Navigation
Papers by Melberg
Elster Page
Ph.D work

About this web
Why?
Who am I?
Recommended
Statistics
Mail me
Subscribe
Search papers
List of titles only
Categorised titles

General Themes
Ph.D. in progress
Economics
Russia
Political Theory
Statistics/Econometrics
Various papers

The Questions
Ph.D Work
Introduction
Cost-Benefit
Statistical Problems
Social Interaction
Centralization vs. Decentralization

Economics
Define economics!
Models, Formalism
Fluctuations, Crisis
Psychology

Statistics
Econometrics

Review of textbooks

Belief formation
Inifinite regress
Rationality

Russia
Collapse of Communism
Political Culture
Reviews

Political Science
State Intervention
Justice/Rights/Paternalism
Nationalism/Ethnic Violence

Various
Yearly reviews

Philosophy
Explanation=?
Methodology

 

[Note for bibliographic reference: Melberg, Hans O. (1997), Culture - An explanatory variable or an interpretive approach?, http://www.oocities.org/hmelberg/papers/970623.htm]

 

Culture - An explanatory variable or an interpretive approach?
A Review of Stephen Welch (1993) The Concept of Political Culture

by Hans O. Melberg

1. Introduction
The content of Stephen Welch's book The Concept of Political Culture can be summarised in the following sentence: There are many problems with the existing uses of the concept of political culture, but these can be solved if we use a phenomenological approach. Thus, we are faced with two questions. First, what are the problems with the existing use of the concept of political culture? Second, what is the phenomenological approach to political culture. Third, in what way is the phenomenological approach better than the other approaches to political culture?

I shall argue that Welch's answer to the first question is quite well done, although he exaggerates the problems with the existing uses of the concept of political culture. In essence, he interprets problems of degree as qualitative proofs against an approach. His answer to the second question is more confusing, but a reconstruction reveals some valuable ideas. Most importantly, the idea that culture consists of a set of socially constructed meanings. Finally, the third question is less than convincing as is demonstrated by his constant appeals to vague criteria such as fruitfulness and the integrative power of the phenomenological approach. I shall argue that the interpretative approach makes most sense if it is viewed as explaining a sub-class of beliefs which, in turn, are useful when we use culture as a variable to explain behaviour. Thus, it does not constitute an alternative use or better use of the concept of political culture. Before embarking on this critique, it is necessary to discuss the concepts involved.

2. The concepts
Welch identifies two main broad uses of the concept of political culture. The first, which he calls the behavioural approach, is characterized by the use of culture as an explanatory variable. For example, some authors use the presence or absence of a democratic political culture to explain the degree to which a country has a well-functioning democracy. [1] Another example is the use of culture to explain the stability of political systems; the theory being that a gap between the culture of the population and the workings of the system is bound to lead to collapse.[2] The second use of the concept of political culture, which Welch names the interpretative use of political culture, is to view it as an approach to politics - a way of organising data and suggesting new lines of research without trying to use culture as an explanatory variable. An example of this approach, is Robert Tucker's writings on Stalinism. [3]

I have two arguments against the distinction between the behavioural and the interpretative approach. First, the two uses of the concept are not mutually exclusive. To use culture as an way of sorting your data does not exclude the use of culture as an explanatory variable. Welch's answer to this is that "... the lack of 'mutually exclusive and jointly exhaustive' categories help us understand the domain we are studying. The latter is to be considered the purpose of a 'useful' criterion of distinction." (p. 2). This is certainly contrary to conventional wisdom which holds that the most useful (in terms of advancing understanding) classifications produce mutually exclusive and jointly exhaustive categories. However, I will not split hairs over this issue which is not a substantive point. Welch is only using his classification as a structural device - a way of organising his book.

A second line of attack, is to enquire into the precise meaning of the interpretative approach. A good starting point for doing so the following often cited argument by R. Tucker:

"Might not the central value of the concept like that of political culture be that it assists us to take our bearings in the study of the political life of a society, to focus on what is happening or not happening, to describe and analyse and order many significant data, and to raise fruitful questions for thought and research - without explaining anything?" 4

It seems clear that Tucker is suggesting to use the concept of culture in a different way that the behavioural-explanatory project, however I think this precise nature of this alternative is unclear.

