Resumé

In their well-known volume on 'Size and Democracy' (1973), Robert Dahl and Edward Tufte argue that small units are likely to be more homogeneous, whereas larger units are likely to exhibit more diversity. This study of the microstates of the world and of selected control groups of states supports this view only in part. In terms of attitudinal diversity, smaller units are indeed more homogeneous. In terms of ethnic and religious diversity, however, no significant differences emerge between small states and large states. This suggests that categoric differences are transformed in larger units to a greater extent into attitudinal differences. Bearing in mind that most microstates are island states, the capacity of microstates to manage ethnic diversity may in several cases be due to the intimacy of island communities which binds members together in mutual solidarity.

Introduction

In political philosophy and political science, a long line of thought has argued the virtues of the small polity. To Plato, unity was a defining characteristic of a state, and this had consequences for the outlook of a state. As large units were likely to display disunity and class conflicts, only small units could constitute states (e.g., Bratt 1951, 8-9). In like manner, in The Spirit of Laws, Montesquieu (1961, 131-34) argued that it is in the nature of a republic to have only a small territory, and that the form of government adopted by any state is closely related to its size. Modern political science literature likewise accommodates several notions that emphasize the importance of size. The fact that political entities, like states, municipalities or organizations differ in size, it is said, carries implications for phenomena, structures and events that are related to such entities. For instance, in party research many authors seem to feel that size is an important factor in relation to internal party cohesion, mobilization of party members and bureaucratization (Panebianco 1988, 186-90). Also, in the study of international politics the belief is widespread that the size of states influences the behavior of states in, for instance, matters of security and alliance options (e.g., Amstrup 1976).

---

*Dag Anckar, Department of Political Science, Åbo Academy, Biskopsgatan 15, FIN-20500 Åbo, Finland. E-mail: dag.anckar@abo.fi

This article deals with size differences between nations and investigates one specific belief, which connects small size and homogeneity. The belief is that the smaller a political system, the more homogeneous it is likely to be, and, conversely, that the larger a political system, the more politically competitive is likely to be. This assumption is presented and discussed at length in the by now classic volume by Dahl & Tufte (1973) on Size and Democracy, which is the source of inspiration as well as the target for this research. Small units, Dahl & Tufte propose, are likely to be more homogeneous respect to beliefs, values and goals, whereas larger units are likely to exhibit more diversity (ibid., 13-14). An increase in size, the authors presume, contributes to and is associated with greater heterogeneity (ibid., 97). 'With increasing size,' they write, 'persistent and overt differences political outlooks, interests, and demands are likely to appear' (ibid., 91).

Dahl & Tufte also explain why this would be the case. The proposition is widely held to be true, the authors assert, that if a system is small and therefore homogeneous, 'variations in behavior are fewer, a higher percentage the population adheres to a single code, the norms of the code are easily communicated by word and example, violations are visible, sanctions are easy to apply by means of both gross and subtle forms of social interaction, and avoidance of sanctions is difficult' (ibid., 92). The view that the smaller the society, the fewer the distinct parties and interests can be found also in The Federalist Papers (Hamilton, Madison & Jay 1961, No. 10, 83), and several modern political science authors have echoed similar arguments. For instance, Muir & Paddison (1981) have argued that the smaller the size of a jurisdiction, the higher the cohesiveness of the community, and Lijphart has suggested that small states offer better requisites for the creation of a spirit of co-operativeness and accommodation. small states, according to Lijphart (1977, 65-67), political leaders will be more united through personal acquaintance and interaction. Other size-related factors work in the same direction. Small units are often dependent in their economies on a certain sector or product, the prospects of which become a common and unifying concern, which adds to homogeneity. 'Let there be no mistake,' de Backer writes (1993, 10), describing unease of the Dominicans over the future of their banana trade and industry, the banana competition 'leaves no Dominican, whether small farmer, civil servant, business man or housewife, indifferent.'

