplying an aggregated index based on Freedom House’s political and civil liberties dimensions with a measure on corruption, supposed to tap honest and law abiding elite behaviour. The authors have used different indexes in different articles and books, with the “Corruption Perception Index” from Transparency International used in for example Welzel et al. (2003) and the “Control of Corruption” score from the World Bank in the study from 2005. After normalizing the corruption scores to a measure ranging from 0 to 1 (X) and FHI from 0 to 100 (Y), “effective democracy” is given as X*Y. X*Y is then a proxy for “how much freedom people actually have rather than how much freedom they have on paper” (Inglehart and Welzel 2005: 196). 4.0 The benefits of the effective democracy measure Inglehart and Welzel have described and also defended their “effective democracy” measure against criticism in different articles and books (Inglehart and Welzel 2005, and Welzel and Inglehart 2006). These authors give clear presentations and well-argued discussions on the 8 benefits of this measure. I will therefore not go as deeply into its benefits as the measure maybe deserves. I will however go through a couple of positive characteristics of the measure. The “effective democracy” measure is explicitly constructed in order to tap also the “actual functioning” of institutions and individual rights and not only their form. In addition to the measure’s relatively parsimonious nature, constructed as a product of two separate indexes, this is the main benefit of this operationalization. Functioning is specified as whether the institutions and society in general is plagued by corruption. There is a clear cut theoretical argument behind this reasoning. Corruption means the abuse of public positions for personal ends. This again implies the lack of rule of law. A further consequence is then that the ordinary citizen’s formally guaranteed democratic rights and opportunities can not be practiced in the way they are constitutionally described to be. In a case where the constitution is guaranteeing elections and freedom of speech, but the country and polity is plagued by rampant corruption, democracy is said to be “formal” and not “effective”, and hence the country will score relatively low on the measure. As I mentioned earlier, this is especially important in a time period where observers and researchers are claiming that the “deficient democracy”, the “semi-democracy” or the “electoral authoritarian regime” is becoming a very common type of political regime. Equating free elections and freedom of speech with popular rule over collective decision making in its strictest and most egalitarian sense is of the mark empirically, also in OECD-countries. To name a few: Italy is often being accused of clientilism (in the south) combined with a not fully independent media. French and Japanese politics are often labelled as elite driven and oligarchic. American politics is perceived to be heavily influenced by organized business interests and other narrow lobby-groups. However, the most obvious discrepancies between “formal” and “effective” democracy are to be found in the developing world; in Africa, Asia and Latin America. Cases are numerous, as for example some of the six comparative studies in Diamond and Morlinho (2005) exemplify. Clientilism and personalistic politics is a particular political trait that seems to devoid formal democratic institutions of what many see as its normative purposes. Jean Francois Medard (1996) claim that African politics is laden with structures referred to as “neo-patrimonial”. Formal state institutions exist, but these are not the arenas where the actual political action is. Politics is rather managed through vertical personal-based ties, in a “patron-client” fashion, where personalization of politics is an inevitable result. Chabal and Daloz (1999) in their analysis of African politics claim that these kinds of structures and practices make democratic institutions function badly. Another example is Philippine politics. Politics in the country is according to Sidel (1998) characterized by “bossism”, and abuse of electoral practises is 9 widespread. Especially landowners have “used” elections in order to maintain power by means not always recognized as true to democratic ideals. Generally, when elites manipulate democratic practices, neglect the rule of law, and undermine ordinary citizens rights by corrupt practices, formal democratic institutions and rights are rendered ineffective. There is another benefit to the “effective democracy index” when compared to for example the FHI. The FHI, and also other indicators like Polity, have been accused of differentiating too little between regime types, and this goes especially for the most democratic regimes. It is a problem for the FHI indicator that several empirical cases are assigned the highest score, in practice equating the regimes in these countries with perfect democracy. The most problematic issue for any analysis on causes and effects of democracy is that some of these countries have a larger degree of “popular control over collective decision making” than others, but all of them pass the ceiling-criteria for being given a score of 1 on political rights and civil liberties. Not to say that Freedom House is completely insensitive to distinctions among what is generally viewed as the world’s more democratic regimes. The four large Western European countries for example all received a 2-score in civil liberties at some point in time during the last couple of decades. France was denied the top score among others because of existence of political corruption, and the UK because of Thatcher’s union bashing, limiting the freedom of assembly. “Effective democracy” as an interaction term produces more graded measures, and also diversifies more at the top end of the scale. The degree of corruption here provides the factor that creates this divergence. Inglehart and Welzel explicitly recognize this as one important benefit of the indicator when compared to Freedom House: “If we took the formal democracy measure at face value, we would conclude that Latvia and Slovakia are just as democratic as Britain and Germany. But in reality they have relatively corrupt elites who devalue the constitutional rights their people theoretically possess” (2005: 195). 