Let me first examine the one of Tucker's suggestion - that of using culture to order our data. There are many ways of organising data, such as chronologically, alphabetically, or thematically. The nature and usefulness of these sorting mechanisms are often obvious. We all understand what it means that names and numbers in a phone-book are listed alphabetically. Moreover, it is immediately obvious that an alphabetical listing is more useful than a catalogue in which the numbers are listed numerically, starting with the smallest phone number. However, it is much less clear what it means to sort historical and political data according to a cultural criterion.

Maybe we could get a better understanding of the interpretative approach by examining the its application in the study of Stalinism. According to Welch there are

"Two arguments in favour of seeing Stalinism as a culture. One is an argument in terms of continuity or revival of culture, mediated through Stalin's personality. The other places emphasis on the idea of a cultural revolution, and draws attention to the comprehensiveness of both the aims of that revolution and its implementation under Stalin ... The strength of the second, more consistently interpretative argument is its integrative power. By integrating the manifold way ways in which and massive extent to which Stalinism penetrated society, the interpretation offers us what appears to be an enriched understanding of the phenomenon." (p. 91)

Ignore, for the moment, the appeal to the rather vague criteria of fruitfulness, integrative power and enriched understanding. The importance of the quotation for the current discussion, is Welch's argument that Stalinism represents a comprehensive attempt to change the culture of a society. A book about Stalinism could then be structured by presenting how Stalin tried to change the different elements in the Russian culture - one chapter on the change in work ethic, one chapter on the up-bringing of children and so forth. This is, at least, one possible way of organising your material.

Yet, surely the interpretative approach implies more than a method of classifying your data. Indeed, Tucker writes that "within political culture research, the defining feature of interpretivism is a conception of political culture as the 'meaning' of political life, or the meaningful aspect of politics" (p. 5). However, it is not immediately clear what is meant by the term "the meaning of political life."

A third possible defining feature of the interpretative approach is Welch argument that "The only common feature of interpretative methodology is ... its distrust of quantitative analysis" (p. 5). At least this is a relatively clear demarcation, but it is also a position with which I disagree. If the aim is to describe the features of a culture, it seems to me a that we need a wealth of evidence, not a single in-depth analysis of one feature such as the Balinese cockfight. To focus on one in-depth case-study is to run a very high risk of committing the fallacy of composition - of making the invalid inference from one case to all cases. It is also, as Welch notes, to run a high risk of over-interpreting a single feature - of finding "deep" meanings where no such meanings exist.

Welch presents several argument against quantitative analysis. One of these is Charles Taylor's notion of inter-subjective and common meanings. Once again I have great problems in understanding exactly what this phrase means and how it prevents quantitative analysis. To be concrete: Suppose you want to investigate the attitudes of children in missionary schools. Assume that a survey reveals that these children have almost the same attitudes as children who are brought up in a Christian environment in your own country. Taylor's argument is then that this similarity is deceptive since the attitudes of the children the missionary school are not "rooted in social practice." For instance, both group of children may respond in a questionnaire that stealing is wrong. However, this similarity could conceal a difference since the two groups might disagree on the definition of stealing. Taking a tool from another house might be stealing in one society, while it is not stealing in another since the 'social practice' of that society is that some tools are collectively owned. Thus, similar attitudes as revealed through a survey do not imply cultural similarity. [5]

While I agree that surveys are imperfect at revealing cultures, I think Welch and Taylor exaggerate this problem. [6] First, there is no law against being sensitive to context when you are doing surveys. What is wrong with the common-sense idea that it is the combination of quantitative data and qualitative information (sensitivity to context) that produces the most reliable conclusions? Second, the problem is a question of degrees - not an absolute proof that surveys in general are meaningless. Some questions are highly context-dependent, others only weakly so. Third, additional questions could be used in an attempt to assess the importance of the context - for example by asking questions about the definition of stealing. Fourth, we can ask questions about behaviour in concrete situations, not only about attitudes towards highly subjective concepts (such as democracy). For instance, we may ask whether a person would charge interest rates on a loan to a friend.[7] We may even devise experiments in which we observe actual behaviour with real rewards (for example, bargaining experiments).[8] While none of these proposals solve the problem of different contexts, they at least reduce its damaging consequences.