However, little systematic empirical knowledge exists in matters of this kind. In his treatise Democracy and Development, Hadenius asks what empirical research has actually demonstrated about the relationship between size and democracy; his answer reads: 'Not much, actually' (Hadenius 1992, 125). The same can be said about research into the link between smallness and homogeneity: whereas ideographic observations of

individual systems abound, there is a lack of systematic overview and systematic knowledge. Let it also be said that ideographic observations do not convey a uniform pattern. Some small states are described as homogeneous, some as heterogeneous. The maneaba system in Kiribati, one observer notes, embodies a mono-culture and mono-class system: 'society is basically egalitarian and is maintained through the ideas perpetuated in the maneaba' (Tabokai 1993, 27). On the other hånd, one observer notes about Mauritius that 'in a society of class, religious, color, caste, and linguistic differences there is no end to the demands on political leaders from thedifferent Mauritian communities' (Bowman 1991, 68).

It is the aim of this article to provide a systematic view of the extent to which small states in fact display homogeneity. A related aim is to test the correctness of Dahl & Tufte's assumptions. This testing involves internal as well as external comparisons, i.e., comparisons within a group of small states and comparisons between small and large states. In these exercises, two kinds of findings are disturbing from the point of view of the Dahl & Tufte theory. On the one hand, if there are great differences between small nations in terms of homogeneity, the theory is obviously disconfirmed. Smallness, then, does not produce homogeneity, as small nations may be homogeneous as well as heterogeneous. On the other hand, if there are no big differences between small and large nations in terms of homogeneity, it again follows that Dahl & Tufte's assumption appears questionable. The more similarities there are between small and large nations in this respect, the less likely it becomes that homogeneity is indeed a function of small size.

The article has six sections. Following this introduction, the second section introduces some observations on the concept of size and identifies a population of very small states. For this population, the third and the fourth sections, the third dealing with categoric diversity and the fourth with attitudinal diversity, provide empirical evidence to support or repudiate the belief that small units are also homogeneous units. In these two sections, systematic comparisons are made between small nations and various sets of larger nations. The fifth section suggests an explanation for the finding that categoric differences are transformed into attitudinal differences to a larger extent in large units. The article is closed by an endnote on size differences within the small size category.

Operationalizing Smallness

To determine whether or not small states are homogeneous, a set of small states needs to be identified. However, this is not an altogether easy task. It is stated in the literature that no wholly agreed definition of a small state exists, and that any definition is to some degree arbitrary (Sutton & Payne

1993, 581). This is certainly true. However, one indicator which is common in the literature is population size: states with a population of less than one million are classified as small (ibid., 582). This measure will be applied here. One definite advantage of the method is that it singles out states that are truly small; in fact, such states may be referred to as microstates, to make use of still one size-related term. There are indeed differing definitions of microstates, and little consensus exists about the ceiling population of such states (e.g., Ogashiwa 1991, ix). However, to settle for one million as the ceiling population is not a highly controversial solution.

Of the nearly 200 independent states in the world, 43 are microstates. These are listed in Table 1, which also reports area and population figures for each state. Almost half of the states are island states, confirming the geographical notion that the smaller the territory, the more likely it is to be an island (Caldwell et al. 1980). From a theoretical point of view, the large amount of islands is a relevant as well as welcome feature. The very concept of island implies at least a certain degree of remoteness, and remoteness, again, as it is likely to become a unifying factor and a frame of reference shared by most members of the island community, fosters fellowship and a sense of community, also homogeneity in terms of values and attitudes. Individual populations, however small, can easily evolve a strong sense of a separate identity,' Newitt notes in an introductory essay to a volume on the political economy of small island states (1992, 11). If one wants to question the validity of propositions that link smallness to homogeneity, small islands are therefore ideal cases. If the propositions do not survive tests in this homogeneity-loaded context, their overall chances appear slim, indeed.