4.2 Validity and reliability problems of the measure The effective democracy measure as constructed and used by Inglehart and Welzel has been heavily criticized by Axel Hadenius and Jan Teorell (2005) on several accounts. One claim is that Inglehart and Welzel are wrong in multiplying two different concepts into one measure, claiming that “[T]he mistake, thus, is to confuse basic and quality criteria” (2005: 90). Welzel and Inglehart (2006) answer the criticism by pointing to the fact that both a lack of formal 10 institutions and deficiencies in the actual practising of formally guaranteed rights and liberties can deprive people of their actual democratic rights and liberties. I would go even further than Inglehart and Welzel, who claim that the combination of “two different things”, democracy and elite integrity, make a very sensible measure. Both formal institutions and corruption are relating to the very same underlying concept, namely democracy defined as “popular control over collective decision making”. Therefore, I am in principle sympathetic to the approach by Inglehart and Welzel, which is trying to incorporate a quality dimension when measuring “democracy” properly conceptualized. However, Hadenius and Teorell produce other criticisms that are indeed very relevant to the validity of the effective democracy measure. I will reintroduce some of these criticisms, and elaborate upon them. I will also present some novel criticisms which cast grave doubts on the reliability and validity of the measure. 1) FHI’s nature as a partially “quality-oriented” index One of the most controversial claims of Inglehart and Welzel in different articles and books is that Freedom House with their index only measure the existence of formally guaranteed political rights and civil liberties. This assumption stands in stark contrast to the explicit intention by Freedom House to measure “the real-world rights and freedoms enjoyed by individuals” (2005: 1). The index is constructed in such a way that it is not bound to be restricted by the formal institutions and constitutional properties of a country. Among others referring to the FHI, Przeworski et al. claim that it entails “highly subjective judgments” (2000: 57). Hadenius and Teorell criticize Inglehart and Welzel for assuming that the FHI only limits itself to formal institutions and rights, since it also in their view incorporates “actual upholding of such rights” (2005: 90). Welzel and Inglehart however agree that Freedom House is not only looking at the legalization of democratic liberties but that the FHI also incorporates whether these rights and liberties are free from obvious violations. Nevertheless they still claim that the FHI is insensitive to more subtle ways actual political processes break with and circumvent formal rights and liberties, outside obvious open violations (2006). I want to list some of the check questions (Freedom House 2005: 780-782) here in order to validate my claim that Freedom House takes quality aspects into consideration when rating. In my view, the FHI index is actually almost as open for attack from the “opposite front”, namely that it contains elements that should be separated analytically from democracy, and rather be rated as causes or effects (XXXX 2006). For political rights, check question B3 11 reads: “Are the people's political choices free from domination by the military, foreign powers, totalitarian parties, religious hierarchies, economic oligarchies, or any other powerful group?” Question C2 should be considered an even more obvious “qualitative component”, especially for the proponents of the effective democracy measure: “Is the government free from pervasive corruption?”One of the further specifications on this check question is actually: “What was the latest Transparency International Corruption Perceptions Index score for this country?” C3 also provides a qualitative component asking: “Is the government accountable to the electorate between elections, and does it operate with openness and transparency?” This goes to the core of actual practicing of people power, beyond institutional formalities, asking whether ordinary people can directly influence politics and use their formal rights. For civil liberties, D1 asks: “Are there free and independent media and other forms of cultural expression?” The under-questions show pretty clearly that the measure is intended not to capture only constitutionally guaranteed rights to free speech: “Is self-censorship among journalists common, especially when reporting on politically sensitive issues, including corruption or the activities of senior officials?” Check question F5 sounds: “Does the rule of law prevail in civil and criminal matters? Are police under direct civilian control?” There are many other examples as well. Believing these “qualitative” components do not affect the overall grading of the FHI-measure obviously seems unwarranted. If one still doubts, the country studies show that several countries score much lower than their formal institutions and rights would suggest. Jamaica for example was scored a 2 in political rights and 3 in civil liberties (2,5 average) through the mid-nineties, and the main reasons given were the violence plaguing the country as well as corruption rendering formal rights less effective in practice. India was actually averaging 4,0 in 1993 through 1995, showing once again that countries are subtracted for dismal functioning institutions and rights. 