In sum, I am still uncertain about what exactly is meant by the interpretative approach, and the meanings I have managed to extract - using culture as a way of sorting your data; viewing, viewing culture as the meaning of political life; only focusing on qualitative research - appear respectively as a use different from the essence of the interpretative approach, vague and wrong.

3. Welch's criticism of existing uses of the concept of political culture
After presenting the behavioural and the interpretative approaches, Welch goes on to distinguish between two different uses of the concept of political culture within the two main groups. Within the behavioural approach, he identifies two major uses of political culture: A comparative and a sociological. The comparative use "amounts to an attempted explanation of the presence of stable democracy in some countries and its absence in others in terms of pre-existing political cultural conditions" (p.15). "The sociological project", on the other hand, "consists in an investigation of the social conditions under which democracy functions" (p. 15).

Welch then makes quite a lot out of the argument that there is a conflict between the comparative and the sociological approach. As he writes

"The comparative aims of the study [Almond and Verba] entail a relative blindness to sub-national cultural variation. Yet, its attempt to produce a 'scientific theory of democracy' yields a sociological theory of a degree of richness and complexity that not only is comparison prohibited, but an ample resource is offered for use by researchers not similarly constrained by the comparative goal ..." (p. 29).

Once again I have to admit that I find this difficult to understand. Why should the comparative project necessarily lead to blindness to sub-national variations? In any case, the problem does not seem to be one of inconsistency or mutual excludability. Both projects are legitimate, and doing one does not imply that the other cannot be done. The problem is rather of a practical sort: A complex theory of democracy would entail a lot of variables, and there with such a theory there are not enough data make reliable comparisons between countries. As Welch writes in a comment on Almond and Verba: "... comparisons among only five countries would be unlikely to be compelling" (p. 16). This is true, but there is nothing which prevents us from using data for many more countries. Furthermore, the data set can be greatly enlarged if we also include different historical periods, and different geographic areas within one countries. For example, Robert Putnam, in his book Making Democracy Work, used culture to explain variations in the effectiveness of (local) democracy in different Italian regions. Hence, I think Welch's emphasis on the difference and conflict between the sociological and the comparative project is exaggerated: It is a solvable practical problem, not a theoretical and inherent inconsistency.

Within the interpretative approach Welch distinguishes between the method of thick description and the phenomenological approach, although the two are not strictly separable. Instead, Welch writes that the interpretative approach "contains an idealist tendency, as well as a phenomenological potential" (p. 9). Regardless of the vagueness of the distinction, we need to examine the definitions of both.

The characteristic of a thick description, is its emphasis on elaborating the meaning of a culture. For instance, C. Geertz analysed the Balinese cockfight in order to decipher its meaning - which is then interpreted as saying something about the Balinese culture in general. A phenomenological approach, as developed by A. Schutz, is defined by its insistence on the social construction of meaning. For example, the meaning of Polish identity emerged as the aggregate result of processes at the level of the individual (especially negation i.e. the meaning of "Polishness" came to be defined in opposition to German identity during Bismarck's Kulturkampf: Being Polish was the opposite of being German).

Thus, we now have four categories: Political culture can be used as an explanatory variable (to compare countries or in a one-culture sociological investigation), or it can be used to interpret (thick vs. phenomenological interpretation). After making these distinctions, Welch presents six problems with the two behavioural uses of political culture, and one problem with the thick-interpretative approach. This part of the book is maybe the most useful, although I will argue that his criticism is exaggerated.

Welch believes the comparative project is made difficult by three problems: complexity, salience, and indexicality (p. 75). Complexity refers to the problem of using political culture to explain differences in outcomes across nations when there are many variables at work and a limited amount of data available to distinguish between the effects of the variables. Salience means that differences in attitudes do not have the same meaning in different countries. A low response to the question of whether you have any influence on the laws of your country (i.e Almond and Verba's subject-competence dimension), does not necessarily imply a un-democratic and fatalistic attitude, since the person may compensate for the inactivity in the rule-producing phase, with active manipulation in the rule-application phase (using informal channels of influence - connections and corruption). Finally, indexicality refers to the problems which arises because historical symbols have many possible interpretations. Thus, the popularity of Pilsudski in Poland can be both interpreted both as a sign of a patriotic culture, and an anti-democratic culture - Pilsudski is both the hero of the war against Russia (1920), and the villain who was responsible for the anti-democratic coup in 1926. In sum, there are many problems with the comparative use of political culture which makes it difficult to use culture to provide reliable explanations of cross-national differences.