However, as explained earlier, isolating a population of very small states is not enough. If and when there are great differences in terms of homogeneity these states, the link between smallness and homogeneity may certainly be questioned, as similarities in the independent variable (smallness) are matched by dissimilarities in the dependent variable (homogeneity). Still, the variation notwithstanding, it may be the case that large states display even higher degrees of heterogeneity. If and when this is the case, differences between small states are smoothed out. The variations in the dependent variable are now of less significance; in fact, as they are overridden by the differences between large and small, they start to look like similarities, thus producing a correspondence between smallness and homogeneity. they no longer question the belief that smallness goes hand in hand with homogeneity. To check for this, control populations that introduce thresholds of size are needed. However, control populations must be composed on the basis of specific demands and criteria, so it is unlikely that one and the same population can be adjusted to serve several analytical purposes. Identification of control populations will be dealt with in proper contexts later in this article.

Categoric Diversity

Homogeneity is a multi-faceted concept. It can, however, be made manageable by means of a distinction, proposed by Dahl & Tufte (1973, 30) in their treatise on size and democracy. When talking about the complexity and diversity of political systems, the authors define a system as being more complex (1) the greater the number of categories of actors whose attitudes, interests, wants, preferences, demands and goals have to be taken into account; and (2) the greater the variation in their attitudes. While the authors in their own empirical presentations and illustrations choose to focus on the first criterion, differences in categoric diversity, they neglect the second criterion because of a lack of available data (ibid., 30-31). From a theoretical point of view, however, the second criterion, which is about attitudinal diversity, is equally important. It taps another and different dimension of homogeneity: clearly, for instance, value and goal conflicts may exist among population strata that are fairly homogeneous in terms of race, language or religion. Categoric homogeneity is one thing, and attitudinal homogeneity is another. Here, therefore, separate measures of the two dimensions will be provided, this section dealing with categoric diversity and the next section dealing with attitudinal diversity.

To measure the extent of categoric homogeneity, several distinctions and classifications can be used. Some are about cultural diversities, others are about socioeconomic diversities (ibid., 31). Some, however, are more basic than others. The most troublesome ones in terms of conflict management and democratic performance are generally adjudged to be the ethnic and the religious cleavages. They are characterized by more of a dichotomy, and they have, in addition, a profound emotional significance for the groups concerned (Hadenius 1992, 112-13). To measure cleavages of an ethnic and religious nature, a variety of methods, techniques and approximations are at the disposal of the researcher. One method, employed, for instance, by Vanhanen (1990, 110-11), and recommended also by Dahl & Tufte (1973, 31), is to apply an index based on the percentage of the largest homogeneous ethnic group, this term referring equally to racial groups, tribes, national groups, language groups or religious communities. As pointed out by some authors (e.g., Hadenius 1992, 114-15; Anckar 1998, 145-46), this method has severe shortcomings. The focus on the largest group disregards the simple fact that the rest of the society may be differently shaped in different countries. For instance, two states may be similar in terms of the size of the largest group, but may differ considerably when it comes to the existence and size of other cleavage groups: the method suggests that the two states are similar, although they clearly are not.

Here, following the method applied by two Finnish scholars (Anckar &
Eriksson 1998), the fragmentation index suggested by Rae for the

calculation of party system fragmentation is used to compile the extent of categoric homogeneity in each and every small island state. The theoretical rationale for the Rae formula is that it represents the frequency with which pairs of voters would disagree in their choice of parties if an entire electorate would act randomly (Rae 1971, 55-56); the rationale is modified by Anckar & Eriksson to describe the probability that randomly selected samples of one percent of the population consist of individuals belonging to different ethnic groups. This index, which runs from 0 to 1, the values indicating the more homogeneity the more they approach 0, builds upon the total number of distinct groups as well as the size of the dominating group and therefore avoids the pitfalls mentioned above. In their study, for all states of the world, Anckar & Eriksson compile separate indexes that describe cultural diversities: one is about ethnic and linguistic homogeneity, another about religious homogeneity. These two indexes are used here. The two indexes are also added to provide a total diversity index.

We need to return at this point to the question of control groups. The dependent variable being here about ethnic and religious orientation, i.e., qualities that are present in any state, although in varying degree, no other considerations apply for the design of groups than those that concern differences in size. Four control populations are used. The first features all states in the world with a population of approximately 3-5 million; there are 28 states in this category, the population size ranging from 2.8 to 5.2 million. These states are in the following referred to as small states; they are larger than microstates but they are still small. The second control group consists of the medium-sized states of the world, i.e., states with populations in the range of 9-11 million. There are 17 states in this category, the precise range running from 8.7 to 10.9 million. In the third group are all large states of the world, numbering 21. These are states with populations in the range of 20-50 million, the precise range running from 20.3 to 44.1 million. Finally, a group of gigantic states is identified, comprising all states of the world with populations exceeding 75 million. There are 12 states in this category.

