U.S. AND BRITISH WORKERS
Same Consciousness, Different Opportunities
Much of the accepted wisdom about America’s weak class consciousness is based, sometimes implicitly, often explicitly, on comparisons with the older European societies. Foreign observers, beginning at least with Alexis de Tocqueville in 1835, have filled their commentaries on the United States with glowing accounts of America’s “inherited ideology of classlessness” (Bottomore, 1966:51). The American Creed promised the opportunity to get ahead and thus constructed an image of American society as a “scheme of gradation” (Ossowski, 1963). The inequality that existed was continuous; American workers, unlike European workers, could not see themselves divided into a class separate from their bosses.
Research neither confirmed nor challenged the accepted wisdom; mostly, it ignored the issue. Other facets of the American Creed came under research scrutiny. But even as sociologists gathered the data to question America’s more rapid mobility rates, they continued to endorse without examination the idea that American workers were protected from class consciousness by an “ideological egalitarianism” (Lipset and Bendix, 1967).
Occasionally, someone would remark on the lack of adequate comparative research on class consciousness. Treiman and Terrell (1975), in a study of mobility paths, cite the common belief that class is less salient in the United States than in Europe, but in a foresighted footnote they acknowledge that this belief rests on virtually no rigorous empirical research. Kahl (1957:174) had earlier noted the lack of systematic evidence for this widely held belief. Everybody seems to have overlooked the implications of a 1948 UNESCO study comparing class placements in nine countries (Buchanan and Cantril, 1953).1 In this study, the United States had the third highest level of working-class identification, exceeding such presumably class-conscious countries as France, Germany, Italy, and Norway. Only in Britain and the Netherlands did more people see themselves as working class. This surprising result led Buchanan and Cantril to question the “glib generalities” that America was a middle-class society. They called instead for a revision or, at least, a careful re-examination of the prevailing wisdom.
The research reported in this chapter attempts this re-examination. We will compare the class perceptions studied in Chapter 4 with similar data from Great Britain. The weight of the evidence suggests that there is, in fact, little difference in the way American and British workers perceive class divisions. At least in the separation of a working class from a middle class, the popular U.S. and British definitions are remarkably similar.
What makes the similarity in class perceptions especially interesting is that the psychological similarities coexist with continued political differences between the two societies.2 The research reported in the second half of the chapter demonstrates that British politics are indeed more polarized by class divisions than are U.S. politics; the British Labour Party mobilizes working-class voters far more effectively than does the U.S. Democratic Party. We suspect it is the greater class structuring of British politics that sustains the belief in greater British class consciousness. But our results suggest that political differences may arise without any underlying difference in class consciousness.
Our strategy to detect cross-national differences in class perceptions relies on a comparative analysis of the class self-placement question. In effect, we recompute the analysis of class placements separately for each country and compare the results. If class is more clearly perceived in Britain, British workers ought to see themselves more uniformly as working class than do American workers; and British managers ought to see themselves more uniformly as middle class. Statistically, this ought to reveal itself in stronger relationships of the objective class divisions to the subjective class placements. But if the statistical relationships are similar in the two countries, then class labels are being assigned according to the same rules, and there is no noticeable superiority of British workers to identify their class membership.
The most critical requirement for cross-national survey analysis is to select equivalent samples responding to equivalent questions that are coded into equivalent categories.3 Fortunately, we begin with research studies that employed quite comparable designs. The U.S. data are from three of the American election surveys (1968, 1970, and 1972) in our Election Sample (see Chapter 4). The British data are from two surveys (1963 and 1964; supplemented in the voting analysis by a 1966 survey and 1970 reinterviews) from the study Political Change in Britain (Butler and Stokes, 1969, 1974). The resulting samples comprise 1,900 American and 1,163 British men, and 1,932 American and 1,485 British women.
Working class placements in Britain and the United States
The two sets of surveys asked virtually identical class self-placement questions: in each, respondents placed themselves in either the working or middle class. Britons selected the working-class label more often than Americans (74 percent versus 53 percent).4 By themselves, the less common working-class placements in the United States imply nothing about less class consciousness. Americans are more often managerial, college educated, and affluent than the British; given more of the attributes of middle-class position, Americans should more often label themselves accordingly.
The more interesting question is whether a U.S. and a British worker in equivalent positions in the social structure are equally likely to label themselves working class. The trick here is to create statistical equivalency of social position in the two samples. Data in the two surveys allow us to come close.
The stratification measures
Both surveys have adequate occupational codes from which we can construct a managerial class dichotomy and a prestige ranking according to the NORC scores (Siegel, 1971). Both also identify the self-employed. (The remaining class dimension, authority, was identified only in the British surveys and so was dropped from this analysis.)
