The conceptualisation and measurement of occupational hierarchies: A review, a proposal and some illustrative analyses

(2011) - Erzsebet Bukodi, Shirley Dex, John H. Goldthorpe

Journal: Quality & Quantity
Link:: http://link.springer.com/10.1007/s11135-010-9369-x
DOI:: 10.1007/s11135-010-9369-x
Links::
Tags:: #paper #NCDS #SocialClass
Cite Key:: [@bukodiConceptualisationMeasurementOccupational2011]

Abstract

Occupational data are central to much research in the field of social stratification. Yet there is little consensus on how such data are most appropriately classified and scaled. We evaluate occupational scales currently in use on the basis of a fourfold typology. This crossclassifies scales, on the one hand, according to whether they are intended to be ‘synthetic’ or ‘analytic’ and, on the other, according to whether they are based on ‘subjective’ or ‘objective’ data. Focusing chiefly on issues of validity, we argue that scales of the analytic-objective type are those which, for most purposes, can be used to best advantage in stratification research. We illustrate our argument by applying scales of occupational earnings and occupational status in analyses of the worklife occupational mobility of men in Britain, using the data-set of the National Child Development Study.

Notes

“we argue that scales of the analytic-objective type are those which, for most purposes, can be used to best advantage in stratification research” (Bukodi et al., 2011, p. 623)

“The first is that the concept of occupation already involves some degree of abstraction. What we call ‘occupations’ are categories that bring” (Bukodi et al., 2011, p. 623)

“together particular jobs, deemed to be in certain respects similar” (Bukodi et al., 2011, p. 624)

“The second point is that although occupations are of central importance in the study of social stratification, they may also have features—for example, distinctive subcultures or ‘communities’—that, while of obvious sociological interest, cannot themselves be readily given any ordering in terms of socioeconomic advantage and disadvantage” (Bukodi et al., 2011, p. 624)

“those scales that we treat as synthetic aim to order occupations on a single dimension, on which it is believed that a number of their more specific features are brought together in a meaningful way” (Bukodi et al., 2011, p. 625)

“those scales we call analytic aim to order occupations in respect of one specific feature only. Note here our emphasis is on what scales—according to their authors—aim to do.” (Bukodi et al., 2011, p. 625)

“Occupational scales of this type are almost entirely scales aiming to measure the prestige of occupations in popular assessment” (Bukodi et al., 2011, p. 625)

“The criticism that has most commonly been made of SIOPS, and of national scales of the kind from which it is derived, is that they have poor criterion validity.” (Bukodi et al., 2011, p. 625)

“That is to say, they do not show correlations with certain other variables that are as high as would be theoretically expected, and that are lower than those produced by other occupational scales (see e.g. Hauser and Warren 1997, p. 190)” (Bukodi et al., 2011, p. 625)

“However, we would further suggest that the lack of criterion validity of occupational prestige scales of the kind in question stems from a more fundamental problem: that is, from their lack of construct validity or, in other words, from their failure to capture adequately what, analytically, they are supposed to capture—the prestige of occupations in the sense of their positions within a hierarchy of social status” (Bukodi et al., 2011, p. 625)

“In short, so-called prestige scales do not work well in the role of explanatory variables because they do not measure effectively what they claim to measure” (Bukodi et al., 2011, p. 626)

“As our example of this type of scale we take the Hope–Goldthorpe (H–G) scale which is based on survey respondents’ ratings of the ‘social standing’ of occupations—in essentially the same way as in studies of the kind from which SIOPS is derived” (Bukodi et al., 2011, p. 626)

“they propose that their scale should be taken as tapping respondents’ evaluations of the ‘general desirability’ of occupations in which perceptions of job requirements and job rewards are of main importance (Goldthorpe and Hope 1974). In other words, they would represent their scale in such a way that, in terms of our typology, it is to be treated as synthetic rather than analytic.” (Bukodi et al., 2011, p. 626)

