A large descriptive and conjectural literature has evolved during the past decade suggesting that rapid urbanization—conceived primarily as a massive demographic shift from country to city—constitutes an important agent of political instability in much of Latin America, favoring the growth of alienation and radicalism. This effect is frequently attributed to the frustration of migrant expectations for economic improvement and social mobility in the large cities, or to the processes of personal and social disorganization allegedly inherent in the migration experience. Although field researchers have long suspected the inadequacy of such theory, derived largely from the urbanization experiences of advanced Western nations, for explaining political instability in Latin America, there have been few systematic attempts at testing the various propositions and unconfirmed generalizations advanced by the “urban instability-crisis-and-chaos” theorists. This paper attempts to explore, if not to test, some of the empirical implications of the general theoretical conceptions of the urbanizing process and its socio-political consequences as developed in recent social science literature on Latin America. Mexico is selected for analysis both by virtue of its extremely rapid rate of rural-urban migration in recent years and because of the opportunities afforded by the availability for Mexico of detailed survey data from a number of independent sources to examine systematically a wide range of theoretically relevant variables and relationships posited in the urbanization literature. While in certain areas (as indicated below) the particular nature of the Mexican political system and developmental pattern may render these findings imperfect as predictors for other parts of Latin America, the analysis presented here should serve to identify the major inadequacies of existing theory and illustrate the need for new conceptual models which can encompass both stabilizing and destabilizing concomitants of rapid urban concentration in a developing nation.

Political Science, like Leo Durocher, is not distinguished by an excessive concern with losers.

Although the reasons are not difficult to fathom, theories of recruitment and representation which omit systematic examination of those who are rejected, in addition to, those selected, will only imperfectly comprehend the political conditions and consequences this process implies.

This study, part of a broader cross-national examination of legislative recruitment, focuses on a largely-ignored but perennial figure among the ranks of American political losers: those candidates who challenge incumbents for seats in the House of Representatives.

Few elective institutions in American political life have achieved the degree of stability which since the depression has characterized membership in the House. At no time since 1932 has the percentage of first-term members exceeded 26 percent and in the last 15 years, it has been rare when more than 15 percent of those incumbents who actively sought reelection were defeated. The 1964 election was considered a striking anomaly when “only” 80 percent of the membership returned to the 89th Congress. It is not an exaggeration to concede that, collectively, congressional challengers stand somewhere between the Harold Stassens and Norman Thomases of campaign history in their impressive disregard for the first objective of American major-party politics: winning.

The analyst of campaign strategy in American elections is usually confronted by the apparent impossibility of demonstrating that campaign behavior actually matters. Although candidates possess numerous options in strategy and tactics, it is clear that incumbent members of the majority party (commonly joint phenomena) usually win and opposition candidates of the minority party usually lose. Lewis Froman has even argued that “ … the behavior of the candidates during the campaign is one of the least influential factors in determining electoral outcomes.

The effects, if any, of campaign behavior are the residue to be examined after the partisan distribution of the electorate, the nature of the times, the power of incumbency and any number of variables beyond the ready control of the candidates themselves have been employed to explain or predict the popular vote distribution. The perceived general ineffectiveness of candidate strategy and tactics is usually explained by the low public saliency of the typical election contest, in which most voters are unlikely to be aware even of the names of the candidates, much less their policy positions or, presumably, any other of their behaviors.

But this circumstance may change when “ordinary” election contests are connected to contests of much higher saliency, especially that for the Presidency. It is at this time, at least, when the strategic decisions and behaviors of certain classes of candidates may attract the attention of voters otherwise preoccupied with that more dramatic battle for power.

The rapidly expanding research literature in comparative state politics typically measures its dependent and independent variables with references to means, averages, and per capitas for each state. (For example, environmental inputs may be expressed as per capita personal income, percent of population living in urban places, median school year completed by the population over 25, etc.; political characteristics may be expressed as average voter turnout levels, percent of the total vote cast for the winning party, etc.; policy outputs are often expressed as per capita expenditures for education, average monthly payments for old age assistance, per capita tax revenues, etc.) With measures such as these, the comparative state politics research has systematically explored many of the linkages between environmental inputs, political system characteristics, and public policy outcomes.

Perhaps the most serious reservation regarding this research is its failure to examine distributive and redistributive aspects of state politics. Both dependent and independent variables are generally expressed as levels or amounts or averages for whole states; these can be neatly arranged for comparative analysis both longitudinally and cross-sectionally. But what about the distribution of wealth within a state? Or the distribution of public monies among high and low income groups, rural and urban populations, or other divisions within a state's population. The linkages between the distribution of resources within states, the distribution of influence within state political systems, and public policies reflecting distributional decisions, remain largely untested.