It is difficult to make education equivalent across the two countries. In Britain, the type of secondary schooling people receive (secondary-modern schools versus the more selective grammar and public schools) and the source of their postsecondary education (whether universities, teachers’ and technical colleges, night schools, or apprenticeships) help determine their occupations and earnings (Treiman and Terrell, 1975; Burawoy, 1977; Kerckhoff, Campbell, and Trott, 1982). These qualitative distinctions ought to be important for middle-class placements as well and thus should be included in the analysis, together with the usual scale for years of education. The problem is to construct comparable qualitative measures from the U.S. data. Our solution has been to seek rough equivalents in the U.S. educational data wherever possible; where this is not possible, we add the remaining British distinctions separately.
The sample means for these variables are reported for each country in Table 7.1. As we expected, the U.S. averages are substantially higher on all variables except technical schooling. By using these dimensions as statistical controls, we can compare the class placements of U.S. and British workers in roughly equivalent positions. That analysis will tell us more about the levels of class consciousness in the two countries: if British workers are more likely than similarly placed U.S. workers to identify as working class, that would constitute some confirmation of the prevailing wisdom about weak American class consciousness.
To present the results, we abstract four hypothetical composites with widely different positions on these dimensions. We then calculate the estimated class placements of these four composite cases in each country. For the men, the four composites are as follows:
SOURCES: American Election Surveys, 1968, 1970, 1972; Butler and Stokes (1969), 1963 and 1964 surveys.
aStandard deviations are in parentheses.
bMen’s jobs are defined for all men in the samples and for all women who are not themselves listed as heads of household (the jobs coded are for the reported head). Women’s jobs are defined only for women who are employed.
cnd = not defined.
2. A machinist (working class with an occupational prestige of about 47) with 12 years of education, including some technical school, earning $8,000.
3. A steelworker (working class with an occupational prestige of about 33) with 10 years of education, including some technical school, earning $4,000.
4. A janitor (working class with an occupational prestige of about 19) with 8 years of education, earning $2,000.
None of these is self-employed, nor does any in Britain have an elite secondary or university education. For women, we substitute a teacher, a bookkeeper, a hairdresser, and a waitress, occupations with the same class positions and prestige as the male occupations. Because for women’s class perceptions we must also consider the jobs of the “heads of household” (see Chapter 8; Jackman and Jackman, 1983), we will assume that the women are married to men with the same class position and occupational prestige—to the four male composites, for example.
Figure 7.1 reports the estimated likelihood of a middle-class self-placement for each of these composite cases in each country. There are two important conclusions to draw from this figure. First, the levels of British and U.S. class placements are remarkably similar. The biggest difference is between the steel workers, who are 14 percentage points apart; most others are within a few points of each other. Once we compare equivalent British and U.S. workers, the Americans are just as likely to think of themselves as working class—and the main exceptions run counter to the assumption that there is a stronger British working-class consciousness: it is the Americans who more readily choose a working-class label.5
Second, and perhaps more important, the gap between U.S. managers and workers is as great as the gap between British managers and workers. If class divisions were less clear in the United States, we would expect the class placements to be less polarized: all the estimated class placements would collapse toward the middle of the scale. In an “ideology of classlessness,” more U.S. workers would assume middle-class placements, and more U.S. managers would revert to working-class placements. In contrast, in the class-conscious British society, virtually all managers and professionals would be clear about their middle-class positions, and British workers would be similarly clear about their working-class positions. Such differences do not emerge in Figure 7.1. U.S. class perceptions are just as clear (or just as uncertain) as British class perceptions. The accepted wisdom about a unique American classlessness is not supported by this empirical test.
FIGURE 7.1. Estimated class self-placements of four composites
SOURCES: American Election Surveys: 1968, 1970, and 1972. Political Change in Britain Surveys: 1963 and 1964.
Comparing the impact of specific variables
We need to look beyond these overall comparisons, however, because specific class divisions may still be more important to British class perception. For instance, it might be that in Britain the difference between our hypothetical schoolteacher and waitress is primarily a consequence of their different managerial class positions, while in the United States the difference might be a result of their different educational levels. If the class divisions were more important in Britain, we would still be justified in concluding that Britain is the more class-conscious society.
The more detailed analysis shows that each determinant of class perceptions is also similar in the two societies. We illustrate the results by comparing how much class placements would change in each country with changes in each class and status variable. Table 7.2 considers hypothetical British and U.S. workers with an equal probability of placing themselves in the working and middle classes. In the United States a man’s managerial position by itself increases middle-class placements by 18.3 percent, in Britain by 24.5 percent.6 Although this difference suggests that British class perceptions are somewhat more sensitive to actual class position, the 6.2 point difference is smaller than we might expect by chance. Similarly, although the results suggest other small differences between the two countries, only one of these is greater than chance expectations. The exception is self-employment, which has a stronger impact on British class placements. This difference might be interesting except that we have already noted (in Chapter 4) that in the American Election Sample the measure of self-employment seems strangely irrelevant to a variety of relationships found in other U.S. surveys. On the other hand, the American GSS found a significant self-employment effect on class placements that more closely parallels the British results. For this reason, we are reluctant to draw any conclusions about cross-national differences based on this single significant result.
