This is not so much a rebuttal of Kimber (1989)* as an attempt to examine the problems from a different angle. There is no quarrel with the view that much of what is said about democracy' is confusing. But I argue that it is possible to categorise political regimes without too much difficulty. In categorising empirical political scientists need models not too sophisticated: indeed not unrelated to those understood in 'ordinary discourse'. Perfect analytic rigour is impossible and any attempt to demand it is unrealistic and will eventually detract from the very necessary attempt to comprehend different types of political regime.

In his recent article Kimber 'attempts to provide a modern, universal, conceptualization of democracy' that 'may be useful analytically' (p. 199). He speaks of clarifying the concept and putting forward 'a serviceable formulation that may then provide a solid starting-point for empirical analysis'. It is common knowledge that 'democracy' is a notion that has often lacked clarity in its usage. Firstly and obviously because politicians have muddied it for their own purposes. Secondly, commentators within our discipline defined the concept differently and thus have added to the confusion. For example, there was the post-war argument about 'the two theories of democracy', variously described as 'people's democracy' and 'bourgeois democracy', and alternatively as 'totalitarian democracy' and 'liberal democracy'. Some have seen democracy in terms of certain social and economic objectives such as a highly egalitarian society; others as dependent upon the assertion of certain moral values; and others, again, as a decision-making system, albeit operating with guaranteed freedoms of expression.

This brief article, however, is not concerned with the differences between
political theorists, but with the methods of political scientists. Differing

---

* Richard Kimber. 'On Democracy", Scandinavian Political Studies, Vol. 12(3), pp. 199-219. page references are to this work.

temperaments, attitudes and experiences doubtless explain why political scientists take different approaches to studying politics. To simplify, one can divide them into two types, those who begin by getting their boots dirty, studying political activity in the field, often a long-winded process; and those who start by enunciating first principles. The one activity is largely empirical, the other abstract and rational. For the purpose of convenience I shall call these two schools 'empiricists' and 'rationalists'. The latter ask 'What does democracy mean?' and examine various countries to refine the concept. The former identify similarities and dissimilarities and enquire 'What are the distinguishing features?' Comparison should lead to further understanding of regimes and differentiation between them. Categorisation may follow. Rationalists and empiricists may even quarrel about the nature of a 'concept'. Longman's dictionary provides two definitions:(1) conceived in the mind; a thought, notion', (2) 'a general idea covering many similar things derived from study of particular instances'. Rationalists begin with abstractions and are likely to accept (1) as a starting-point. Empiricists, who start from particular instances and move to generalisations, will prefer (2).

There is much in Kimber's article that I agree with. Most importantly I accept that much nonsense has been spoken about minority rights as a precondition of democracy. No decision-making process can be limited to unanimous votes without reducing itself to not making any decisions. (This is what happened with the Polish liberum veto.) Majorities' powers to act can only be limited in respect of democratic rights, as Kimber says; the necessary precondition is that the majority does not decide to deprive any minority of its rights of participation and self-expression. But the consequence is that democracies can act unjustly and still remain democracies. it is easy to accept that democracy cannot be quantified because so many aspects of human behaviour are unquantifiable. Thirdly, I agree that voting in elections makes governments accountable to the voters, but not to a great extent. I shall say more about this last point later.

My main purpose in this article is to survey some of the problems Kimber experiences in defining democracy and to argue that problems of categorisation, at least for empirical research, are not nearly as insurmountableor confusing as he appears to think. I write as someone who has not ventured far into theory; but who has been examining democracy for nearly half a century. It is, I suppose, of little relevance that I feel that I fought for it against those who would have suppressed it. In the 1930s and 1940s there was more emotional resonance about the term than there is today (or was until very recent events). I only mention this as a possible explanation of why I believe I begin, like so many empiricists, from intuitive impressions which can reinforce book-learning in making distinctions between democracies and non-democracies. If I could cite two