The purpose here is to explore certain aspects of the philosophy of science which have serious implications both for the practice of social and political science and for understanding that practice. The current relationship between social science and the philosophy of science (or the philosophy of the social sciences) is a curious one. Despite the emergence of a considerable body of literature in philosophy which is pertinent to the methodological problems of social science, there has been a lack of osteusive ties between the two areas. A justified concern with the independence of social scientific research has contributed to a tendency toward isolation which is unfortunate in view of the proliferation of philosophical problems which necessarily attends the rapid expansion of any empirical discipline. Although in the literature of contemporary social science there are frequent references to certain works in the philosophy of science and to philosophical issues relating to methodology, these are most often in the context of bald pronouncements and shibboleths relating to the nature of science, its goals, and the character of its reasoning. But what is most disturbing about the fact that social scientists have little direct and thorough acquaintance with the philosophy of science is not merely that there has been a failure to carefully examine the many logical and epistemological assumptions which are implicit in social scientific inquiry, since this task might normally and properly be considered to be within the province of the philospher of science.

In an article published in the December 30, 1967 issue of the Dutch weekly magazine Vrij Nederland, Mr. Frans Grosfeld suggested the following modification of the procedure of parliamentary representation of political parties. Instead of the conventional method of allocating to each party a number of representatives proportional to the number of votes cast for the party, he proposed to allocate a number of representatives which is proportional to the square of the number of votes. Thus, when there are n parties and when p1, …, pn are the proportions of the total number of votes obtained by these parties, the idea is to allocate a fraction qi of the representatives to the ith party determined by

Mr. Grosfeld made this suggestion for the Dutch political system, because in his view n is so large and even the largest pi belongs to a minority party, so that on many occasions several months of negotiations were necessary to obtain a coalition cabinet of one form or another. By squaring the pi's one obtains much larger relative differences between the larger and the smaller parties, and one party or a pair of politically related parties then has a much better chance of obtaining a parliamentary majority. Also, the system (1.1) provides an incentive for parties to merge, which contributes to a reduction of the degree to which the country is politically divided. This system should be contrasted with another system which is applied on a rather large scale, viz., that of declining any seat to those parties whose pi is less than a pre-assigned fraction θ (for example, θ = .03).

Recent comparative, quantitative studies of state politics conclude that party competition and other political variables have little or no impact on important state policies such as per pupil expenditure, old age assistance, unemployment compensation, and aid to dependent children. These are rather unexpected and disturbing conclusions for they disconfirm relationships predicted by some of the most important theoretical formulations concerning democratic politics. Before re-examining the bases for these conclusions, a review of theory is in order.

V. O. Key, Jr., set the context for examining the effects of political variables on state policies. He stressed the importance of two-party competition, or bi-factionalism in one-party states, as a determinant of policy. Key sees the degree of party competition as crucial because it reflects the extent to which politics is organized or unorganized. Party competition by producing some semblance of an organized politics lessens the difficulty of lower status groups in sorting out political actors and issues, thereby enabling them to promote their own interests more effectively. Since state social welfare policies are undoubtedly relevant to the interests of “have-nots,” we can utilize state expenditures in this area as a measure of the success that these groups have enjoyed. Key's formulation, then, would lead to a simple two variable model:

P = e1

S = k1P + ez

Where P is inter-party competition, S is expenditure on social welfare, and e1 and ez represent error or variables left out of the system, and k1 is a constant.

Students of Latin American political parties have long assumed a strong correlation between social class and party identification, yet this assumption has never been tested empirically in any of the Latin American nations. This is probably due in large part to the lack of reliable survey data; however, even the mass of aggregate voting data has seldom been analyzed systematically. As a result, most of what we know—or think we know—about voting behavior in Latin America is based upon the intuitive assumptions of North American scholars. “If I were a member of the Chilean middle class, I would probably vote for the Radicals or Christian Democrats, but on the other hand if I were quite wealthy I would probably vote for the Conservatives.”

Students of Argentine politics assume that the Conservative parties, always representing the interests of the nation's aristocracy, have received the bulk of their votes from the large landowners; that the interests of this group consistently have been opposed by the Radical parties who receive their electoral support from the urban middle classes; and that the more recently formed Peronist parties have championed the cause of, and been supported at the polls by, the industrial workers. It is the purpose of this article to test these assumptions, primarily through analysis of aggregate voting data, but also by examining the social backgrounds of party leaders and their actions while in power.

Professor Gunnell seems to be arguing that if political scientists follow the deductivist path they will almost certainly fall into a bottomless and sterile abyss. My understanding of the deductivist conception of science, and my verstehen about social phenomena lead me to believe that this is a very real risk. The risk derives from the apparent complexity of social phenomena and the relatively limited ability of the human mind to deal with complexity in a rigorous deductive manner. It is just possible that social phenomena are sufficiently complex that they will never be manageable within the neat framework of deductive models. It is precisely for this reason that, at least at this point in time, I would prefer not to have the whole of the discipline tread this path. But why should even one man risk a lifetime of scholarly endeavor against such a possible outcome? Well, if there is another possible outcome to which he attaches a sufficiently high utility, he will take the gamble. I suggest that there is another possible outcome, and hope that, for some, its utility will be sufficiently high to warrant the gamble. That outcome consists in the production of a cumulatively reliable body of knowledge about those social phenomena called political. Consider the worth of being able to reduce the errors of calculation on the parts of voters, office seekers, secretaries of state, etc.