SOURCES: American Election Surveys, 1968, 1970, 1972; Butler and Stokes (1969), 1963 and 1964 surveys.
aA t-test of greater than 2.0 indicates a difference between effects that is greater than chance (p > .05). The statistics are calculated from the interaction terms of the probit analysis.
bMen’s jobs defined for all men in the samples and for all women who are not themselves listed as heads of household (the jobs coded are for the reported head). Women’s jobs are defined only for women who are employed.
c* = p < .05.
dNot defined. Elite secondary and university schooling are defined only for the British sample. The reported t-test compares the British coefficient against zero.
The overall similarity of the results implies that the distinction between the middle class and the working class is made as easily in the United States as in Britain. Indeed, the cognitive rules used in relating objective position to subjective placement are almost identical. An additional year of education or a proportional rise in income will increase the likelihood of middle-class placement as much in the United States as in Britain. And in both countries, managerial class is an equally crucial determinant of class placement.
Other aspects of class consciousness
Thus far we have compared only class perceptions in the two countries. We should also consider more fully committed aspects of class consciousness that might yet explain differences between British and American class conflict. Even if Americans and Britons see class divisions in much the same way, they may feel quite differently about these perceived divisions. Perhaps these deeper feelings create the politically important differences between the two countries.7 Our data are not perfect for investigating these differences, but what data we do have provide little support for British-U.S. differences at any level of class consciousness.
Class closeness. One wave of the British survey (1963) and three waves of the U.S. survey (1968, 1970, 1972) asked respondents whether they felt close to their chosen class or no closer to people in that class than to people in other classes. If members of the British working class are more class conscious, we would expect them to claim this closeness as an expression of their class solidarity. But there are negligible cross-national differences in response to this question. Among self-identified working-class people, 60.0 percent in Great Britain and 58.8 percent in the United States report feeling closer to their chosen class than to people in other classes; this difference is smaller than chance expectations; therefore, it appears that U.S. workers claim class solidarity just as often as British workers do. At this level of class consciousness, too, there is little evidence for American exceptionalism.
Class salience. Another dimension of class perception sometimes studied in surveys is class awareness or salience: the degree to which survey respondents report themselves as thinking in class categories. In both surveys, respondents were first asked whether they thought of themselves as belonging to any class before being asked for their specific class placements.8 Of the U.S. working class, 63.5 percent agreed to thinking in class terms; of the British working class, 64.1 percent. Again, the difference is negligible and not statistically significant.9 Thus, research results for class awareness, like those for class closeness and for class perception, provide no support for the supposed weak class consciousness of U.S. workers. The consistency of this result, together with the lack of any prior research finding U.S. differences, must call into question the conventional wisdom.
Perceptions of a dominant class. The foregoing analyses almost exhaust the capabilities of our data, but other comparisons are worth speculating about. First, British-U.S. differences that do not seem to occur along a middle-class-working-class division might still be observed if respondents were asked about a dominant or ruling class. A greater British consciousness of a ruling class would fit with the British history of feudal privilege (important to Louis Hartz, 1955, among others).
This intriguing question must go unanswered for now, as we can only look where the data currently exist. But where we do have data, the result is not a trivial matter. The popular definition of the working class has been thought basic for progressive forces in industrial societies. It is confusion over the boundaries or even the existence of a working class that has been blamed for the lack of a genuinely socialist alternative in the United States. But the research reported here suggests, on the contrary, that U.S. workers are as clear about the working class as British workers.
Perceptions of inequality. Second, though our data do not permit us to compare more general attitudes toward economic inequality, a study by Wendell Bell and Robert Robinson (1980) found no support for any greater class consciousness among the British. They compared small samples from London and New Haven, Connecticut, and found that the New Haven residents perceived more inequality than the London residents: for example, stronger agreement—53 percent to 32 percent—with the belief that “people of higher social classes in this country get easier treatment by the police and the courts than people of lower classes do.” Moreover, the Londoners expressed equally egalitarian values condemning the class differences that do exist (Robinson and Bell, 1978): for example, similar levels of disagreement with the proposition that “it’s fair that rich people who can pay fines can stay out of jail while poor people may have to go to jail for the same crime.” These results reverse the conventional wisdom (and the researchers’ own expectations) that Americans hold more egalitarian values but are less conscious of existing inequalities.