personal experiences to support this statement. In 1954 I went to Franco Spain for the first time and observed a society, nearer the Ancien Regime than a totalitarian state. Its comatose repressive nature was very evident. The contrast with France where at the first stop returning home on the railway, I saw a crowd of cheminots in a yard behind the station holding a rowdy strike meeting, was remarkable. Similarly, an hour's bus journey in 1984, taking me from Bratislava in Czechoslovakia to Vienna, transported me from an ambience where freedom of expression was visibly absent to one of bustle, light and noisy argument. It is from impressions of this kind that empirical investigation often begins. Clearly the differences between two countries in both cases, if those of degree, were of extreme degree. Doubtless they could be explained by differences of culture, standards of living and national temperament. But even if this were the case it would be necessary to explain why such traits had produced such different political systems in adjacent countries. Visible contrasts give rise to the questions: How do we explain the differences between these regimes? How do we characterise and categorise them?

Categorisation is an investigatory convenience. As few instances of human collective behaviour are exactly similar, an element of approximation almost inevitable. A strictly definitive categorisation may result in each entity having a category of its own. But this is not convenient and hence the practical question of differentiation arises. What similarities and what dissimilarities determine our taxonomy? Dogs can be put in the same category as elephants if we want to separate four-legged warm-blooded mammals from the rest of creation. It does not follow from this that we can find no visible and ascertainable criterion to distinguish them. Nor does it prevent us from producing a categorisation of different breeds of dogs. The practicality of categorisation depends upon our ability to draw boundaries between different types. (With dogs there will be mongrels, but they can be put in a category of their own.) If the limits of the categories are easily ascertainable then there should be a successful outcome to the process. Once the categories are defined meaningful comparison can proceed. If the differences between the types are blurred, and/or not easily ascertainable, then comparisons between categories may not be meaningful. It is from attempts at comparisons that understanding often commences.

Another practical problem may arise where there are many similar features between potential types; but one obvious feature may be disparate. For example, one feature of swans, it was thought, was their 'whiteness'; but when Australian black swans appeared, similar in all other features, it was decided that whiteness was not a definitive feature. Presumably this was a practical decision by zoologists and swan-breeders. On the other hand, the absence of an important feature in another case may exclude it

from a category. How the researcher decides what feature is important,
and what unimportant, is bound to some extent to be subjective.

How do we categorise democracy? The taxonomy of regimes is likely to be more difficult than that of animals. Yet does not Kimber make too much fuss about it? Are the difficulties insurmountable because differentiating features are not visible, ascertainable and easy to define? My answer to the latter question is 'No¹. The empirical political scientist should choose a definition of democracy whose identifying features are not highly abstract and sophisticated. Nor should it be one that rejects the demotic view. A Prague housewife I recently saw on television, when asked what democracy was, replied, 'to be able to say what I like about the government without being arrested. That does not seem to me to be too bad an attempt from someone who had never known democracy in her lifetime. (Of course, the absence of democracy and the presence of oppression may very well make people sensitive to democracy's essentials.) So a political science stipulative definition that distinguishes between democracy and non-democracy should not be too complex and should incorporate, as far as is consistent with analytical rigour, the notion as it is understood in 'ordinary discourse'. It would be absurd if freedom of speech were omitted as an essential feature. All distinguishing features of democracy should be ascertainable by the ordinary processes of research and fieldwork, i.e. the perusal of documents and the observation and interrogation of political actors.

Robert Dahl's 'dimensions' of democratization (Dahl 1971, 2-16) are summarised in two terms 'public contestation' and 'inclusiveness'.¹ If and when a country proceeds to democracy, he said, it will move along both dimensions until it reaches 'polyarchy', a synonym for pluralist democracy. But if 'public contestation', besides meaning freedom of speech and the press, means 'freedom to form and join organisations' (usually called freedom of association), as Dahl implies, then pluralism is enclosed within the framework of public contestation. 'Inclusiveness' means the right to participate in decision-making, including the right to be eligible to be elected to public office. It is true that Dahl refers to these characteristics as elements of 'democratization', not democracy; but he goes on to say that a 'full' polyarchy would have no restrictions on either public contestation or inclusiveness. Polyarchy is not possible without both features being present. Where either is missing (and of course both are missing in many states), polyarchy does not prevail. '. . . polyarchies are regimes that [are] . . . highly inclusive and extensively open to public contestation' (Dahl 1971, 8). It is a relative position: a 100 percent prototype does not exist.