“the H–G scale is open to exactly the same criticism as SIOPS and the individual scales on which SIOPS is based: i.e. that it appears to lack criterion validity in not performing especially well as an explanatory variable” (Bukodi et al., 2011, p. 626)

“With scales of this type we need to provide two quite different examples. The first is that of a scale of the ‘socioeconomic status’ of occupations, the International Socioeconomic Index (ISEI).” (Bukodi et al., 2011, p. 626)

“More specifically, the authors of the scale argue that it is through individuals’ occupations that their education is most importantly linked to their incomes: occupations are the ‘engines’ that convert education into income (Ganzeboom et al. 1992, pp. 10–11 ).” (Bukodi et al., 2011, p. 627) important

“captures the indirect effect of education on incomes” (Bukodi et al., 2011, p. 627)

“The second example we give is that of the Cambridge Scale—a number of versions of which for countries other than Britain have by now been developed” (Bukodi et al., 2011, p. 627)

“scale differs from the ISEI—as also from the H–G scale—in that the occupational hierarchy is seen as being determined not by the requirements and rewards of different occupations, whether assessed via popular perceptions or by factual information on education and income, but rather by patterns of social interaction” (Bukodi et al., 2011, p. 627)

“aims to reflect ‘social distance’ in some quite general sense of ‘combined material and social inequality or advantage and disadvantage’ (Prandy 1990, p. 630; Prandy and Lambert 2003, p. 400 our italics).” (Bukodi et al., 2011, p. 627)

“In evaluating synthetic-objective scales, we again see issues of construct validity as being of key importance” (Bukodi et al., 2011, p. 627)

“Education is obviously one important individual input to the labour market; but it is also known that on-the-job training, informal as well as formal, is a further major determinant of earnings and, of late, increasing attention has focused on the earnings returns to individuals’ non-cognitive attributes (see e.g. Bowles and Gintis 2000).” (Bukodi et al., 2011, p. 627)

“the issue of construct validity is here not so much inadequately treated as entirely by-passed.” (Bukodi et al., 2011, p. 628)

“Rather than starting out with some attempt to specify what it is they are seeking to measure and then developing an instrument for this purpose, the validity of which is open to test, the authors of the scale begin with a measure—the dimension they extract from their data on the occupational bases of friendship or marriage—and then simply assert that this provides a synthetic or general scale: i.e. one that captures the structure of social inequality or of social advantage and disadvantage in some quite comprehensive way. No explanation is offered of why this should be so, and insofar as the issue of validity is considered at” (Bukodi et al., 2011, p. 628)

“While poor criterion validity may well point to poor construct validity, as in the case of occupational prestige scales, seemingly good criterion validity does not guarantee good construct validity” (Bukodi et al., 2011, p. 628)

“The first example is a scale of occupational earnings or, to be precise, of average hourly earnings rates, constructed for Britain on the” (Bukodi et al., 2011, p. 628)

“basis of data—for men only—from the 1975 New Earnings Survey” (Bukodi et al., 2011, p. 629)

“author (Nickell 1982, p. 43) explicitly recognises that while income is one criterion of such success, others can be envisaged, and it is of particular interest that he should here refer also to status” (Bukodi et al., 2011, p. 629)

“Our second example is a scale that aims, just as specifically, to measure occupational status or, it might better be said, social status as expressed via occupation. The scale is one constructed in the same way as the Cambridge Scale: that is, through extracting a principal dimension from data on social interaction among members of different occupations—in this case, data on close friendship.” (Bukodi et al., 2011, p. 629)

“authors ofthescale(Chan and Goldthorpe 2004) believe that in this way the relative social status attaching to occupations is effectively captured, but only social status and not other, more ‘material’ aspects of the occupational hierarchy” (Bukodi et al., 2011, p. 629)

“implement empirically the Weberian conceptual distinction between status, on the one hand, and class, understood in terms of social relations in economic life, on the other (Chan and Goldthorpe 2007).” (Bukodi et al., 2011, p. 629)