2) Additive and multiplicative measures and their implications for ranking countries One of the concerns of Hadenius and Teorell is that certain quality criteria can compensate for the lack of basic democratic criteria. Their fear is that for example relatively incorrupt authoritarian regimes will get higher scores on the “effective democracy” indicator than corrupt democratic regimes, thereby reversing the “correct” ranking between these countries (2005: 89). Welzel and Inglehart’s (2006) reply is that this could have been the case if one was using a additive measure, adding the components of formal democracy and “quality” measured as corruption. However, Inglehart and Welzel point to the interactive nature of their 12 measure. They point out that “high scores in effectiveness cannot compensate for low scores in democracy”, since effectiveness is only a weighting factor of a given level of democracy. The mathematical intuition is simple. Zero multiplied with something is still zero, and a small number multiplied with one is still the small number. However, even if we take the “interactive weighting argument” at face value, there arises a methodological problem due to the way the FHI is constructed. The implications is that the effective democracy measure de facto functions partly like an additive measure, which Inglehart and Welzel claimed is inappropriate. As we saw above, it seemed obvious at least a decent number of the FHI’s check questions are output or “quality” oriented. Now denote FHI as A and the CPI as B. These are intended to respectively capture “formal institutions and rights”, X, and “quality or actual functioning”, Y. In the ideal case where A perfectly reflects X and B reflects Y, the “effective democracy” measure A*B perfectly captures X*Y. When the degree of formally guaranteed rights and democratic institutions becomes “very small”: X 0 A 0 A*B0 This is precisely Inglehart and Welzel’s argument formalized. Let us now say that an arbitrary 1/4 of the FHI points are deducted from “quality”-considerations, which I do not find to be an overestimation. This implies that A = (3/4X + 1/4Y). Take the case of a regime lacking formal institutions and rights, but where there exists rule of law and no corruption, no political violence, etc.; that is X0 and Y1: X0 & Y1 A ¼ & B1 A*B ¼. In this case, the effective democracy measure resembles an additive measure, since a regime lacking formally guaranteed rights and institutions, but doing well on economic performance and keeping firmly to its constructed, authoritarian laws, is given a score of ¼. This is a higher score than the 0 received by the totally corrupt formal democracy. More generally, if z is the share of “quality-based” points in FHI, and z >0, a regime lacking all traits of “formal democracy” will be assigned a value on the effective democracy score of (z*Y)*Y = z*Y 2. This is not equal to zero unless Y is zero. Therefore Inglehart and Welzel’s argument that due to the interaction-characteristic of their measure they avoid the criticism from Hadenius and Teorell (2005) is invalid. 13 3) Corruption as a narrow, and sometimes misleading, measure of democratic quality Inglehart and Welzel are relatively clear on the point that corruption stands as a proxy for the wider phenomenon of lack of rule of law. There can be raised two questions: First of all, is corruption a valid and reliable measure for the rule of law? Second, are there not other ways than lack of rule of law that can render formally guaranteed rights ineffective? The answers to both these questions in my view point to the insufficiency of using corruption scores as a measure of “quality” or “effectiveness” of political regimes. It is important to notice as Hadenius and Teorell (2005: 90) that CPI is actually mainly based on perceptions of among others businessmen and country specialists, rather than ordinary citizens. A country keeping its trade with large scale international businesses relatively free from corruption would therefore gain a relatively fair CPI-score, even if it was corrupt in dealing with its citizens. The World Bank measure later used by Welzel and Inglehart (2005), trying to capture political corruption therefore represents an improvement when it comes to validity. As history shows however, there are several ways through which formally guaranteed rules can be rendered inefficient. Political leaders with lust for continued power have creatively used a variety of means in order to restrict popular participation in the political process. The mechanisms can be as subtle as implicit threats of sanctions (loss of position or benefits), that make formal freedom of speech useless. Violence and terror are other ways of restricting actual exercise of formal rights. Media manipulation is a third, and in general lack of responsiveness makes the chain of representation most modern democracies pivot around dysfunctional. Corruption is thus only one way to render formal rights for the populace ineffective. In the anthology edited by Diamond and Morlino on the quality of democracy, the authors recognize at least “eight dimensions on which democracies vary in quality” (Diamond and Morlino 2005: xii). These are listed up as: rule of law, participation, competition, vertical accountability, horizontal accountability, respect for political and civil freedoms, political equality and responsiveness. Why then only single out only corruption, which again is intended to stand as a proxy for the rule of law? The question then becomes why one should weigh corruption so heavily, thereby influencing the effective democracy score to an extremely large extent? Corruption