Attitudes on class. Cross-national research on other class attitudes—what are sometimes called higher levels of class consciousness (Mann, 1973)—faces a crippling dilemma. On the one hand, surveys can ask about broad ideologies. Past research indicates that workers often appear quite conservative in such surveys (Free and Cantril, 1967; Mann, 1970; Chamberlain, 1983). The problem with these results is that such general attitudes do not determine any behaviors of consequence (Schuman and Johnson, 1976).10 Only attitudes about specific policies or groups have been shown to affect subsequent actions. A prerequisite for research on class consciousness, therefore, ought to be that the ideas and attitudes being studied actually predict some class-relevant behaviors (but see Kluegel and Smith, 1981:49 for how rarely this is done)11. The class self-placement question, whatever its limitations, does determine political preferences; most other supposedly class-conscious attitudes fail even this test.
On the other hand, attitudes towards specific objects pose a different problem for cross-national research. If questions are asked about unions, political parties, or the government, American workers will inevitably interpret these questions in the context of American unions, political parties, and government. If they then appear more skeptical than European workers of these institutions, their reluctance may not be a failure of class consciousness but merely a reflection of the real differences in these American institutions. Most radicals distrust the AFL-CIO, the Democratic Party, and the U.S. government; they should not then expect American workers to endorse reforms that would depend on these institutions.
Given these problems, it is understandable that there has been so little cross-national research on class consciousness.12 But neither the analysis in this chapter nor Bell and Robinson’s results support the contention that American workers are less class conscious or that Americans have an “inherited ideology of classlessness.” We can speculate that other psychological differences may exist between the United States and Britain, but where we have reasonably accurate data, we cannot find evidence for this particular one.
How then do we explain the widespread assumption that class consciousness is much greater in Europe than in the United States? Such a belief did not arise out of nothing. We suspect that the conclusions about class consciousness derived from observations about real differences in the political systems. Class position determines party affiliation and voting more often in Britain than in the United States (Alford, 1967); from this difference, it was inferred that British voters must be more class conscious than U.S. voters. This is again the familiar reductionist fallacy: all differences in behavior are attributed to differences in consciousness. The alternative explanation is not considered: that U.S. voters do not have class-oriented parties to vote for, while the British voters do.
Kay Schlozman and Sidney Verba tell a revealing anecdote about the Englishman trying to explain American politics to an English audience: ‘Tn America, there are two political parties. There is the Republican party, which is roughly equivalent to our Conservative party. And there is the Democratic party, which is roughly equivalent to our Conservative party” (1979:292). This “tweedledum-tweedledee” interpretation of U.S. politics describes the choice offered to U.S. voters, not characteristics of the voters themselves. We argue in this section that it is this structural choice that makes U.S. politics exceptional, but that the choice does not result from any weaker class consciousness of the voters.
Studying politics requires us to alter our research statistics, since we no longer have a simple two-category outcome to predict. In the British surveys, while most of the electorate expressed a Conservative or Labour preference, approximately 10 percent identified with the Liberal Party. And both nations include many citizens who identify with no party or who do not vote. Politics thus present us with many categories that do not fall easily along a linear scale, and these nonlinearities cause problems for our usual statistical techniques. Prior research has sometimes tried to escape these complexities by dropping nonvoters, independents, and third-party voters from the analysis, thus simplifying the statistics to a simple Republican versus Democrat (or Conservative versus Labour) choice. Alternatively, the other groups might be coded as a midpoint between the two major parties. But both of those strategies, in forcing the multiple categories of political choice into a linear scale, obscure one of the most important facts to be discovered about U.S. politics. Therefore, we use statistics (discriminant function analyses) designed to analyze categorical dependent variables; we find, happily, that the methodological innovation uncovers a critical substantive insight.
Both surveys asked respondents about their usual party affiliations. These affiliations, especially in the United States, provide a stable measure of political orientation and a convenient baseline for our later analyses of actual voting choices.
The discriminant function analyses find the “best” ordering of the party affiliations that maximizes the fit between the political categories and the array of class and status variables. For party affiliations (but not, as we shall see, for voting) the socioeconomic ranking of the parties yields no surprises (see Figure 7.2). Republicans and Conservatives top the scales; Democrats and Labourites hold down the bottom; independents (in Britain, respondents who refused to choose a party) are in between—in fact, almost equidistant from the two major parties. The Liberals in Britain rank quite high on these socioeconomic scales, resembling Conservatives more closely than they do Labour partisans.
A more revealing outcome of this analysis is the positions of our hypothetical composites along these socio-economic-political scales. As the discriminant function analysis creates the best ordering of the political parties, it also computes what combination of class position, income, education, and so on, determines people’s positions along these political scales. When we compute the positions of the four composites along these scales (Figure 7.3), the contrast with the class-placement results (Figure 7.1) is immediately apparent. The distance separating plant managers from janitors, and teachers from waitresses, is narrower in politics than in class perceptions. This is no surprise, for we expect people’s class positions to determine their class placements more closely than they determine political affiliations. More important are the striking U.S.–British differences in politics that were not found for class perceptions. The political distances are much larger in Great Britain than in the United States. British janitors and waitresses are more Labourite than U.S. janitors and waitresses are Democratic; British plant managers and teachers are more Conservative than U.S. plant managers and teachers are Republican. The British system is thus more polarized along class lines.