The problems with the sociological use of political culture overlaps to some extent with the comparative critique, and we can discuss three additional problems under the following headings: cultural lags, retreating cause, and retreating effect. Cultural lags refer to the fact that there is a time difference between changes in culture and changes in the political structure. For example, values taught through socialization are not put into effect until adulthood. Thus, a change in culture is not immediately reflected in a change in the workings of the political system (the growth of democratic attitudes does not immediately tramsform a dictatorship into a democracy), and changes in the political structure are not immediately reflected in a change in culture (democracy does not immediately produce democratic attitudes). This creates a problem because some theories of modernisation become unfalsifiable. These theories argue that the universal process of industrialisation creates a certain political culture which, in turn, will make the political systems in the world become more similar - the convergence thesis. Any exceptions to this patters is simply explained by cultural lags - and it is always just a question of time before the exceptional country will also converge.

The problem of the retreating cause and the retreating effect are summarised by Welch as follows: "In political culture research, we will argue, when investigation begins by specifying the effect, it becomes difficult to discern the cause; and when it begins by specifying the cause, it is hard to discern the effect" (p. 67). An example of the retreating cause, is John Miller's analysis of why the Kazaks had a much higher level of party-membership than the Balts in the former Soviet Union. Starting with this effect, one may try to explain it by differences in culture. However, Miller also shows that the difference could be explained in terms of interests since the Balts had a greater probability of attaining sought-after positions without part-membership than the Kazaks (because of their relative higher level of education). If our strategy is first to eliminate all differences that are explained by interests, then very few differences may be left to be explained by reference to culture. In more general terms, the research strategy a priori reduces the role of culture when we start with an effect and only explain the "unexplained" with reference to culture (Elkins and Simeon's residual).

An example of the retreating effect, is provided by D. Elzar's account of the American political culture (p. 71-72). He starts by identifying three different sub-groups of the American culture: the individualistic, the moralistic, and the traditionalistic. These cultures are supposed to exist in various degrees in different geographic areas depending on the origins of the immigrants who first populated the area. Ira Sharkansky has tried to test the explanatory power of this theory by statistically examining whether 23 variables (indicating the political culture) were significantly different between the various areas. His result was that 15 variables were significantly different, but Welch argues that "The problem lies in ensuring that the differences that contributed to the original descriptions are distinct from the ones held to be the 'effects' of these sub-cultures ... The effects retreat from view as effects in proportion to the fullness of the description of the cause. No amount of mathematical sophistication can account for this deficiency" (p. 72). Hence, starting with an assumed cultural difference, it is difficult to prove the effects of this difference independent of the definition of culture, and this is a problem for explanations based on political culture.

I do not disagree with the argument that these are serious obstacles to the use of cultural variables in producing reliable explanations. However, as argued previously, these are problems of degrees that can be reduced. More data can be gathered to reduce the problem of complexity (see for example R. Inglehart's large scale project) and questions may be phrased so as to reduce the problem of indexicality and salience. The problem of salience and indexicality can also be reduced by focusing on comparable data (see Hofstede's study of people from different cultures who work in the same positions in multi-national companies). The retreating cause - that culture seems to disappear as an explanatory variable if we first seek explanations in terms of interests - is true, but also justifiable if we believe that people generally are rational and selfish. This is not to deny that norms and non-rational behaviour takes place, only that as a first assumption we should work on the hypothesis that people do act in their own (perceived) interests.