For the microstate population as well as the four control populations, data concerning ethnic and religious diversity are given in Table 2. Because of missing data, Belau as well as Sao Tome and Principe are excluded from the microstate group. There is in the data a weak tendency in the direction of larger units being less homogeneous than smaller units, but this tendency is weak and also less than consistent. The general impression is that the pattern is fairly similar over size categories; apparently, size does not make that much of a difference. Also, as clearly indicated by the range figures, mini-sized units do not display a higher internal homogeneity than larger states. In terms of ethnicity, microstates like Tonga (0.02), Cape Verde (0.04), Kiribati (0.05) and Iceland (0.08) are homogeneous, but other microstates,like (0.74), Micronesia (0.75), Qatar (0.76) and Djibouti

(0.81) are anything but homogeneous. On the other hand, whereas certain gigantic and large states like Indonesia (0.80) and Kenya (0.90) are indeed heterogeneous, others like Bangladesh (0.05) and North Korea (0.00) are even extremely homogeneous. Concerning religious diversity, Andorra (0.01), Comoros (0.01) and Malta (0.05) are examples of homogeneous microstates; among the heterogeneous microstates are Swaziland (0.72), Guyana (0.73), Cyprus (0.76), Vanuatu (o.B2)and others. Several large states, like Kenya (0.82) and South Korea (0.86) are heterogeneous, others, like Thailand (0.11) or Nepal (0.19), are fairly homogeneous. Among the gigantic states Nigeria is heterogeneous (0.70), whereas Pakistan is not (0.06). The outcome of this mapping, therefore, does not substantiate the view that small is homogeneous. Rather, the results repudiate this view.

To sum up: it is not true that small states are more homogeneous in terms of categorical diversity than middle-sized or large states. Size does not appear to make much of a difference. In his study of democracy and development, Hadenius concludes that the often very small island states are not as special as they appear to be on simple inspection of the political geography (Hadenius 1992, 127). Given the context and the frame of this statement, its overall validity may be questioned (Anckar & Anckar 1995); in the context of the present research, however, the conclusion seems well substantiated. Among the microstates, many of which are islands, one may find units that are fairly homogeneous, units that are less homogeneous, and units that are anything but homogeneous. In terms of an internal comparison, therefore, the Dahl & Tufte theory lacks validity. Furthermore, small states on the one hand and large states on the other, no significant differences in terms of homogeneity emerge. The findings from the internal comparisons are thus substantiated by the findings from an external comparison.

Attitudinal Diversity

The task of measuring attitudinal diversity is approached here by the application of a simple measure of political plurality and competition, namely the number of political parties in the units. The theoretical point of departure for this choice of an operational device is a notion of parties as the most visible and in many ways the most important organizations involved in conflict resolution (Dahl & Tufte 1973, 96). The following reasoning applies: the more parties there are, the more dissenters from the majority perspective; the number of parties, therefore, reflects attitudinal differences. Indeed, in their treatise on size and democracy, Dahl & Tufte suggest that in the politics of homogeneity parties scarcely exist, and that in the process of party emergence, homogeneity initially tends to ensure the dominance of a single party (ibid., 97). The use of the number of parties as a shortcut for attitudinal diversity is therefore very much in line with the theory to be tested here. Indeed, it is hard to think of any other convenient and appropriate device for measuring attitudinal homogeneity in research that taps many countries.