might not for example be a good proxy for democratic quality in richer and medium income countries, since threats to democratic quality might be of a different nature in these countries. The case of Singapore is an extreme example. There are elections taking place in Singapore, and it is relatively free of corruption. However, there are several other channels through which the rights Singaporeans have are rendered almost 14 useless. Manipulation of media and voting districts, as well as jailing of opposition leaders on cumbersome charges are among them. Richer and medium income countries with doubtful quality of democracy might therefore not be punished by a measure focusing on corruption, whereas poorer countries always tend to get their democracy scores reduced, because of the strong correlation between low income and high corruption. If corruption functions as an unbiased estimator for “quality” this would be a problem only because it generates a lot of uncertainty in the estimates of “effective democracy”. However, there is strong reason to believe that corruption is systematically linked to other variables. First of all, corruption is because of several reasons highly correlated with income level. The Weberian insight that poorly paid officials might tend to expand their earnings by illegal means is one obvious mechanism. That corruption in the economic sphere creates badly functioning economies, independent of surrounding political institutions is also clear to most development economists. Corruption in the economic sphere therefore is both a cause and an effect of a low income economy. Giving perceived corruption a large influence on a “democratic quality” measure is therefore bound to secure the measure a very large correlation with income and therefore also other variables correlated with income such as education or “emancipative value-indexes”. The correlation coefficient between income measured as GDP per capita in 2000 (PPP) and corruption as measured by CPI was 0,87 in 1999. This is perhaps the most troublesome aspect related to using Inglehart and Welzel’s “effective democracy” measure. The fact that both formal democracy and corruption are strongly correlated with income and other economic variables, tend to make the correlation between income and effective democracy an extremely strong one. All causal analysis relating effective democracy to income, or variables that are highly correlated to income, would therefore tend to produce stronger results than are actually warranted, even if there are strong theoretical and empirical reasons to believe that actual functioning of democratic rights is correlated stronger with income than formal democratic institutions. Table 4.1: Empirical correlations between FHI, CPI and effective democracy in 1999, with GDP per capita in 2000 GDP 2000 FHI 1999 CPI 1999 Effective democracy 0,62 0,87 0,87 15 4) The subjectivity-factor: Validity and reliability issues Suppose for a minute that FHI, X, measures the real degree of formal democracy, A, plus a random measurement error, ε. CPI, Y, measures “functioning of institutions”, B, and an error term, γ. This gives: X*Y = (A+ ε)*(B+ γ) = A*B+A* γ +B* ε + ε* γ. Finding estimates for the variance of a product of variables is obviously not a straightforward task (Goodman 1962), even if the variables are independently distributed and measures are unbiased. The multiplicative formulation of the “effective democracy” measure has the implication that the size of the total variance, and also error variance is relatively large when compared to any of its componentsii. Moreover, the assumptions of independence and unbiasedness are probably not holding when it comes to FHI and CPI. The indexes are strongly positively correlated. There are also reasons to believe that biases in one estimate on one index increases the probability of having a bias in the same direction on the other index, alas the error terms are correlated as well. One obvious point is that the CPI actually is an integrated part of a check question of FHI, and that FHI is one of the surveys the CPI uses when giving its estimates. We also have to remember that these two indexes are constructed as relatively subjective measures. As Bollen points out (1999), judges can be biased in their scoring of such indexes, and they might be systematically influenced by certain other variables. In this case, not only the indexes but also their error terms are not independent of each other, which could lead to further problems. Economic variables, like GDP-growth, might influence judgements systematically. Might Singapore and Botswana for example be judged as more democratic and less corrupt than they actually are, because of their shining economic performances? Judges are not objective and totally unaffected by wider concerns than those they are supposed to judge. If the error terms of the CPI and FHI are systematically correlated, this will exaggerate the size of the error term given by ε*γ. The new error term will probably not be normally distributed, and the reliability of the measure will suffer. There is also validity problem related to the rating of corruption and freedom of press and free flow of information. If the perception of corruption is influenced, even for businessmen and country experts, by what they are able to access through the media, then democracies might actually be rated systematically as more corrupt than actually warranted when compared to authoritarian regimes. A free and functioning media, with allowed flow of information might contribute to highlighting the need to combat the scourge of corruption in a given country, and their revelation of large scale scandals might actually drive CPI numbers higher than if these scandals were not to be discovered. Belgium, France and Italy might suffer relative to 16 Singapore and China because of their democratic qualities. This effect is detrimental for the validity of CPI as a democratic quality component. Couched in the terms of the equation above: The expected value of A* γ is not equal to zero. Even if E(ε) and E(γ) are both equal to zero, E(X*Y) would not be E(A*B). 