The problem of the retreating effect is more damaging: It is very difficult to start with a cultural trait and derive the likely consequences of this trait. However, consider the following quotation in which Welch discusses the consequence of the problem of the retreating effect:

"This outcome [a holistic definition of culture and interpretivism] ... is in fact impossible to avoid; the attempt to avoid it, to continue, in other words, to see political culture as a separable factor in comparative explanation while beginning with an account of cultural differences, is doomed. The more fully cultural differences are specified, the less easy it is to separate them from their putative effect." (p. 71)

The reasoning is false as I will show with an example. Define culture as a set of stable beliefs, norms and values that distinguishes one group of people from another (i.e. a subjective definition). Assume next, a supposed difference in culture, such as the Russians being more authoritarian (e.g. to place a high value on order, and believe in collectivism as a means to create a good society). The proposed effect of this difference, is that Russia will have a more centralised and less democratic political system than America. To prove this argument we have to make two steps. First, to show that the Russian culture really is more authoritarian than the American. Second, to prove the link between an authoritarian culture and the authoritarian system. I agree with Welch that these links are difficult to prove, but I disagree with the argument that this problem is bound to lead to an interpretative and holistic approach to culture. In fact, the problem is caused by the holistic definition of culture.

First, to prove the link we would need some either some historical proof such that every time a people has had authoritarian attitudes there has been an authoritarian state, or we need to show the precise causal mechanisms which leads from authoritarian attitudes to an authoritarian state. Second, the separation of cause and effect - of culture and structure - is simply a question of defining your terms carefully. If we define culture subjectively as values, beliefs, and norms, we avoid making the nature of the state (authoritarian) a part of the culture - and we have a clear separation between cause (authoritarian attitudes) and effect (authoritarian states). Thus, Welch is wrong in suggesting that the retreating effect necessarily must lead to a holistic definition of culture.

Welch also presents a second implication of the problems of the retreating cause and the retreating effect? He writes that:

"If neither cause nor effect can be given primary attention without the other fading from view, the explanation that the two are not in fact distinct presents itself rather forcefully." (p. 72)

Once again I must admit that I think this is plainly wrong. The practical problems we may have in separating cause and effect, can never be taken as an indication that the two are conceptually the same. Causes and effects exist in the real world, regardless of the problems we have in distinguishing between the two in our academic studies. Moreover, I have already argued that a subjective definition of culture enables us to separate cause from effect. Saying that the two are the same is conceptually confused and devoid of meaning.

Turning to the problems with the interpretative use of culture, Welch has fewer objections. The only main problem - he argues - with the interpretative "thick-description" approach, is its idealistic tendency. By this he means the danger of confusing the interpretation which makes the events meaningful for the interpreter, with the interpretation which makes the event meaningful to the participant. For instance, he argues that the concept of atomisation, used by Tucker to understand Stalinism, represents an example of idealism. Atomisation may make Stalinism more understandable or meaningful for Tucker, but it was not the meaning ascribed to the events by the people who living during Stalinism. On the contrary, Welch argues, Stalinism sometimes strengthened the bonds between people, such as when managers and workers had to unite in order to evade the impossible demands made upon them in the five-year plans. Thus, the thick-description is found to be flawed by its idealistic tendency.

In sum, Welch correctly presents some of the major problems of using culture as an explanatory variable. However, he sometimes exaggerates these problems, presenting them as qualitative proofs that make the comparative project "impossible", while they in fact are problems of degrees that can be reduced. Furthermore, he draws the wrong conclusions; The problems do not lead to a holistic definition of culture (in fact, it leads to the opposite), nor do they imply that the explanatory project is "doomed".

4. The phenomenological approach
After discussing all the problems with the existing uses of political culture, the stage is set for Welch's last - and recommended - use of the concept of political culture: the phenomenological version of the interpretative approach.

As mentioned, the phenomenological approach is characterized by its focus on the social construction of meaning. According to Welch "... phenomenology serves to anchor interpretivism to concrete social reality, while at the same time arguing that such reality is a construct needing to be continually reproduced" (p. 91). I have already tried , somewhat unsuccessfully - to examine what Welch means by the interpretative use of the concept of political culture. One of its distinguishing features was then, according to the quote by Tucker, that it did not try to explain anything. However, despite the relatively uncomplicated assertion to the opposite (recall the phrase "without explaining anything"), Welch thinks Tucker did note really mean that his approach was non-explanatory. As he writes: "What Tucker must have in mind is the sort of explanation using political culture that has been criticised in previous chapters; explanation in which political culture appears as a variable." And he continues, "Just what sort of explanation remains possible when this has been ruled out will be exemplified in the following discussion and elaborated in the succeeding chapters." (p. 81, emphasis in the original)