Later Dahl accepts that polyarchy is a feature of 'relatively democratized nation-states' (Dahl 1982, 4). Hence a simple operationally practical model of democracy, now accepted by many academics, exists. Are the problems of applying it to the 161 states in the world great? In my recent article

(Bealey 1987) I used this simple categorisation to identify democratic regimes which had been overthrown. The difficulties in carrying out this exercise were not great. I needed to do some reading and to consult experts on Latin America, Africa and Asia. It was a matter of empirical research. Empirically the division between democracy and non-democracy, using my simple criteria, was fairly easy to establish. (Of course, if one were attempting to sub-categorise non-democracies that would be much more difficult.) Where public contestation is infringed it is commonly infringed severely. The division between democracy and non-democracy proves to be almost black and white. Of the three countries Kimber mentions as being in a 'grey area', Singapore seems clearly not a democracy: opposition politicians are in gaol and publications critical of the government are liable to be censored. Argentina at the moment seems clearly a democracy with freedom of expression, universal suffrage and multi-partyism. The fact that democracy has been overthrown three times in this century by generals' coups does not nullify the model. A country can move from non-democracy to democracy, and back again, that is, from 'black' to 'white' or from 'white' to 'black' without being 'grey'. Finally, Nicaragua needs more fieldwork before it can be categorized, though after the recent election it may settle down as a democracy.

As we have observed, Dahl's requirements for democracy include freedom
form and join organisations'. It is surprising, therefore, to find
Kimber writing (p. 208):

There is no contradiction in principle between the single-party concept and that of democracy.
causes single-party regimes to be largely undemocratic in practice is their failure
to ensure the implementation of all the principles necessary for full democracy.

Unless there is some ambiguity about the phrase 'in principle' this point is valid only where there is a single party because no group has chosen to organise themselves in opposition to the one existing party (which presumablyis power). Hypothetically, this could be so even where there is freedom of association. Indeed, democracy could exist where there was no party at all. Political scientists, however, have reserved the term, singlepartyregime' a situation where only one party is allowed to operate. This is usually in totalitarian countries. It is true that in some developing countries one finds competitive politics within a single-party framework. Tanzania is a one-party state although it has no competitive politics except within the Tanzanian African National Union (T.A.N.U.), the only permittedparty. candidates for election in Tanzania can only stand if they submit to screening by T.A.N.U. which will not allow anyone to stand who opposes its general policies and guiding role (Bienen 1970, 354). Only two candidates are allowed to stand in each constituency and both must have the T.A.N.U. imprimatur. Hence, in spite of an element of electoral

competition, freedom of association does not prevail in Tanzania

and it

cannot qualify by Dahl's simple model. In practice, regimes exist where the single party forbids others to operate. Thus it seems to me there is little difficulty if one uses two criteria of democracy - the presence of public contestation and inclusiveness - in separating democracies from nondemocracies.

The simple categorisation I used for my article, with its two criteria, would not be regarded as adequate by all commentators. One of Kimber's principles is 'upward control', the capacity of citizens to exercise constraint upon their leaders, and to influence them towards expressed objectives. Democratic governments must be accountable to, and responsive to, the demos. In my recent book (Bealey 1988), I also used accountability as an essential additional component of the concept of democracy. Accountable government is, after all, part of the classic theory of democracy and, as I explain (Bealey 1988, 5-6, 45-53), the presence of public contestation and inclusiveness does not universally guarantee accountability of governments. It is, however, a much more difficult criterion to deal with because the degree of accountability in any country is extremely difficult to assess. It is not nearly so obvious as the presence of free speech and free media; it is not measurable as the extent of the voting right; it is not often clear how much democratic politicians' actions are the result of paying attention to and responding to the voters. Nor is it clear how much the voters want to make their rulers accountable.