FIGURE 7.2. Discriminant function analyses of party affiliations
SOURCES: American Election Surveys: 1968, 1970, and 1972. Political Change in Britain Surveys: 1963 and 1964.
The comparison of the two figures tells us that while U.S. workers consider themselves just as much members of the working class as British workers do, the U.S. workers do not so readily translate that class identity into a political affiliation (and while U.S. managers are just as middle class, they do not translate that identity as readily into a Republican affiliation). For U.S. workers, the class feelings are there, but the political loyalties do not follow.13
FIGURE 7.3. Estimated political partisanship of four composites
SOURCES: American Election Surveys: 1968, 1970, and 1972. Political Change in Britain Surveys: 1963 and 1964.
The simplest explanation for these results is that the U.S.–British difference lies not in different levels of class consciousness but in a different party system. British politics offer British voters a choice between a middle-class and a working-class party; U.S. politics offer no such choice. Thus, once Americans recognize their class position, there is still little they can do about it.
What happens to U.S. working-class voters when they see no class differences between the major parties? The answer is found when we analyze actual voting choices rather than party affiliations. The surveys encompass several national elections in the two countries. The British surveys include voting for elections in 1964 and 1966; each election is analyzed and reported separately. The American data include the 1968 (Humphrey-Nixon-Wallace) and 1972 (McGovern-Nixon) elections.
What is most interesting in the voting analyses14 is the ranking of the voting choices along the socioeconomic scales (see Figure 7.4). Unlike the results for party affiliation, these results show an important difference between the U.S. and the British rankings of the political outcomes. British voting appears much like British party affiliations: Conservatives at the top, Labour at the bottom, and nonvoters somewhere between. The Liberal Party voters are again more similar to the Conservatives than to Labour, even outranking them in the 1966 elections. The U.S. ranking is completely different. While Republican voters are at the top, it is nonvoters who are lowest in socioeconomic status; Democratic voters are in the middle. This pattern reaches its extreme in the 1972 election, when McGovern voters are virtually indistinguishable from Nixon voters along this principal socioeconomic dimension.
In every election, the main social division in the U.S. electorate is between voters and nonvoters, not between voters for the different parties. The typical working-class response in the United States is to abstain. This is the argument of Walter Dean Burnham: “The ‘real’ class struggle, the point at which class polarization is most salient, is not found in the contests between Democrats and Republicans in the active electorate, but between the active electorate as a whole and the non-voting half of the adult population as a whole” (1980:37).
FIGURE 7.4. Discriminant function analyses of voting choices
SOURCES: American Election Surveys: 1968 and 1972. Political Change in Britain Surveys: 1964 and 1966.
The contrast with Great Britain tells us what has happened to U.S. politics. The Democrats occupy a position that is occupied in Britain by the Liberal Party. It is U.S. nonvoters who hold the same position as the British Labour Party; the United States does not offer a party to capture the bottom of the socioeconomic ladder. Throughout the industrialized West, these workers vote for the Left: for a Labour Party in Britain or Australia; for a Social Democratic Party in Sweden, Austria, or Germany; for the Communist Party in Italy and France. The United States has no leftist party, so these workers sit out. It is this working-class nonvoting that is exceptional in U.S. politics: “The large decline in participation after 1900 and the exceptional working-class abstention rate today very much resemble a gap in the active American electorate that was filled elsewhere by socialist parties” (Burnham, 1974:679). The United States has one of the lowest rates of voting in the democratic world. In a typical presidential election, only half of the voting-age population casts a ballot; in off-year Congressional elections this proportion falls to one-third. Even India, facing obstacles of enormous poverty and massive illiteracy, has a better record than this.
The pieces of this cross-national puzzle fit well together: the United States has one of the highest rates of nonvoting among democratic countries. U.S. nonvoters are far more working class than voters for either of the major parties. In British elections, nonvoters are not distinctively working class. There, as in most other democratic systems, workers vote for the Left. The United States does not offer workers a leftist political alternative. Without a party to express their class interests, U.S. workers lose interest in the electoral process. What is exceptional here is not the consciousness of U.S. workers but the structure of the U.S. political party system.
Structural determinants of politics
All of this is more easily understood if we drop any assumption that politics, even democratic politics, simply reflect the ideas of the voters. That is, we ought to seek explanations of political behavior directly in the dynamics of political institutions; inferences about the motivations of the actors are likely to be mistaken in attributing psychological differences.