His answer to the question above, is that "the nature of the explanation offered by the interpretative use of political culture is phenomenological" (p. 117). This, of course, only raises the question of exactly what a phenomenological explanation is. One of Welch's answers is implied in the following:

"Phenomenology thus contributes to the methodology of Verstehen or thick description an explanatory component: an explanation of the origins of social meaning, and hence an explanation of the utility of the interpretative method. Essentially, because 'culture', in the form of assemblage of emergent typifications, is a creation of social actors themselves, it becomes of use to the investigator in reaching an understanding of their behaviour." (p. 110)

At this point I begin to understand a little more of what Welch means. A concrete example may help. Assume a person colours his hair green in order to signal his rejection of "everything" in society. Phenomenology is then an approach we can use to explain why colouring your hair green signals a rebellious attitude - to find the social origin of the meaning of green hair! The mechanism giving green hair this meaning, is negation: If "the establishment" has normal hair colours, then non-normal hair colours must be anti-establishment. This, it seems to me, is the type of explanation Welch claims is provided by phenomenology.

Culture, in this view, become a set of meanings - that green hair signals rebellion; that nodding your head means yes; that wearing jeans in a business meeting is not a smart move. Groups of people clearly differ in terms of these meanings - in some cultures nodding your head means no! Political culture, in this view, would then be meanings attached to political symbols - that red is the colour of the socialists; that 1789 has a very specific meaning in French politics; that the phrase 'democratic centralism' has a connotations beyond its literal interpretation in Russia.

5. Evaluation of the phenomenological approach
How should we evaluate Welch's suggestion to focus on the phenomenological uses of the concept of political culture. To answer this it is necessary to engage in a small digression into the philosophy of the social sciences. I shall do so using the works of Jon Elster.

In the social sciences we try to explain facts (e.g. the election of Bill Clinton as president) and events (e.g. the presence of a Republican majority in Congress). [9] To explain an event is "to give an account of why it happened" [10] . From this we can deduce that theories may fail in two respects: theoretically and empirically.[11] A theory fails theoretically if we can prove that it is logically inconsistent, if some of the reasoning involved is invalid, or if it does not derive unique predictions. Empirical failure occurs when the theory is falsified by empirical facts: that some of its empirical premises fail to hold, that it does not conform to historical data, or that its predictions turn out to be wrong. Thus, there are two ways an approach can fail: logically or empirically.

With this basic digression in mind, Welch's praise for the interpretative-phenomenological approach weak. An approach should be praises for being able to explain concrete events, however Welch frequently uses more vague notions to praise his phenomenologica-interpretive alternative. Consider the following terms used either to praise the phenomenological-interpretative approach, or to attack other approaches: It - the phenomenological approach - is an "anchor" (p. 9); it focuses out "attention" on important features - while other approaches "ignores" (frequently used) or "entail a relative blindness to" (p. 29) certain factors; it "transcend the dichotomy of culture and social structure" (p. 79) - while other approaches "deny" this duality; it "throws considerable light on" (p. 135) while other approaches are "constrained" (p. 45); it has great "integrative powers" (p. 106); it provides a descriptive "richness" (p. 106); it "helps us to find out feet" (p. 115, also p. 105); it "structures our understanding of real social arrangements" (p. 111); it helps us "more fully to recognize social reality" (p 134); it helps in "placing emphasis" on important processes (p. 135); it "invites" empirical investigation (p. 146-147). In short, "... the phenomenological analysis ... provides a posture that sensitivizes us to these matters [eg. the indexicality of meaning], and alerts us to the source of the dynamism that we may locate in political culture" (p. 164).