As I wrote in my book (Bealey 1988, 47):

... the amount of upward control will not only depend on institutional arrangements, but
on the political maturity of the electorate.

All governments will try to escape from voter pressures if the voters, or their representatives, do not try to hold them to account. Periodic elections provide a degree of upward control; but how much is very difficult to estimate. The volume and complexity of modern policy-making may allow politicians to take refuge in obfuscation. They mislead the voters about who is to blame or to praise for economic failure or success and, as few voters understand economic policy, or even remember the history of it, they find attribution of praise or blame difficult. Governing politicians, anyway, may know less about policy than they ought because policy may be drawn up by unelected technocrats. Between elections governments may respond more to pressure groups than to what they perceive majority opinion to be. Finally, some of the time-honoured mechanisms of accountability, as British Parliamentary questions (Chester & Bowring 1962) have been revealed as of little value as instruments of upward control. Perhaps the best that one can hope for is that:

. . . upward control must include the chance to query minor decisions post hoc and to debate
major issues before policy is finalised (Bealey 1988, 272).

One hundred percent accountability, or even perhaps 50 percent, cannot be envisaged. Accountability or upward control is, therefore, a relative notion which is not only impossible to quantify, but difficult to assess. It is not, therefore, a useful guide for empirical categorisation.

This explains why I did not use accountability as part of my model in my article (Bealey 1987), where I was examining the causes of the breakdown of democracy. The other two criteria were much more ascertainable in both presence and degree. Moreover, it is only occasionally that there is no accountability of government where both public contestation and inclusiveness exist. Upward control without them would be well-nigh impossible. the other hand, in my book I was concerned with demonstrating how much many aspects of the contemporary democratic state affect the control citizens may exercise over their governments. The use of two different models will strike most researchers as quite legitimate as long as what is stipulated in different instances is made quite clear. Of course, this flexibility may not find much favour from rationalists. Yet the sort of conceptual typology Cohen proposes, with its 25 points (Cohen 1973, part 4), rightly criticised by Kimber (p. 214), would be quite impractical as a basis for empirical operation; and even the latter's simpler, and therefore more realistic, scheme (p. 213) raises several problems. For example, the stipulation that 'there must be no restrictions on the placing of items on to the political agenda' presents a formidable difficulty for the analyst of political activity. What is the 'political agenda'? Where is it? In fact, the term is a metaphor, and a misleading one, unfortunately more and more in use. Agendas are formalised things - written lists of items to be discussed. With regard to the so-called 'political agenda' one never knows at what stage it comes into existence. Does one in Britain place an item on it by writing a letter to The Times; discussing the matter with someone on a train; mentioning it to an MP, etc., etc.? If it can be done merely by asserting an opinion in public, then where the right of free speech exists there will be no problem about placing an item on the political agenda. If that is the case, however, freedom to place items on the 'political agenda' does not need to be a separate feature of democracy.

Another outcome of using complex models to categorise states is that it is likely there will be no state in accordance with all the features and no state that does not possess one of the features. It is likely that such a model will reveal the truth that every state has something unique about it, even as every dog does. No Alsatian is exactly the same as another Alsatian. Is this a sufficient argument to persuade us that using categories is a waste of time? One cannot answer the question unless one returns to an earlier question. Why and how do we categorise? The answer is that we categorise

for convenience of description, understanding and explanation, and that we categorise when members of the entity under examination have similaritiesin which distinguishes them from members of another entity who also have features in common, though features different from the first group. So it is with states. The process by which we discriminate between groups is bound to have some degree of approximation; of course, the less the better. The easier it is to draw a clear distinction, or boundary line, between categories the more confident we can be about making statements about them. But when we deal with political science data there will only be rare occasions when we can define categories which theorists might call 'analytically useful'. 'Empirical usefulness' (if that is the correct term), is all we can normally hope for. As I have already said, an element of approximation will creep in.

