Thinking About Stratification: Reflecting Back on Week Five
Last class I wanted to get you talking about the significance of occupations (as opposed to classes), including
(1) the function of occupations (as the essential component of the division of labor, as developed by Durkheim)
(2) the utility of occupations as a base for calculating or measuring inequality in a population of working persons
(3) the importance of occupations as the basis for generating and sustaining inequality (particularly, income or earnings inequality)
Since we had already done this for class (two weeks earlier) it seemed a fairly reasonable way to proceed. It seems, however, that this was a much more difficult task, at least partly because I'm not convinced that Grusky et al. make a compelling case for occupations. My disinclination to use occupations as they do and my reluctance to purge occupations from my analysis of class, makes my personal position somewhat confusing (not just for students but for colleagues as well). In this case, this may have been a barrier to our discussion. For this, I apologize.
After class I had a discussion with Bob Perrucci about the class and the problems with Grusky, Sørensen, and even Erik Olin Wright. Bob suggested that we need to keep in mind the fact that all of these folks are attempting to develop classification and coding systems that others (their students, of course, but their sociological posterity, more generally) will use. In that regard, although they are couching their discussions in theoretical terms—particularly in Durkheim and Marx, their concerns are primarily methodological (question 2, above).
Durkheim's theory of the division of labor argues against Herbert Spencer and Adam Smith, who focus on the economic function of the division of labor—it is more efficient and effective and is adopted by societies that are developing economic capacity in evolving from warring barbarian tribes toward modern industrial societies. Durkheim was concerned that the shift from small, isolated, subsistence settlements to modern urban industrial centers would undermine social integration and regulation and the sense of being a part of something greater than oneself—solidarity. He believed, at least initially, that occupational associations might take on some of the regulatory and integrative functions that family and religion had served, although he became increasingly concerned with the continuing need for family and religion or some sort of solidarity based on sameness (rather than interdependence).
Grusky wants to convince us that Durkheim was right about the importance of occupations while Marx was wrong about the importance of class. Thus the argument between Grusky and Wright has developed into a debate on the relative merits of class and occupation, along with efforts to develop better measures of each.
So, with that said, I'd like to consider Abigail's questions: What difference does it make, as a practical matter? Are their policy implications?
The simple answer is, yes. If occupations are both major integrative mechanisms and the basis for sorting people into strata (or more or less prestigious or honorable statuses, or more or less well paid positions) then the process of fitting persons to occupations is critical. Clearly, families are expected to raise their children to be good citizens, but schools are particularly important in both teaching and sorting children toward their future occupations. The failure of schools to prepare children for the occupations (positions) that society needs to fill is the fundamental problem from this perspective. The problem, in Indiana right now, is being defined as the problem of bad teachers hiding behind unions (which are interfering with the free market). Schools need to be changed into more efficient systems of preparing children for their future careers. Essentially, the problem is the integration of political, economic, and cultural systems. Schools (including universities) are supposed to be preparing students for work in the real world, but teachers and professors are not living in the real world. Schools need to be run more like businesses, with teachers hired and fired and rewarded for the results that they produce—as evidenced by student performance on standardized tests.
Needless to say, a Marxist approach would be more critical of the economic system and argue that class relations (regardless of particular occupation) alienate individuals from their true productive, sociable nature by imposing a system in which owners of the means of production exploit a large population of workers who are alienated and miserable but have no choice since they have nothing to sell but their labor. From this perspective, education reproduces class privilege and contributes to the dominant ideology of free market (bourgeois) principles of equal opportunity and rewards for the best and the brightest and those who work the hardest. Here the problem is private ownership of the means of production and exploitation of labor on that basis. Once the economy has been reorganized by the proletarian revolution, each will work according to ability and receive according to need. Until then, the rich will get richer and the poor poorer.
The problem, of course, is that neither Durkheim's not Marx's vision of the modern world fits the reality that we face today, so we are left with various attempts to build on one or the other in an effort to understand inequality in contemporary society. Might we try to combine the two theories—solidarity and exploitation models? Max Weber tried a version of that in his interactive contingency model of class, status, and party competition and conflict. To some extent, labor market theories and liberal feminist and racial inequality theories all attempt to do something similar. Personally, I'm not convinced that they are successful. Ultimately, you can add Weber to either Marx or Durkheim and get something like Tilly or Hauser, but neither would be considered a synthesis of Marx and Durkheim.
With regard to the more practical/methodological issue of the relative merits of continuous (gradational) versus categorical measures, we did manage to cover that in class. The gradational measures tend to explain more of the variance in income inequality—which is, essentially, their goal. The categories are more effective in analyzing income gaps, particularly differences between classes—see discussion in Hogan (2005:650).
This brings me to Nick's question. Can't we use both occupations and classes? Again, the simple answer is yes. This is the path that both Bob Perrucci and I would recommend. Rather than attempting to classify professionals as a class we can recognize that there are employers and employees among the professionals. At the same time, using all of the occupations in census code (or coding them for occupational prestige) might increase the explained variance but it is likely to explain away the categorical class and occupational measures and leave us with no better understanding of how the distribution of people over occupations explains earnings.
Ultimately, if we knew everything about everyone and could take all of this into account we could distinguish each individual and "explain" all of the differences between them. The statistical problem, however, is that we can't include more variables than cases. We lose "degrees of freedom" with every variable that we add. If we have very large datasets, like the CPS data that I analyze in my "orphan paper" (on website) we can include lots of variables (I use 19). Even here, however, the sample of black women and black men is only about 2,000 (each), so we might not want to use all 901 census occupation codes because we would have lots of empty cells and would not be able to calculate reliable means and standard deviations.
Even if we had ten times as many cases, how could we explain the effects of everything we know about everybody? What is important? Is it not more effective to include a few categories rather than all of them? Consider, for example, the idea of "core" and "peripheral" industries. This is a very effective way of summarizing a series of labor market effects that are associated with higher or lower earnings. We could, instead, include more dummy variables to represent all of the detailed industrial codes. Assuming that all of these produced significant effects they would increase the explained variance in the model, but would this help us to understand how labor markets work? This is where theory comes in. Our theories should help us to determine which variables we need to include.
Here we should probably turn to the third question (above). It is pretty clear how class relations generate inequality (through exploitation), but how occupations generate inequality requires a theory of the ranking of occupations (like Davis and Moore). Micro-economic theories (like Grusky and Sørensen) simply view occupations as the mechanism through which people maximize their resources (more or less efficiently). If we could code every resource as a variable then we could better estimate the earnings of each person. As we shall see, however, labor markets are not as open as micro-economic and status attainment theories suggest. We already spoke about "closure" with regard to family, occupational/professional/industrial organization, and government regulation, licensing, certification, etc. We will talk more about this in considering race and ethnicity and gender.