It needs to be emphasized that what is counted here is the totality of actually existing parties. Since the theoretical point of departure does not discriminate against any kind or type of parties, no 'intelligent counting' (Sartori 1976, 121-25) is used which discounts parties that seem irrelevant on the basis of this or that counting rule. As they are clearly relevant for the measuring of attitudinal homogeneity, small and insignificant parties are on equal footing with big and important parties. However, the identification of parties in different countries encounters methodological as well as technical difficulties in several cases, and the results of the count are therefore by necessity somewhat imprecise. It is stated in a reference guide on political parties that 'the dividing line between a political party and a protest movement or pressure group is necessarily imprecise and varies in accordance with the context in which an organization operates' (Lewis & Sagar 1992, vii); the editors of the guide also explain that they have in their effort accepted self-definition as a starting point for inclusion, although editorial judgments have had to be made on a case-by-case basis (ibid.). Inevitably, the figures presented here sustain the same kind of imprecision and vagueness. To ensure comparability in the identification of parties, only a few sources have been consulted which can be expected to have applied equal standards in the registration of parties.¹

For a specific reason, i.e., the nature of regimes, it is not possible to make use here of the same microstate population and the same control populations as in the preceding section. Obviously, if homogeneity is measured by the existence of parties, only countries that allow free party formation qualify as units for study. States which do not recognize this right

to an open political competition become uninteresting from the point of view of this investigation, simply because they restrict the space for an interplay between the independent and the dependent variable. When this criterion is applied to the universe of microstates, eight out of the total of 43 states must be disqualified because they do not allow parties or allow only a restricted number of parties. These authoritarian or semi-authoritarian regimes are: Bahrain, Bhutan, Brunei, Djibouti, Maldives, Qatar, Swaziland and, for obvious reasons (Duursma 1994, 413-25), the Vatican City. The ministate population which is used here is thereby reduced to 35 cases.

What is needed in terms of comparison and control is a set of countries which (a) differ in size and (b) do not prevent the establishment of political parties; for this set, information is also needed about the number of existing parties in each country. Fortunately, this information is available. In his recent doctoral dissertation, Anckar (1998) studies the relation between country size and party fragmentation. The 77 countries included in his study are selected on the basis of criteria which fit the requirements of this study rather nicely: countries which (a) grant or at least do not withhold the right to party formation, the classifications being based on the Freedom House ratings for the year 1995 and (b) are stable democracies in so far as they satisfy a criterion of at least three open parliamentary elections by 1995. For these countries, the number of parties in each country is established,² the party conception being very similar to the one used here for the microstate population. The figures are therefore comparable with the corresponding microstate figures of this study.

The states in Anckar's sample are dealt with here in the following manner: (a) from the 77 states, 24 microstates are removed, all of which are already included in the microstate population of this study; (b) of the remaining 53 states, 21 are large states with populations exceeding 20 million; these states form one control group; (c) of the remaining 32 states, 20 states with populations between 5 and 20 million are classified as medium-sized states and are grouped to form another control group. For the microstate population of this study and the two control groups, Table 3 shows the average number of parties as well as the actual range within each category. From the figures, two main observations emerge. First, the

expected link between size and attitudinal homogeneity is certainly there. There are, on average, clearly more parties in medium-sized states than in microstates; furthermore, there are, on average, clearly more parties in large states than in medium-sized states. It would now appear that Dahl & Tufte are right in emphasizing the relationship between size and homogeneity:the the size, the less attitudinal diversity.

Second, however, as suggested also by the range figures, the findings from an internal comparison do not invalidate the view that smallness may be linked to heterogeneity and homogeneity. In fact, if one looks at individual cases, one may find among microstates a bewildering variety of party systems, ranking from systems with no parties to systems that are characterized by atomization. For instance, there are cases like Belau, the Federated States of Micronesia and Tuvalu which have no political parties. The absence of parties in these miniature democracies derives from several interrelated factors associated with diminutive size, insularity, excessive geographical noncontiguity and cultural heritage (Anckar 1997, 248). There is one party in the Marshall Islands and one in Nauru; furthermore, these groups may be classified as parties only if and when the electoral criterion is relaxed (Anckar & Anckar 1999). On the other hand, there are 10 parties in Vanuatu, 16 parties in Fiji, Guyana and Surinam, and no less than 18 parties in Equatorial Guinea and 19 parties in the Comoros. The issue of small being homogeneous is therefore open to interpretation. If attitudinal homogeneity is regarded as a continuous phenomenon (more or less), the idea that small is homogeneous certainly appears valid. If, on the other hand, attitudinal homogeneity is classified as a dichotomous phenomenon (either or), the idea rests on more shaky grounds.