This has been the problem with many explanations of cross-national differences in political behavior. The difficulties arise when we try to attribute the lack of class voting in the United States to the lack of motivation among the individual voters. U.S. voters seem not to be class conscious because they rarely vote or organize politically along class lines. But it may not be the case that the psychological predisposition is lacking; it may be only that there is no opportunity provided to express that predisposition.
How does a political system develop to frustrate working-class consciousness? We must remember that politics involve a contest between opposing forces—not just a tallying up of people’s wishes, like an opinion poll. Success in this conflict depends on many factors besides numerical strength. If one of the forces in politics is exceptionally strong—if it has great wealth, virtually uncontested social status, and a firm grip on existing political institutions—it can shape the political choices offered to the voters so that its dominance will not be seriously threatened.15 This is the argument we sketch in the final chapter: what is exceptional about U.S. politics and about U.S. class conflict in general is the extraordinary power of U.S. capital. There are good reasons to conclude that U.S. capital has had more power than the capitalist classes of other Western democracies. In the face of this overwhelming power, the U.S. working class has had a more difficult time constructing political and class organizations to defend its interests. The typical mistake that scholars have made is to infer weaker class consciousness from these greater difficulties. The evidence in this chapter suggests that class consciousness is not the explanation.
Several restrictions were placed on our two samples to achieve equivalence and analytic clarity. Since much of the research focuses on the role of class divisions and occupational structure, the samples were restricted to men in the labor force and to women who were themselves in the labor force or whose “household heads” were in the labor force. We also restricted the samples to the 21–65 age range. As in the earlier analyses, minorities present special problems for research on class perceptions (see Chapter 10; also Jackman and Jackman, 1973 and 1983; Goyder and Pineo, 1974; in Canada, Pineo and Goyder, 1973). We have therefore restricted the cross-national comparison to whites in each country, there being insufficient numbers of nonwhites in the British sample for an adequate comparison.
The final counts are not the effective degrees of freedom for the analysis, since respondents are weighted by several criteria. First, the sampling design of the British survey imposed weights (ICPSR, 1972:iii–viii). Second, the British survey was a panel study, so the 1,263 interviews of British men represent only 721 distinct individuals. To be conservative, we weighted the British samples to reflect the number of distinct individuals, not the number of interviews. Third, we adjusted for the different sample sizes in the different surveys so that each year had an equal weight (the harmonic mean of the sample sizes) in the final analysis. Finally, in the analyses where we combined the U.S. and British data, we weighted the two nations equally (to the harmonic mean of the number of individuals). The result was a weighted sample of 1,045.3 men and 1,260.1 women in each country.
Prestige scores. For the cross-national comparison, we use the NORC scores on the assumption that the U.S.-derived scores are not substantially different from British rankings (Hodge, Treiman, and Rossi, 1966; Treiman, 1977).
Income. Income data are available for the total family in the United States and for the head of household in Britain. The U.S. family income data are multiplied by 0.8 to approximate head of household income (see Lebergott, 1964). For both countries, income is translated into 1967 U.S. dollar equivalents by adjusting for the prevailing exchange rate and for changes in the U.S. consumer price index. The logarithm of this adjusted income is again used because it seems reasonable that class placements reflect proportional rather than absolute increases in income. After these adjustments, the British reported an average income only one-third that of the United States. This is undoubtedly an underestimate of the relative British incomes and probably an even greater underestimate of the relative standards of living.
Education. Four dummy variables have been created to estimate the effects of the qualitative differences in education. First, trade-oriented postsecondary schools are identified as night schools and apprenticeships in Britain, and as “vocational and technical training programs” in the United States (ICPSR, 1975:184–85). Second, a college-versus-noncollege dichotomy is created for both countries. The distinctions identify the more elite tracks of the British system: one for university education (as distinct from teachers’ or technical colleges), and another for the kind of secondary school: the basic secondary modern track versus the elite tracks (grammar school, public school). No equivalent U.S. variable could be constructed, although if the data were available, it would be interesting to test whether the American “prep” school—admittedly a more limited phenomenon—might not have the same class-defining characteristics as the British grammar and public schools (cf., Mills, 1956; Domhoff, 1967) and whether a college curriculum within U.S. high schools had an effect equivalent to the British grammar school education.
Party affiliation. In the United States, the surveys asked:
Generally speaking, do you usually think of yourself as a Republican, Democrat, independent, or what? (ICPSR, 1975:63).
In Britain, the surveys asked:
Generally speaking, do you usually think of yourself as Conservative, Labour, Liberal, or what? (ICPSR, 1972: 180).
Note that the U.S., but not the British, surveys included an explicit choice of independent. The first wave (1963) of the British survey did not even include the final “or what?”
In addition, both surveys probed further for strength of party affiliations. We have ignored these further distinctions, since a separate discriminant function analysis revealed that strong and weak party identifiers did not fall along a linear scale of socioeconomic position.