What does the rather long list above tell us? First, Welch seldom judge approaches according to their logical or empirical validity. Instead there are tendencies, relative blindness and tensions (not inconsistencies). Second, some of the terms are highly subjective. Consider, for instance, the use of fruitfulness as a criterion to distinguish between good and bad approaches. For some people taking a hot bath may be a good way of getting new ideas! In other words, what is fruitful for you need not be fruitful for me - the criterion of fruitfulness is subjective. That is why I want to stick to the non-subjective criteria of logical consistency and empirical validity. A theory which is logically inconsistent is logically inconsistent for everybody, a theory which is fruitful need not be fruitful for everybody.[12]

Criticising Welch for the use of vague criteria to praise the phenomenological approach, does not imply that it is entirely devoid of positive content. In the following I will argue try to exemplify and argue how phenomenology is most useful within the approach using culture as an explanatory variable i.e. not as an alternative to the approach of "using-culture-a-variable".

Recall the meaning of culture in the interpretative-phenomenological approach as a set of meanings (green hair means rebellion etc.). Knowledge of these kind of meanings is certainly valuable in order to explain behaviour. For example, a Martian who came to earth would need to know these meanings in order to understand why people behave as they do. Why did Lech Walesa have a picture of Pilsudski behind himself when he spoke on television (or why Walesa grew such a large mustach!). Clearly, the answer would have to include a discussion of the "meaning" of Pilsudski as a symbol - a meaning a Martian unread in earthly politics has no way of knowing.

Return, again, to the question of why Walesa used a picture of Pilsudski. It seems to me that the explanation would go something like this: Walesa believed Pilsudski is associated with certain positive connotations for many people in Poland (a belief about the meaning of Pilsudski). Walesa wanted to be associated with this meaning, so he put a picture of Pilsudski behind him when he spoke on television. This is a perfectly fine rational explanation involving preferences and beliefs. There is, however, one particular twist: It involves a belief about a meaning. In a previous essay I developed a definition of culture as a set of distinct and stable preferences (such as a strong preference for justice), beliefs (of different categories: factual, causal and strategic), and norms (e.g. norms of secrecy) (for more on this see chapter 2.3 of The Cultural Approach to Russian History - How reliable?. See also, The information required for perfect prediction). Groups of people may differ from another in one or more of these traits, and I defined this as a cultural difference.[13] After reading Welch's book, I think I need to add another category: beliefs about meanings.

The use of the phenomenological approach is then to explain one of many factors which enter into the explanation of behaviour; Beliefs are one variable of several that explain behaviour; beliefs about meanings is one of many types of beliefs (and differences in beliefs is only one of several types of cultural differences - preferences and norms being the two other). This puts the importance of the phenomenological approach into a more correct perspective. In fact, one might question whether phenomenology really "explains" the meanings that we attach to political symbols. It is more a case of simply pointing out that some meanings are socially constructed, but phenomenology itself does not say exactly how. One such mechanism is, as mentioned, negation: We can "explain" the meaning of "being Polish" by arguing that its is a negation of "being German". However, phenomenology does not explain the meaning, negation does. On this account phenomenology seems a rather obvious (be aware of meanings and how they are constructed) and not very powerful approach. Lastly, the popularity of one mechanism, negation, suffers from the problem of non-uniqueness. The opposite of "I am a Christian" could be "I am a Satanist" or "I am an atheist" or "I am an agnostic." Thus, saying that the some meaning emerged in opposition to another, is more complicated than it might appear at first sight.

6. Minor flaws and virtues
Before I conclude, I want to make a few comments on some minor flaws and virtues. Among the virtues, I would mention the emphasis on methodological individualism; the many examples combined with an extensive overview of the literature; and, finally, the very detailed discussion of the weak points of the use of culture as an explanatory variable. Among the minor flaws I could mention a mistaken interpretation of Daniel Bell (p. 36 - Bell himself later often pointed out that he never meant that all ideologies were dead, only that existing ideologies were dead and that new ideologies would be inspired by the third world). A similar flaw is the argument that Geller's theory of nationalism is "arbitrary" (p. 123). It may be focused on how the elite tried to construct a national consciousness using myths, but the does not automatically imply that the culture created is "arbitrary." Finally, Welch's style is often rather complex. The language is sometimes very abstract and it is difficult to understand what he means (not always, he is also - sometimes - reasonable clear). Maybe the subject mater is complicated, but maybe too the language betrays the origin's of this book as a PhD dissertation - a type of manuscripts in which the author is required dress up the obvious in confusing language.