A question often posed by students of American state politics is: “Do state political systems leave a distinctive imprint on patterns of public policy?” Prior to recent years the nearly automatic response of political scientists was an unqualified “yes.” More recent research has led to a qualified but increasingly confident “no.”

Several recent publications have explored relationships between various indices of state politics, socio-economic characteristics, and public policy. The general conclusion has been that central features of the political system such as electoral and institutional circumstances do not explain much of the variation in policy. There are occasionally high correlations between individual measures of voter turnout, party competitiveness, or the character of state legislatures and some aspects of governmental spending. But these political-policy correlations seem to disappear when the effect of socioeconomic development is controlled.

These are disturbing findings. They have not gone unchallenged. But the challenges, rather than reassuring those who have asserted the relevance of parties, voting patterns, and government structures, have demonstrated that the burden of proof now rests on those who hypothesize a politics-policy relationship. The problem has not been resolved.

Part of the problem may rest on the conceptualization and measurement of the central variables. Electoral balance or alternation in office is not “inter-party competition,” except in the most mechanical sense. Compare Massachusetts' loose-knit party structure to the centralization of Connecticut's. “Party competition” is not the same as “party organization.” And party competition, voting habits, and patterns of apportionment fall far short of being equivalents of “political systems.”

It is impossible to tender a reply to Professor Gunnell's essay, “Deduction, Explanation and Social Scientific Inquiry,” that would be both brief and adequate. It would be impossible to be brief because Gunnell conjures up a tagraggery of issues, none of which he seems prepared and/or disposed to resolve. But no matter how extensive a reply might be, it would still be impossible to conceive it as adequate for I am not sure that I, or anyone else, can determine precisely what he means to say in the essay before us. It is impossible for me to determine with any specificity whatsoever, for example, what it means to say:

Logical empiricism as an approach to the philosophy of science has been concerned with developing formal representations or reconstructions of the logical structure of scientific explanation and with a meta-logical analysis of the language applied to science. In this view there is a very strict correlation between the philosophy of science and formal logic.

I haven't the foggiest notion what a “metalogical analysis of the language applied to science” might mean. I had always understood “metalogic” to refer to discourse employing logic as its object language—just as metamathematics would be a language employed in the analysis of mathematics as an object language.

We are now in the midst of a notable revival of interest in the politics of the American states. During the last decade many studies have been conducted of the social, political and economic determinants of state policy outcomes. Several of these writers have argued that the relative wealth of a state, its degree of industrialization, and other measures of social and economic development are more important in explaining its level of expenditures than such political factors as the form of legislative apportionment, the amount of party competition, or the degree of voter participation. It has been claimed that such factors as the level of personal income or the size of the urban population are responsible both for the degree of participation and party competition in a state, and the nature of the system's policy outputs. By making this argument these writers have called into question the concepts of representation and theories of party and group conflict which, in one form or another, are the foundations for much of American political science.

There is a growing awareness, however, that levels of expenditure alone are not an adequate measure of public policy outcomes. Sharkansky has shown, for example, that levels of expenditure and levels of actual service are seldom correlated; presumably, some states are able to reach given service levels with much less expenditure than others. Besides establishing the appropriate level of expenditure for a program, policy makers must also decide about the program's relative scope, provisions for appeal from administrative orders, eligibility requirements, the composition of regulatory boards and commissions, and many other matters which have little to do with money.

Since both Mr. Goldberg and Mr. Gregor have limited their responses principally to my specific criticisms of the deductive model, it would seem appropriate to emphasize once more the purpose of this criticism. The issue which I wish to raise, using the discussion of the deductive model as a vehicle, goes far beyond the model itself and even the approach to the philosophy of science which it represents. It should be apparent that there may be little chance of resolving the differences that stand between the respondents and myself regarding either the character of the philosophy of science as a discipline of inquiry or its product, but to some extent these differences can be separated from the more general problem of the relationship between social science and the philosophy of science.

I suggested that there is a significant intellectual lag or gulf between political science and contemporary work in the philosophy of science. This situation need not, in principle, be viewed as odd or even undesirable since the distance between philosophy and other fields of science is at least as great. What is unique about political science is that, although it has lost touch with philosophy, it has to a large extent derived its conception of science and its notion of the procedural rules of empirical inquiry from a restricted body of literature in the philosophy of science, i.e., traditional logical empiricism.