The figures in the text are computed from the detailed equations reported in Tables 7.A-7.D, below. We have endeavored to make the text presentation as straightforward as possible without ignoring the complexities of the results. In this regard, we have made two decisions that require further comment.
Pooling men and women. Throughout most of this chapter, we pool the men’s and women’s data to present a single set of results. Both the pooled and separate equations are reported in the following tables. The pooled results simplify the presentation; in general, the results for men and women are similar. We make an exception in reporting both genders when we focus explicitly on the occupational composites, since occupation is so closely linked to gender.
The analysis of class placements does reveal some differences between men and women. In the separate equations, some British-U.S. differences emerge that are not found in the pooled equation. We have not commented on these differences because there seems to be little pattern to the results. None of the male differences meet the usual criterion for statistical significance (p < 0.05). For women, two of the effects (income and head of household’s self-employment) are more important in Britain; also, elite secondary schooling, a distinction defined only in Britain, has a significant impact on women’s class placements. If we relax the criterion somewhat (p < 0.10), three of the male differences and one more of the female differences are statistically significant. But the patterns are not consistent across gender. Only self-employment shows a statistically significant cross-national difference for both genders.
Discriminant function analyses. We rescale the discriminant function results reported in the tables in order to make them more equivalent to the probit analyses of class placements. We approximate equivalent scaling factors by standardizing the constructed party and voting scales, much like the standardized cumulative normal that underlies the probit analysis. That is, we scale the discriminant function results so that the political party and voting scales (the dependent variables) have zero means and unit standard deviations. They are then transformed to 0–100 scales through the cumulative normal function, thus approximating the same scale used in the probit analyses of class placements.
SOURCES: American Election Surveys, 1968, 1970, 1972; Butler and Stokes (1969) 1963 and 1964 surveys.
aMen’s jobs are defined for all men in the samples and for all women who are not themselves listed as heads of household (the jobs coded are for the reported head). Women’s jobs are defined only for women who are employed. If a variable is undefined for any variable, the respondent is given a score of 0. A dummy variable (“defined”) identifies those respondents for whom these variables are defined (coded 1). Thus, the coefficient for the men’s managerial position represents the difference between men (or husbands) who are managers and men (or husbands) who are workers; the coefficient for “defined” represents the difference between respondents for whom these variables are defined and those for whom they are not defined. Managerial position includes the census classification of managers and professionals but excludes technicians.
b* = p < .05. Standard errors are in parentheses.
cnd = not defined.
dElite secondary and elite university are defined only for the British sample.
eSo that this variable yields meaningful results, the income, education, and occupational prestige variables have been centered at $5,000, 12.0 years of school, and 40.0 prestige. The coefficient therefore represents the difference between U.S. and British workers who are neither managerial nor self-employed, with an occupational prestige of 40, an income $5,000, 12 years of school, no technical school, no college, and no elite secondary or university.
SOURCES: American Election Surveys, 1968, 1970, 1972; Butler and Stokes (1969), 1963 and 1964 surveys.
aThese coefficients are scaled somewhat differently from the coefficients reported in conventional analyses. Usually, the coefficients of the independent variables are scaled so that the discriminant function constructed from them has zero mean and unit standard deviation. The centroids of the groups are then computed along this function. We reverse this so that the groups are scaled to have a zero mean and unit standard deviation, and the discriminant function coefficients are scaled to predict these group scores. Our results are analogous to an ordinary least-squares regression of the political groups (coded to the computed centroids) on the independent variables.
bMen’s jobs are defined for all men in the samples and for all women who are not themselves listed as heads of household (the jobs coded are for the reported head). Women’s jobs are defined only for women who are employed. If a men’s job or women’s job variable is undefined for any respondent, the variable is given a score of zero. A dummy variable (“defined”) identifies those respondents for whom these variables are defined (coded 1). Thus, the coefficient for the men’s managerial position represents the difference between men (or husbands) who are managers and men (or husbands) who are workers; the coefficient for “defined” represents the difference between respondents for whom these variables are defined and those for whom they are not defined.
cnd = not defined.
NOTES: aCoefficients are standardized partial regression coefficients. Discriminant functions are computed from the socioeconomic variables from Table 7.B.
SOURCES: American Election Surveys, 1968 and 1972; Butler and Stokes (1969), 1964 and 1966 surveys.
NOTES: See notes to Table 7.B.
1. Kahl (1957:182) took note of the UNESCO study but dismissed its results for unspecified reasons. The most recent edition of that textbook (Gilbert and Kahl, 1987) omits any reference to either the study or the problem.