One possible objection to the findings suggests that the number of parties in one country is affected heavily by the electoral formula which is in use in that country and that the impact of the formula therefore must be controlled for. For instance, reference could be made to the well-known and much researched generalization known as Duverger's Law, stating that a plurality electoral system leads to a two-party system (Duverger 1964, 216-28). Although the differences between categories in Table 3 are clear enough to suggest that the electoral system impact may be negligible, a control for this factor has been effected. Unfortunately, however, due to a heavy bias in the sample of medium-sized states in favor of proportional electoral systems, a control for this size category is not possible. Within the microstate group, a difference emerges which is supportive of Duverger's Law: nations with proportional or mixed systems (N 13) have, on average, 8.1 parties, whereas nations with plurality or majority systems (N 22) have, on average, 5.1 parties. Within the group of large states, the difference is smaller: nations with proportional or mixed systems (Nil) have, on average, 26.2 parties, whereas nations with plurality or majority systems (N 10) have, on average,

27.0 parties.³ The main conclusion from this exercise, however, is that the differences between the size categories remain. Small states are more homogeneous in terms of attitudes than large states, and this homogeneity is not a function of the use of one or the other electoral system.

Discussion

To repeat, the findings from this study are in less than full agreement with a central assumption by Dahl & Tufte. Referring to the thoughts of Rousseau and Montesquieu, the authors expect that the small system, being more homogeneous, is likely to be more consensual, whereas the large system, being more heterogeneous, is likely to be more conflictual (Dahl & Tufte 1973, 91). This research, however, has demonstrated that whereas small units are indeed more homogeneous than larger units in terms of attitudinal diversity, there are no significant differences between small and large units in terms of categoric diversity. Translated into the language of Dahl & Tufte, this means that small systems and large systems are equally homogeneous; still, small systems are less conflictual than large systems. This almost suggests a paradox: the mechanisms described by Dahl & Tufte appear to be operative, although the homogeneity premise from which they are derived is not. Or, in other words, rather than homogeneity, smallness counts. The implication of this finding is that categoric differences are transformed to a greater extent into attitudinal differences in large units; in so far as categoric differences are seen as sources of attitudinal differences, small units are more equipped to manage and restrain these sources.

In the search for explanations of this difference, one specific aspect merits close attention. It was stated earlier in this article that islands, because of a remoteness factor, may be expected to display attitudinal homogeneity to a higher extent than other units. When one bears in mind that out of the 43 microstates no less than 29 are island states, conceivable explanations for the capacity of small units to handle diversity may be derived from island characteristics. The relevant mechanism is illustrated in a statement by Sir Baddeley Devesi, Deputy Prime Minister of Solomon Islands, in an interview from 1992: 'We are an island state and a very small place, so everybody knows who's who' (Homer 1992, 32). This statement introduces the concept of intimacy and points at smallness and insularity as specific sources of intimacy; intimacy, again, is one of the system features that, according to Dahl & Tufte, work in favor of consensualism. Indeed, as suggested by Newitt (1992, 16-17): 'the smallness of islands adds a dimension of intimacy to their affairs, for the same families are involved in all aspects of island life and the political elites remain closely connected with farming, business, and local affairs.' However, as clearly indicated by

Newitt (ibid., 16), remoteness may add to the impact of smallness: 'Not all small states are islands and not all island states are small, but the problem of smallness is given an added dimension in the case of an island, and insular isolation can be considerably intensified if you are also small.' Indeed, isolation may promote an increased sense of community, as people share a feeling that they are, so to speak, alone in the world and thrown upon their own resources. About the political culture of the English-speaking islands of the south-eastern Caribbean, it has been said that 'it is a political culture that is essentially inward-looking and insular, where the surrounding sea is a barrier rather than a highway' (Thorndike 1991, 110).