5) Ordinality, cardinality and multiplication A variable is said to have an ordinal measurement level, if the numbers assigned to its values only tells something about the relative ordering of the values, and nothing about the absolute distance between the values. A cardinal measure on the other hand contains information about the absolute distance between values. Therefore in the ordinal case, if two units, P and Q, are assigned respectively 1 and 2 on a variable X, this only implies that Q has “more” of variable X than P. If X has a cardinal measurement level, Q has twice as much of X as P. The FHI and CPI are only variables of ordinal measurement level. Even if a country has half the corruption score or democracy score of another, this does not imply the country is only half as corrupt or half as democratic in a strict metaphysical sense. There is no obvious yardstick that can actually decide what twice as democratic actually means. Normalizing measures does not totally take away the problem either, even if it seems to relieve some contingencies. One mathematical trait of cardinal measures is that they can be multiplied, and that this multiplied measure still retains a meaningful interpretation. Two ordinal measures can not be multiplied together and still be given a meaningful interpretation in the strictest sense. You can not multiply numbers that only contain information about the ranking on respective dimensions. This might be brushed of as a concern for pedantic methodologists, but I will below show that this actually has some serious de facto implications for the validity of the “effective democracy”-measure. This goes especially for the issue of relative distance between units on the new variable, but also on the arbitrariness of rank between units. The arbitrariness of the scoring of the indicators, and the stringent mathematical assumptions underlying the possibility of multiplication, indicates that more subjective and intuitive evaluative methods have to be applied in order to actually judge whether the “effective democracy” measure validly captures reality, or if it is systematically misrepresenting the phenomena of actual degree of democracy. 4.3 Further considerations on the validity and reliability of the measure 17 The above discussion should have made relatively clear that there are no God-given ways of measuring “effective democracy”. Contingencies of measurement based upon arbitrary choice of principles, measures, scaling, variables included and such, make bombastic claims of the correctness of one single operational variable very problematic. Of course, this should not give rise to unconstrained intuitionism or discredit any attempt to classify; leading to a relativist claim that comparison is completely impossible. Theory and empirical generalizations on the historical functioning of certain institutions and how certain traits (clientilism and corruption) have in several cases affected the functioning of these institutions can give rise to a set of possible classificatory principles for “effective democracy”. I will draw an analogy from philosophical debates when elaborating upon the role of principles and overarching rules and their relation to the evaluation of specific cases. One illustration is John Rawls’ (1997: 18-19) treatment of ethical principles versus the evidence of ethical intuitionism on the basis of single cases in ethics, but others have treated the same problems in connection with epistemological issues (Quine 1956). Rawls’ point is that not only is our interpretation of different cases’ characteristics dependent upon which principles and theories we are operating with, but the intuitive and thorough investigation of the cases themselves might actually force through a revision of these principles. Rawls suggests that these mutual revisions of case interpretation and principle revision could go on until we reach some kind of “reflexive equilibrium”, where principles and case interpretation are in harmony. We saw above that the operationalization of “effective democracy” could be questioned strongly on validity grounds. Therefore, according to the “Rawlsian” logic, the more case-based interpretation of degree of democracy becomes an important factor in settling the case on whether the measure actually works as a good yardstick for democracy. The ranking of some cases, and the distance between the cases when interpreted linearly, point to deficiencies in the measure. Cardinal interpretation, trying to make sense of distances between units, is for example underlying linear regression analysis. Concerns on rank between cases were voiced by Hadenius and Teorell (2005), and I will elaborate on the “distance issue”. I will look at empirical characteristics of the effective democracy measure from 1999 systematically. A more comprehensive analysis would have looked more years. Empirical evaluation of the measure Table 4.2 shows the “effective democracy” scores for some selected countries, and figure 4.1 is a scatter-plot with FHI along the X-axis and “effective democracy” along the Y-axis. 18 Table 4.2: “Effective democracy” in selected countries (1999) Effective democracy 100 89 73 61 49 45 43 32 31 30 23 22 10 10 9 6 3 Country Denmark Norway Germany Spain Belgium Mauritius Italy South Korea Latvia Singapore Argentina India Zimbabwe Russia Nigeria Belarus China Figure 4.1: FHI and “effective democracy” (1999) Denmark 100,00 Iceland Canada Australia Ireland Austria United States 80,00 Portugal France Israel Spain Slovenia