Kimber, in criticising Sartori's attempt to distinguish what 'a democracy'
is (p. 217) concludes:

If the same feature is imperfectly present in one system and imperfectly absent in another,
it is hard to see how that feature can form the basis of an analytically useful categorisation.

The implication seems to be that without having 100 percent presence or absence a feature cannot be useful for analysis. If this is the case empiricists must admit that their work lacks the academic rigour Kimber demands. One hundred percent freedom of speech is lacking in both Albania and Britain; in the latter because the government has taken powers to enable it to impose limits on TV reporting in Northern Ireland. It would not be enough to rate the UK at 98 percent and Albania at 2 percent (approximating The basis of Kimber's position (and I apologise if I have misunderstood it) seems to be that one cannot be analytically rigorous about a sociopolitical feature that one cannot measure, unless one can identify either its complete presence or its complete absence. Hence it is not possible to be analytically rigorous about freedom of speech. It is not a concept that can be used as an identifying criterion for democracy. So the contrast between the freedom of self-expression that Albania almost completely lacks and its enormously wider, if not perfect, practice in Britain, is of no analytic use. Indeed, Kimber goes on to say that discussion about states as 'democracies' is 'founded upon a mistake of regarding democracy as an entity' (p. 217).

But does the fact that one cannot be analytically rigorous imply that one cannot say anything useful or meaningful? Should we abandon the notion of different types of 'regimes' because every country has a uniqueness that puts it in a category of its own? If political scientists do, they will be giving analytic rigour precedence over the effort to understand the needs, attitudes, aspirations and vocabulary of the vast mass of normally unpoliticalpeople. people of Eastern Europe in 1989, mostly without

either

either experience of empirical research or any appreciation of analytic rigour, seemed to know what: they wanted. 'Free elections' and 'Democracy' they shouted. They knew what they had lacked for 40 years or more. Democracy was a recognisable entity to many of them.

Political scientists when they compile their dictionaries of terms would be unwise to omit 'democracy'. This is not to dismiss the importance of rigour in analysis; but it is to accept that the subject matter of political science is derived from the political activities of human beings. A term which has great significance for them, even if they are unaware of all its implications, cannot lightly be discarded. For many it expresses the nature of the 'relationship of the individual' to the process of placing 'restrictions upon the actions of the authorities' (p. 204). Empiricists, the 'poor, bloody infantrymen' of our discipline, aware of the unsatisfactory quantitativeness of much of our data, and the inadequacies of much of our terminology, derived as it is from common discourse, must nevertheless soldier on through the snipers in order to understand the vast differences between political systems.

---

NOTES

1. These are the collations of eight "requirements for democracy'. Dahl does also speak of 'a third dimension"; but it is not clear what it is.

REFERENCES

Bealey, F. W. 1987. 'Stability and Crisis: Fears about Threats to Democracy", European
Journal of Political Research 15, 687-715.

Bealey, F. W. 1988. Democracy in the Contemporary State. Oxford: Oxford University Press.
Bienen, H. 1970. Tanzania: Party Transformation and Economic Development. Princeton:

Princeton University Press.

Chester, D. N. & Bowring, N. 1962. Questions in Parliament. London: Clarendon Press.
Cohen, C. 1973. Democracy. New York: Free Press.

Dahl, R. A. 1971. Polyarchy. New Haven: Yale University Press.

Dahl, R. A. 1982. Dilemmas of Pluralist Democracy: Autonomy vs. Control. New Haven:
Yale University Press.

Kimber, R. 1989. 'On Democracy', Scandinavian Political Studies 12(3), 199-219.