“is in fact analytic-objective occupational scales that we would see as holding out most promise in future research” (Bukodi et al., 2011, p. 629)

“In addition, objective data on occupations are in general easier to interpret in this regard than are individuals’ perceptions and evaluations.” (Bukodi et al., 2011, p. 629)

“the validity of the occupational earnings scale is relatively unproblematic, given that the data of the New Earnings Survey on which it draws are known to be of high quality.” (Bukodi et al., 2011, p. 629)

“homophily is indeed being captured—but homophily in respect of status rather than of other attributes; and empirical findings in support of this claim can then be advanced. In fact, only quite modest correlations exist between status as measured by the scale and both income and education (Chan and Goldthorpe 2004).” (Bukodi et al., 2011, p. 629)

“With analytic-objective scales, we would in fact see the more difficult issues that arise as concerning not validity but rather the choice of which aspects of occupations might thus be scaled to good advantage and which not” (Bukodi et al., 2011, p. 630)

“he question can obviously be asked of why one should want to construct a scale of, say, occupational earnings, rather than collect data on individuals’ earnings directly” (Bukodi et al., 2011, p. 630)

“a practical level, it is well known that respondents to social surveys are far more ready to provide information about their occupations than about their earning” (Bukodi et al., 2011, p. 630)

“Second, it can be shown that occupations, and especially insofar as they fall in to different class categories, are associated not only with differing levels of current income but, further, with significant differences in income security, short-term income stability and long-term income prospects (Goldthorpe and McKnight 2006; Chan and Goldthorpe 2007; McGovern et al. 2007, Chap. 3 ).” (Bukodi et al., 2011, p. 630)

“As the main complement to occupational earnings scales, we would therefore favour occupational status scales of the kind developed by Chan and Goldthorpe.8 In other words, as suggested by Nickell, status can be seen as a major reward obtained via occupation in addition to income—but also as one that may to some extent be an alternative to income.” (Bukodi et al., 2011, p. 630)

“In the analyses that follow, we draw on the data-set of the National Child Development Study (NCDS),” (Bukodi et al., 2011, p. 631)

“Table 2 we show how the two scales relate to each other when they are used, in their five-level form, to cross-classify all jobs ever held by males aged 16–46 in the NCDS data-set” (Bukodi et al., 2011, p. 631)

“although a positive correlation exists between occupational earnings and status, with empty or near-empty cells occurring towards the top-right and bottom-left corners of the table, the correlation is still not all that strong” (Bukodi et al., 2011, p. 631)

“As would be expected from results previously reported by Chan and Goldthorpe (2004, pp. 391–392), the former occupations are mainly ones associated with the manufacturing, construction and transport sectors, while the latter are mainly ones associated with administration, sales and personal services” (Bukodi et al., 2011, p. 631)

“Scales that are by intention of the analytic-subjective type—in effect, scales supposedly of occupational prestige, such as SIOPS—do not adequately measure what, conceptually, they are supposed to measure, at all events if prestige is equated with social status in the classic Weberian sense” (Bukodi et al., 2011, p. 637)

“Scales of the synthetic-objective type likewise face problems of construct validity. In the case of socioeconomic status scales, such as the ISEI, where the aim is no longer simply to mimic prestige scales, no compelling rationale has been developed for determining which aspects of occupations and weightings thereof should be taken as best capturing socioeconomic status.” (Bukodi et al., 2011, p. 637)

“For most research purposes, therefore, we argue that advantage will lie with using occupational scales of the analytic-objective kind: that is, scales which aim to order occupations in terms of one specific feature only, for which a relatively clear definition can be provided” (Bukodi et al., 2011, p. 637)

“and an appropriate metric devised using data of a factual kind. In this way, problems of construct validity are at all events made transparent and open to empirically-based assessment” (Bukodi et al., 2011, p. 638)