Chile Estonia Botswana Hungary Belgium Taiwan Italy Mauritius Namibia Czech Republic Uruguay 60,00 Effective democracy 40,00 20,00 0,00 Jamaica Korea, South Singapore Philippines El SalvadorLatvia Brazil Jordan India Argentina Mexico Guatemala Morocco Ecuador Malaysia Colombia Ukraine Venezuela Armenia Honduras Cote d'Ivorie Albania Nigeria Kazakhstan Belarus China Azerbaijan Afghanistan Burma 0,00 20,00 40,00 60,00 Inverted and normalized FHI 19 80,00 100,00 The case of Singapore was put forward by Hadenius and Teorell as an example of a regime that receives a too high democracy score by the effective democracy measure. The degree to which the elite in the PAP has controlled political decision and policy making, as well as curbed civil liberties like freedom of speech, freedom of press and freedom of association makes the regime deserve the badge relatively authoritarian, or at least semi-authoritarian. Singapore exercises elections, but they are arranged in such a way that the opposition has typically won one or two seats in the 81 seat parliament. This result is also due to several other aspects of political life like lack of civil liberties and arrest of opposition leaders. However, the fact that there exist elections, property rights and that corruption is lacking secures Singapore approximately 30% of the maximal FHI-score. This score is almost kept in the effective democracy measure because of the success of the PAP in stamping out economic corruption. However, as the discussion on ordinal measures showed, having 30% of the maximal effective democracy score is not necessarily the same as having 30% of the maximal FHI. If one uses the measure in regression analysis, the rank of cases and the distance between different cases are of utmost concern, and will drive the results. We cannot assign cardinal values of democracy or corruption, but we can try to make our measures keep some “intuitive” appeal, when it comes to distance between units and especially when it comes to rank. Even if democratic rights are rendered less effective by corruption and clientilism in India and corruption and populism in Argentina, I have a problem with seeing how one could claim based on a case-by-case understanding of the politics in these countries that the actual exercise of political and civil rights is larger in Singapore than these two countries. Countries such as India and Argentina are punished “twice” because they are first subtracted from deficiencies related to dysfunctional democratic practices in the FHI, and thereafter by multiplying this score with a low CPI. The same rank problem appears when claiming that Robert Mugabe’s Zimbabwe was more democratic than Russia (before Putin) and Nigeria (which admittedly is plagued by corruption, patrimonialism and shaky democratic foundations). These two latter countries are punished excessively because of their rampant corruption, and one is therefore underestimating the modest actual practice of political and civil rights that existed in these societies in 1999. Even if these countries are not exemplars of societies where the populace determines political decision making, they are not based on my knowledge worse than Zimbabwe, even if in 1999 the Mugabe regime had not yet shown its worst antics. 20 The linear interpretation of distances shows even more clearly the validity problems of the index. Remember that even if the measure is not metaphysically cardinal, one has to give it a cardinal interpretation if one is to use it in a linear regression. I have no large problem when interpreting the scores cardinally comparing the score of Singapore with Denmark’s score, claiming that the degree of popular control over politics in the former is approximately 30% of the latter (whatever this actually means). However, some of the distances when interpreted in light of other distances are clearly counter-intuitive, and I believe fall short of empirical evidence. A closer reading of the political history of Mauritius makes it clear that we are here dealing with a country where the populace has had relatively large influence over politics, with well-functioning elections, changes of government appearing smoothly, popular working class movements coming to power, and freedom of speech and assembly. There has also existed general responsiveness to popular demands through political channels in areas such as health care and educational issues (Bräutigam 1997). The Mauritian populace threw out a government by election when they wanted a change of economic course in the early eighties (Srebrnik 2002: 280). Even if there are claims of corruption in Mauritian society, it is an exaggeration to claim that Mauritius is closer in democracy score to Singapore than Spain, and it is even more disturbing that the distance to Denmark upwards is almost quadruple the distance to Singapore downwards. Mauritius is also closer in effective democracy to China and Belarus than to Denmark. Other examples can be provided. Surely, corruption in Belgium and Italy makes popular control over politics less efficient, but it is non-sense to place these countries closer to North Korea, Myanmar and Cuba than some Nordic countries. Either there is a serious validity problem with the measure, or the reliability is weak, leading for example the score of Belgium to be driven by relatively arbitrary events. Actually, computing the measure for Belgium in 2005, gives the country an effective democracy measure of 74. According to the measure, Belgium has in the last six years undergone an “effective democratization” of 25 points. This is more than Benin could be expected to have undergone if we were to have corruption scores before and after the overthrow of the Kèrèkou regime in 1989-1990. Benin shifted from an extremely oppressive and authoritarian regime to a remarkably