2. Thus, our results should not be interpreted as yet another instance of the substantial “homogenization” of industrial societies. A popular school of sociology has developed the idea that all industrial societies converge toward a more or less common social structure dictated by the necessities of industrial production (Kerr et al., 1960). This industrial convergence theory seemed to be supported by the high cross-national correlations of prestige rankings (e.g., Hodge, Treiman, and Rossi, 1966). In contrast, what we find important is the juxtaposition of social psychological similarities and structural—especially political—differences.
3. Details of the problems encountered in this analysis are reported in a methodological appendix to this chapter.
4. These differences are more pronounced but essentially similar to those in the 1948 data reported by Buchanan and Cantril (1953). Among the nine nations they studied, Britain had the highest rate of working-class identification (60 percent), while the U.S. rate was third highest (51 percent).
5. We do not attribute much importance to this reversal. We are especially suspicious that the income controls overcorrect for American affluence. To test for this, we eliminated the national income differentials by recoding income into percentile ranks within each country. When we control for relative income rank rather than for actual dollars of income, the U.S. workers appear on average slightly more middle class than the British workers. Since the dollar estimates probably overcorrect for income differences and the percentile ranks undoubtedly undercorrect, it seems safest to conclude that neither country is more (subjectively) working class than the other.
6. The results seem to indicate that a woman’s job has little effect on her class placement but that the job of the “household head” does. These data are not appropriate for addressing this issue, however, since we cannot distinguish in the British samples (and therefore do not do so in the U.S. samples) between full-time and part-time workers. The question of women’s class self-placements is quite complicated; see Chapter 8.
7. Landecker (1963), e.g., stresses the difference between cognitive and affective components of class consciousness. The class-placement question addresses the issue of cognitive differences only. It may be that while class position is equally clear to Americans and Britons, Americans invest less affect in this identification. Giddens (1973) and Mann (1973) also propose several levels of class consciousness of which class perception is only the most basic. While we are skeptical of such “armchair introspection” into class consciousness, we would like to investigate each of these possibilities before we draw firm conclusions about cross-national similarities.
8. Previous research using this question (Guest, 1974) has linked such awareness to support for more “liberal” or collective-governmental strategies of social change as opposed to dependence on individual efforts.
9. A more complete loglinear analysis, not reported in detail here, shows that the American middle class reports itself as class aware more often than the British middle class (73.9 percent to 58.8 percent). We are reluctant to read too much into this result; however, a more class-conscious U.S. middle class would be consistent with our theory that the most exceptional characteristic of U.S. class structure is the greater strength of its dominant class (see Chapter 12).
10. Nor are general beliefs a consequence of the class situation or personal economic experiences of survey respondents (Schlozman and Verba, 1979). Ideological beliefs are more often the result of direct socialization by parents, peers, and mentors (see Portes, 1971b, on the sources of radical beliefs among the Chilean workers). Burawoy (1979:201) observes that the broad values socialized during childhood play little role in the class struggles on the shop floor.
11. The problem of behavioral relevance also disqualifies most survey questions that ask respondents how they feel about their own position in the social order. Collective action is motivated by perceptions about collectivities: about one’s class, ethnic group, gender, or neighborhood. In Runciman’s (1966) terms, it is fraternal not egoistic deprivation that is relevant (see also Vanneman and Pettigrew, 1972). Hamilton and Wright (1986:366) also find a sharp disparity between continuing high levels of personal satisfaction as reported in opinion polls and the dramatic drop in public confidence in the government and other national institutions. The perception of social structural problems appears to be largely independent of the perception of one’s own well-being.
12. Wright (1985), using a six-item attitude scale, finds U.S. workers less class conscious than Swedish workers. But much of the difference in his class-consciousness scale derives from different attitudes toward strikes, and once union membership and other background variables are controlled, there is no more class polarization in Sweden than in the United States (1985:276).
13. This can be seen also in path analyses constructed from the discriminant functions (chapter appendix, Table 7.B). The path analyses add class placements as an intervening variable between the socioeconomic scales and the political party scales. The British results show that class placements are an important mediator between objective socioeconomic position and party affiliation: objective position determines class self-placements, which determine party affiliation. In the United States, this causal chain is broken at the second step: as in Britain, objective position determines class placements, but these self-placements are almost irrelevant to party politics. In a separate British regression of the political party scale on all the individual class and status variables, class placement had the largest zero-order correlation and the largest standardized partial regression coefficient with the political party scale. In the United States, class placement was not even statistically significant.
14. These results are reported in detail in the appendix, Table 7.D. The British composite scores are again more polarized than the U.S. scores, although the contrast is somewhat smaller for the voting analysis than for the party affiliations. Again, class position (managerial occupation and self-employment) determines political outcomes in Britain more than in the United States.
15. Again, Burnham (1980:66): “The structure of political choices offered the electorate in the United States, and the major decisions made by political elites, have together produced more and more baffled ineffective citizens who believe that chance rules their world. This implies the long-term paralysis of democracy.”