7. Conclusion
I started this review by arguing that Welch's book could be summarised in one sentence: There are many problems (complexity, salience, indexicality, cultural lags, retreating cause, retreating effects, idealism) with the existing uses of the concept of political culture (comparative-behavioural, sociological-behavioural, thick-interpretative), but these can be solved by a phenomenological-interpretative approach (focus on the social construction of meaning). In this I think Welch is mainly wrong. To be concrete, I think Welch contradicts himself when he writes (and quotes) that:

"Culture, in this view, is therefore merely the extension and elaboration of the normal process of coming to terms with puzzling phenomena: it is 'not a power, something to which social events, behaviours, institutions, or processes can be causally attributed; it is a context, something within which they can be intelligibly - that is, thickly - described'" (p. 105)

To make behaviour intelligible by showing in what sense the act was meaningful to the participant, is precisely to explain it using meaning (culture) as a causally important variable. [14] A Martian would find the importance attached to Brandt's act of kneeling in front of the Warsaw Ghetto very puzzling. An elaboration of the meaning of this act would - in part - explain the act. Together with the meaning of kneeling, we would need to explain the act with reference to an aim (that Brandt wanted to make use of the meeting). Indeed, phenomenology is a way of coming to terms with puzzling behaviour, using culture as a variable. In sum, there is no choice between using culture as a variable and using culture in a phenomenological approach. The phenomenological approach is only a part of the larger project of using culture as an explanatory variable.


ENDNOTES
1 This is the topic of Welch's chapter 1 Political Culture and Democracy.

2 This the topic of a sub-chapter titled The Consonance/Dissonance Theory ....

3 See Welch's chapter 5: Political Culture and Stalinism.

4 Tucker, R. (1973), "Culture, Political Culture and Communist Societies." Political Science Quarterly, vol. 88, p. 179.

5 See Taylor's review of Jon Elster's Logic and Society and Elster's reply for a short exchange on the notion of inter-subjective and common meanings.(Inquiry vol. 23 (1980) no. 2).

6 To be fair, Welch does mention the possibility of reducing some of the problems (by gathering more data, by controlling for other variables etc.), but he more frequently - and somewhat contradictory - writes about the impossibility of overcoming the problems.

7 See, for example, Schiller Robert J., Maxim Boycko and Vladimir Korobov. 1991. "Popular Attitudes Toward Free Markets: The Soviet Union and United States Compared." American Economic Review 81:385-400.

8 See Testing cultural differences: Four experiments for how this could be done. For some actual results, see Bar-Hillel, M. and M. E. Yaari. 1984. "On Dividing Justly." Social Choice and Welfare 1:1-24. See also A. Roth's HomePage for more references.

9 Jon Elster (1989), Nuts and Bolts for the Social Sciences, Cambridge: Cambridge University Press, p. 3.

10 ibid.

11 This distinction is inspired by Elster's article "Some unresolved problems in the theory of rational behaviour", Acta Sociologica (September 1993) vol. 36, no 3, pp. 179-190.

12 I am not denying that fruitfulness might be a criteria. I am only arguing that it is not a very good criteria for judging theories.

13 My views have evolved somewhat since then. I think there is a problem when we say that a true belief is cultural. For example, Russians - more than Americans - believe that rich people are dishonest (see Schiller et all.). Does this reveal a Russian anti-capitalistic culture? Maybe, but if the belief is true 8as it well could be) I would hesitate to call the difference cultural. So far I think the solution to this is to define "cultural beliefs" by their history of creation (and not their content). Cultural beliefs are beliefs that are inherited, they do not arise as the result of a conscious weighting of evidence in favour or against the truth of the belief. (However, this creates a problem when we want to classify self-fulfilling and wishful beliefs).

14 My views are here influenced by Ch. 10, 11 (most important), and 14 in Jon Elster (1989) "Vitenskap go Politick (Science and Politics), Oslo: Universitetsforlaget.


[Note for bibliographic reference: Melberg, Hans O. (1997), Culture - An explanatory variable or an interpretative approach?, http://www.oocities.org/hmelberg/papers/970623.htm]