well functioning democratic situation with multi-party politics, freedom of speech and assembly, even though it falls short of a democratic ideal. However, since Benin, when assigned CPI scores receives below 3, this event of regime change counts as about the same democratization as what Belgium is supposed to have experienced in the last years. I also have problems seeing how the 21 difference in effective democracy can be larger between Denmark and Norway than between Yeltsin-Russia and China, and that the difference between Denmark and Germany should be larger than the difference between Latvia and Zimbabwe. One general problem of the index is that it comprises all relatively low values. Corrupt democracies and semi-democracies are far closer to harshly authoritarian regimes than wellfunctioning democracies. Modestly corrupt and semi-democratic regimes receive scores that are undeservingly low, especially if they go together as they often do (not the least because the FHI already captures qualitative aspects). The reason behind is the mathematical multiplicative operation on the two ordinal indexes CPI and FHI. 0,1 squared is 0,01, which is very close to 0,3 squared (0,09). 0,8 squared (0,64) is however relatively far away from 1 squared. Empirically, we can see the implications by the fact that only 25 countries manages to rack up scores above 50 on the index ranging from 0 to 100 in 1999. The measure has the nice property of differentiating well between the relatively effective democracies, but fails to do so among other countries. The mean score is 31,2 and the median is an extremely low 21.3. Half the countries are pressed together on one fifth of the interval. This is problematic, when using the index in a linear regression, since results would tend to represent differences between the “West” and “the rest”, with the few countries dispersed over the relatively large intervals on the upper end of the scale driving regression coefficients. The claim from Inglehart and Welzel is that corruption scores only function as a grading factor, and that they do not determine “effective democracy” by compensating of political rights and civil liberties (2005: 194). However, this relatively dichotomized argument does not explain away the empirical fact that the CPI is actually higher correlated with the “effective democracy”-measure than the FHI. With a correlation coefficient of 0,95, we could actually use CPI as a decent proxy for effective democracy. The FHI and “effective democracy” have a correlation coefficient of 0,84. Table 4.3 Bivariate correlations between FHI, CPI and “effective democracy” (1999) FHI CPI Effective democracy FHI 1 0,67 0,84 CPI Effective democracy 0,67 1 0,95 0,84 0,95 1 22 The CPI’s empirical distribution is skewed in a way that makes the mass of countries obtain a relatively low score. Once again, I must remind that there is no strict cardinal interpretation of “corruption scores”, how corrupt is for example a 5? The median score for the 1999 survey is a low 3.2, and a score above 5 is only given to 26% of the countries. This explains much of the fact that “effective democracy”-index had a lumping together of countries at the bottom values. By using a multiplicative measure, a below 5-score in corruption guarantees that a country is not going to get more than half the score for “effective democracy”, no matter what the FHI-score should be. 5.0 Conclusion with suggestions for measuring “effective democracy”. When estimating a political regime’s “democratic” characteristics empirically, the need to go beyond the counting and classification of formal institutions and constitutional articles seems obvious to proponents of substantial democracy-definitions. The measure of effective democracy proposed by Inglehart and Welzel takes the issue of actual functioning of political institutions and the degree of popular control over them seriously. However, there are some serious flaws with the measure, indicating that the search for a measure of “true democracy” has not ended. All operationalizations and indexes have their validity and reliability problems. If there is no alternative measure available, the second best option is not to revert from measuring at all, but to be seriously aware of the problems of the index when analyzing and interpreting results from studies using the measure. This is especially important if the construction of the index can lead to systematic biases in the relationship under study. Another option is to use different measures in the analysis, and report the results stemming from usage of different operationalizations. One should not give up on finding a better measure of “actual” or “effective” democracy. One very interesting alternative is the “democracy audits” introduced by Beetham, and also argued by many to be very appropriate in contexts where the “deficiencies” of formal democratic institutions, like Africa, seem obvious (Baker 1999). In stead of relying on the a priori multiplicative assumption of formal democracy interacting with corruption, this approach is more empirical and somewhat more nuanced, asking a large amount of check questions, where the rating is clearly and explicitly related to the actual functioning of rights, liberties and institutions. Actors’ capacities can also be incorporated directly. Its drawback is the 23 amount of work needed to construct scores, the difficulty in securing comparability across countries, when country experts probably are needed in the scoring, and that measures could be less parsimonious, incorporating more than 80 questions (Beetham 1997). If one however wants to stick to a type of parsimonious framework like Inglehart and Welzel’s, one could improve upon the effective democracy measure as well, by making some adjustments. If the claim that FHI underestimates the degree