The core of this reasoning is, then, that small islands are characterized by intimacy and thereby well equipped to deal with categoric differences in a consensual rather than conflictual manner. The reasoning suggests that small islands have a strong sense of local community that binds members of tiny places together in mutual solidarity. In such places, differences among individuals are often ignored, and a confining sense of geographical limits reduces interpersonal friction: 'Small island residents simply know that they must get along' (Richardson 1992, 195). A calculation which is unfortunately by the low number of cases offers at least suggestive support for the validity of this kind of reasoning. There are in all 32 microstates with populations of less than 500,000. Of these 32 states, 24 are island states. The average number of parties in these island states is 4.1 as against 7.3 parties in the remaining mainland microstates.

A Note on Thresholds

One further aspect of the issue at hand remains to be dealt with. This aspect taps the very conception of smallness used in this article. One could in fact argue that smallness has been defined here in terms that are too generous and therefore serve to disguise the link between smallness and homogeneity. In their study of size and democracy, Dahl & Tufte take care to emphasize that their line of reasoning may apply to very small systems only; indeed, so far as the testing of their paradigm is concerned, even most subunits of political systems, like towns, cities, states and provinces, may be too large (1973, 94-95). Although recent research (Anckar 1998) certainly indicates that important elements of the paradigm are valid in comparisons between nations, there are also in the literature empirical findings that suggest that even within groups of very small states, important thresholds of size are indeed operative. For instance, in a study of dominating parties in small island states, Anckar (1997) was able to trace an impact of size on the emergence of predominant party systems only if and when certain thresholds of size are passed: eleven out of the fourteen smallest nations in

the sample confirmed the theory, whereas there were ten out of twelve
disconfirmations within a group of the fourteen largest nations.

As is evident from this article, no specific threshold argument is needed to support the idea that attitudinal homogeneity is a function of size. In the present context, the implication of the threshold idea would therefore be that very small microstates display categoric homogeneity, whereas other small microstates do so to a lesser extent. To control for this aspect, the microstate category is cut down in the following to comprise truly diminutive states only. A further classification of these states into four size categories is used. The first category comprises all microstates with populations of less than 500,000; the second category singles out from these states all cases with populations of less than 100,000; the third comprises states with populations below 50,000; finally, the fourth includes a couple of cases with populations below 20,000. For these microstate categories, Table 4 provides information about the extent to which the states on average display categoric and attitudinal homogeneity. A familiar pattern emerges: again, no significant differences can be detected in terms of ethnic diversity, whereas the number of parties follows rather closely a size dimension. Categoric diversity is no function of size, but attitudinal diversity is. And again: this second pattern is not a function of the fact that the states employ different electoral systems. It would appear, therefore, that the Dahl & Tufte threshold theory does not receive support in this investigation. The link between small size and categoric homogeneity is simply not there, not even among the very smallest of nations.

One obvious implication of this finding is that the relation between the two types of homogeneity is blurred in the microstate context. As a rule, one would expect that countries that have a high categoric homogeneity represent a high attitudinal homogeneity as well. On the other hand, low categoric homogeneity countries, according to the same logic, should represent low attitudinal homogeneity. However, these expectations are false. For 30 microstates with populations of less than 500,000, Table 5 reports the results of a crossing of the categoric homogeneity dimension and the attitudinal homogeneity dimension. The cutting points that are used are given in the table. The result is fairly straightforward: types of homogeneitydo

geneitydonot necessarily correlate with each other. In about half of the cases only, countries with high categoric homogeneity display high attitudinal homogeneity as well. Also, only two out of eight categoric diversity cases display high attitudinal diversity as well.

Of course, the objection is still valid that a full repudiation of the Dahl & Tufte threshold theory awaits intra-state comparisons, states being too large units of research. Two things, however, need to be said in this connection. One is that this study has considered several units that are in fact smaller than the average town, district or province; indeed, to the extent that truly diminutive of size need to be considered in order to detect relevant relationships, this study has fulfilled this condition. The other is that the very notion of a link between smallness and homogeneity somehow becomes uninteresting in fact disappears almost into meaninglessness from the comparative point of view if the units that are considered must be so small that not even the smallest of the small nations of the world merit attention.