of effective democracy as practiced by citizens still holds, but that one recognizes that Freedom House to a certain degree incorporates this functioning, one could weigh FHI stronger, for example by using a formula like X= FHI(2/3) * CPI(1/3). This would of course move away from a clean and intuitively appealing formula, but it should be clear by now that this formulation in any case gives arbitrary results, given the fact that we are not dealing with cardinal measures. This suggested formula would give rankings and differences in empirical democratic characteristics that make more intuitive sense than the originally proposed effective democracy measure: India would be somewhat further away from China, Singapore would be ranked lower, and Italy and Spain would again be somewhat closer to their other OECD partners. Another clear improvement would be taking the recognition that not only corruption affects the quality and actual practice of democratic rights and institutions. Instead of letting CPI be the quality component in an interactive measure, one could introduce a weighted index, where corruption could be incorporated, but also accompanied by other crucial variables like “responsiveness of elites”, “political power distribution” and other “rule of law” indicators. Even if less parsimonious, such a measure would clearly exhibit superior validity in capturing the actual functioning of democracy in a country. 24 6.0 Literature Arrow, K. Social Choice and Individual Values. New York: Wiley; 1951. Baker, B. The Quality of Democracy in the Developing World: why and how it should be measured. Mannheim: Paper for The 27th ECPR Joint Sessions of Workshops, (26-31 March 1999); 1999. Beetham, D. Conducting a Democratic Audit. University of Leeds: Manuscript; 1997. Beetham, D. Democracy and Human Rights. Cambridge: Polity Press; 1999. Bollen, K. Liberal Democracy: Validity and Method Factors in Cross-National Measures. American Journal of Political Science 1993;34:1207-1230. Bräutigam, D. Institutions, Economic Reform and Democratic Consolidation in Mauritius. Comparative Politics 1997;30:45-62. Chabal, P., Daloz, J.P. Africa Works: Disorder as Political Instrument. Bloomington: Indiana University Press; 1999. Dahl, R.A. Polyarchy. Participation and Opposition. New Haven: Yale University Press; 1971. Diamond, L. Developing Democracy. Toward Consolidation. Baltimore: The Johns Hopkins University Press; 1999. Diamond, L., Morlino, L. Introduction. In Diamond, L., Morlino, L. editors. Assessing the Quality of Democracy. Baltimore: The Johns Hopkins University Press; 2005. p. IX-XLIII. Elkins, Z. Gradations of Democracy? Empirical Tests of Alternative Conceptualizations. American Journal of Political Science 2005;44:293-300. Freedom House. Freedom in the World 2005. Lanham: Rowman & Littlefield; 2005. Goodman, L. The Variance of the Product of K Random Variables. Journal of the American Statistical Association 1962;57:54-60. Hadenius, A., Teorell, J. Cultural and Economic Prerequisites of Democracy: Reassessing Recent Evidence. Studies in Comparative International Development 2005;39:87-106. Inglehart, R., Welzel, C. Modernization, Cultural Change and Democracy – The Human Development Sequence. Cambridge: Cambridge University Press; 2005. Lijphart, A. Patterns of Democracy. New Haven: Yale University-Press; 1999. Médard, J.F. Patrimonialism, Neo-Patrimonialism and the Study of the Post-colonial State in Subsaharian Africa. In Marcussen, H.S. editors. Improved Natural Resource Management – the Role of Formal Organizations and Informal Networks and Institutions. Occasional Paper 17, International Development Studies, Roskilde University, Denmark; 1996. Meisenhelder, T. The Developmental State in Mauritius. The Journal of Modern African Studies 1997;35:279-297. 25 Przeworski, A., Alvarez, M.A., Cheibub, J.A, Limongi, F. Democracy and Development. Cambridge: Cambridge University Press; 2000. Quine, W.V. Two Dogmas of Empiricism. In Quine, W.V. From a Logical Point of View. Boston: Harvard University Press; 1953. Rawls, J. A Theory of Justice. Rev.ed. Oxford: Oxford University Press; 1999. Schumpeter, J.A. Capitalism, Socialism and Democracy. 4th ed. New York: Harper Perennial; 1976. Srebrnik, H. ‘Full of Sound and Fury’: Three Decades of Parliamentary Politics in Mauritius. Journal of Southern African Studies 2002;28:277-289. Welzel, C. Are Levels of Democracy Affected by Mass Attitudes? Testing Attainment and Sustainment Effects on Democracy. International Political Science Review 2007;28 (4):397424. Welzel, C., Inglehart, R. Emancipative Values and Democracy: Response to Hadenius and Teorell. Studies in Comparative International Development 2006;41 (3):74-94. Welzel, C., Inglehart R., Klingemann, H.D. The theory of human development: A crosscultural analysis. European Journal of Political Research 2003;42:341-379. XXXXXX Political regime types and economic growth – Are democracies better at increasing prosperity? XXXXXXXXX; 2006. i There are of course several different ways to measure equality, or if one wants the spread of a good over a population. Therefore, it is not always clear in principle whether a situation A incorporates a more egalitarian distribution of a resource (power, wealth etc) than another situation B, and especially how much more. However, in practice one can at least often rank two situations relatively determinately. Whether political power resources is more equally distributed between citizens in Norway and Sweden is debatable, but this is not the case when comparing Norway and Togo, independently of measure. ii See Goodman (1962) for the specific expressions and approximations of variances of products. The article treats both the special case where the variables that make up the factors of the product are independent, and also when they are not. 26