---

NOTES

1. The main sources for the information on microstate parties are Anckar 1998 and the latest issues of Regional Surveys of the World (various volumes).

2. In Anckar's study, the registration of parties is made on the basis of information in various individual sources and reference guides. The main source is the handbook series Political Parties, issued by Longman Current Affairs, which covers all regions of the world.

3. Data concerning electoral systems are from Blais & Massicotte 1997 and Anckar 1998.

REFERENCES

Amstrup, N. 1976. 'The Perennial Problem of Small States: A Survey of Research Efforts,"
Cooperation and Conflict 1!, 163-82.

Anckar, C. 1998. Storlek och partisystem. En studie av 77 stater. Åbo: Åbo Akademis
Forlag.

Anckar, C. & Eriksson, M. 1998. 'Measuring Ethnic and Religious Diversity.' Department
of Political Science, Åbo Akademi University, Occasional Papers Series, No. 10.

Anckar, D. 1997. 'Dominating Smallness: Big Parties in Lilliput Systems,' Party Politics 3,
243-63.

Anckar, D. & Anckar, C. 1995. 'Size, Insularity and Democracy,' Scandinavian Political
Studies 18,211-29.

Anckar, D. & Anckar, C. 1999. 'Democracies without Parties,' Comparative Political Studies,
forthcoming.

de Backer, R. 1993. 'Much ado about . . . bananas,' The Courier, No. 140 (July-August),
10-18.

Blais, A. & L. Massicotte 1997. 'Electoral Formulas: A Macroscopic Perspective,' European
Journal of Political Research 32, 107-29.

Bowman, L. W. 1991. Mauritius. Democracy and Development in the Indian Ocean. Boulder,
Colorado: Westview Press.

Bratt, E. 1951. Småstaterna i idéhistorien. Uppsala: Skrifter utgivna av Statsvetenskapliga
foreningen i Uppsala.

Caldwell, J. C, Harrison, G. E. & Quiggin, P. 1980. 'The Demography of Microstates,' World
Development 8, 951-965.

Dahl, R. A. & Tufte, E. 1973. Size and Democracy. Stanford: Stanford University Press.

Duverger, M. 1964. Political Parties: Their Organization and Activity in the Modem State.
London: Methuen.

Hadenius, A. 1992. Democracy and Development. Cambridge: Cambridge University Press.

Horner, S. 1992. 'Solomon Islands,' The Courier, No. 132 (March-April), 27-31.

Lijphart, A. 1977. Democracy in Plural Societies. A Comparative Exploration. Yale: Yale
University Press.

Montesquieu, 1961. De l'Esprit de lois, Vol. I. Paris.

Muir, R. E. & Paddison, R. 1981. Politics, Geography and Behavior. London: Methuen.

Ogashiwa, Y. 1991. Microstates and Nuclear Issues. Suva: University of the South Pacific,
Institute of Pacific Studies.

Panebianco, A. 1988. Political Parties: Organization & Power. Cambridge: Cambridge
University Press.

Rae,D. 1971. The Political Consequences ofElectoral Laws. 2nded. Yale: Yale University Press.

Regional Surveys of the World. The Far East and Australasia 1997. 28th ed.; South America,
Central America and the Caribbean 1997, 6th ed.; Africa South of the Sahara 1997, 26th ed.
London: Europa Publications.

Richardson, B. C. 1992. The Caribbean in the Wider World, 1492-1992. Cambridge:
Cambridge University Press.

Sartori, G. 1976. Parties and Party Systems. Cambridge: Cambridge University Press.

Tabokai, N. 1993. 'The Maneaba System,' in van Trease, H., ed., Atoll Politics. The Republic
of Kiribati. Christchurch: Macmillan Brown Centre for Pacific Studies, University of
Canterbury.

The Times Guide to the Nations of the World 1994. London: Times Books.

Thorndike, T. 1991. 'Politics and Society in the South-Eastern Caribbean,' in Clarke, C,
ed., Society and Politics in the Caribbean. London: Macmillan.

Vanhanen, T. 1990. The Process of Democratization. New York: Taylor & Francis.