@lambertOptimisingUseMeasures2021

(2021) - Paul Lambert, Camilla Barnett

Journal: The Routledge Handbook of Contemporary Inequalities and the Life Course
Link:: https://www.taylorfrancis.com/books/9780429470059/chapters/10.4324/9780429470059-18
DOI:: 10.4324/9780429470059-18
Links::
Tags:: #paper #SocialClass
Cite Key:: [@lambertOptimisingUseMeasures2021]

Abstract

Many different approaches are available to measure the social stratification position of individuals. It is well known that different approaches can be associated with different theoretical and empirical properties. Nevertheless there is little consistent advice when confronting two important and interconnected considerations that affect many analyses of inequalities: how can we best exploit stratification measures when an intersectional and/or longitudinal understanding is prioritised? This paper will review the features of a number of important candidate measures of social stratification and discuss the challenges and opportunities for adapting conventional practices in ways that can take better account of intersectional and longitudinal analytical considerations.

Notes

"how can we best exploit stratification measures when an intersectional and/or longitudinal understanding is prioritised?" (Lambert and Barnett 2021:188)

"Many measures are 'occupation-based': starting from the premise that occupations are key influences upon the circumstances of inequality experienced by individuals and their families" (Lambert and Barnett 2021:189)

"locality characteristics" (Lambert and Barnett 2021:189)

"asset-based" (Lambert and Barnett 2021:189)

"Occupation-based measures often comprise a social class categorisation with a small number of different classes" (Lambert and Barnett 2021:190)

"one-dimensional metric scales based on occupations are also often calculated" (Lambert and Barnett 2021:190)

"few examples a much higher number of different occupation-based social class categories are studied" (Lambert and Barnett 2021:190)

"Non-occupation-based measures are perhaps more frequently expressed in a metric functional form" (Lambert and Barnett 2021:190)

"It is not yet common practice to do so, but in other reviews we have argued that social scientists ought to undertake extensive sensitivity analysis, in which they derive many plausible measures, then compare and contrast their empirical properties, in order to make a well informed decision" (Lambert and Barnett 2021:190)

"CAMSIS occupation-based stratification" (Lambert and Barnett 2021:190)

"key mechanism of socially structured inequality (e.g. DiPrete and Eirich 2006)" (Lambert and Barnett 2021:191)

"More pragmatically, longer term structural economic transformations ordinarily alter the underlying distribution of stratification measures in a society over time - expansion and re-structuring of educational systems, in particular, means that for different birth cohorts the relative distribution of educational qualifications is often very different" (Lambert and Barnett 2021:191)

"some influential comparative evaluations of occupation-based stratification measures have argued that harmonised international measures work reasonably consistently across time as well as across countries" (Lambert and Barnett 2021:191)

"Within countries, academics have often issued revised and updated measures in response to changing industrial distributions through time, but these approaches do not generally provide guidance on making comparisons between time points" (Lambert and Barnett 2021:191)

"All of these examples can be described as a model of a priori 'measurement equivalence' in terms of comparisons over time: an assertion that measures can be interpreted consistently across time because they are operationalised in a consistent way" (Lambert and Barnett 2021:191)

"when the priority is of consistent interpretation across extended time-periods, there is often a good case for standardisation according to the appropriate temporal distribution" (Lambert and Barnett 2021:192)

"Standardisation is theoretically compelling as social stratification position is often conceived of as relative position within a given inequality structure: what matters for comparison is relative position, not an absolute position which might be contingent upon societal-level change in industrial or economic structures" (Lambert and Barnett 2021:192)

"continuous measures of stratification, arithmetic mean standardisation is an obvious option" (Lambert and Barnett 2021:192)

"categorical measures of stratification such as social class schemes, a comparable approach to standardisation is not conventional. One plausible option however is to use 'effect proportional scaling' to generate scale scores for categories, then mean standardise those scores within the temporal context" (Lambert and Barnett 2021:192)

"Alternatively, a theoretical temporal standardisation might be obtained on a proiri criteria, for instance by defining modal categories, or other key categories that are specific to the context, and making contrasts with them." (Lambert and Barnett 2021:192)

"plausible, however, to deploy a pre-analysis standardisation, in which different stratification measures are derived bespoke to the appropriate temporal context - examples include the derivation of new CAMSIS scales for new time periods in the same country (e.g. Lambert and Griffiths 2018), and the calculation of alternative social class schemes for alternative points in history (e.g. van Leeuwen and Maas 2011)." (Lambert and Barnett 2021:192)

"Most stratification measures have some correlation with age and life-course circumstances" (Lambert and Barnett 2021:192)

"it is still common to find naïve analyses that ignore or downplay the link between stratification measures and the life-course" (Lambert and Barnett 2021:193)

"popular strategy is to fit control variables that indicate life-course stage in a statistical model, but even when doing so, controls might be under-developed, if they are captured in only a rudimentary functional form" (Lambert and Barnett 2021:193)

"At its simplest, it may be compelling to restrict an analysis to a narrower range of ages or life-course circumstances; more ambitiously, we might use strategies that focus upon fine-grained conjunctions of age, life-course stage and stratification position," (Lambert and Barnett 2021:193)

"for instance a statistical matching approach or the tools used to model 'intersectionality' as described in section 4." (Lambert and Barnett 2021:193)

"A commitment to modelling trajectories can involve a data reduction analysis, where latent patterns in trajectories are identified, then subsequently analysed (e.g. Pollock 2007; Wielgoszewska 2018)." (Lambert and Barnett 2021:193)

"defined on a priori grounds - for instance, analysing a measure derived arithmetically from circumstances over time (e.g. Bottero 2005)." (Lambert and Barnett 2021:193)

"also be used to allow for varying patterns in a trajectory, for instance, by fitting person-specific growth curves to repeated contacts data on a stratification outcome (e.g. Jenkins 2011), or in the impact of a stratification measure upon a pattern of growth." (Lambert and Barnett 2021:193)

"Interestingly, there is a natural longitudinal relationship between intersectional conjunctions of inequality, and social stratification position - for instance, the way in which life-course trajectories of social class or stratification position develop may well be different for different groups defined by characteristics such as gender and ethnicity (e.g. Blossfeld et al. 2011)." (Lambert and Barnett 2021:194)

"Weldon (2006) suggests four ways in which intersectionality might be conceptualised empirically: (i) an 'additive' approach, that fits parameters for each different inequality category, suggesting a 'double jeopardy' conceptualisation of accumulating inequalities (typically achieved by modelling dummy variable indicators of each category membership); (ii) a 'multiplicative' approach, that fits both main effects, and interaction effects, for inequality categories, suggesting a 'mutually reinforcing' conceptualisation of intersecting inequality's; (iii) an approach described as 'intersectional only', where intersectionality is conceptualised at something 'qualitatively different' that might only be assessed in a qualitative research design; and (iv) an approach labelled 'intersectionality plus', which allows for any of the above three permutations and allows for their effects to vary over time and place." (Lambert and Barnett 2021:194)

"In quantitative research" (Lambert and Barnett 2021:195)

"theoretical terms (ii) is often preferred" (Lambert and Barnett 2021:195)

"some have also advocated an 'intercategorical' approach (McCall 2005), concerned with comparisons between and within multiple groups. Typically the intercategorical approach involves fitting a model and fully saturating it with all possible interaction effects (or with separate dummy variables for every possible combination)" (Lambert and Barnett 2021:195)

"Evans et al. (2018) demonstrate how fitting 'random effects' ('multilevel models') for the distinctive groups can provide a parsimonious device to this end" (Lambert and Barnett 2021:195)

"measures of stratification position might reasonably be standardised within a suitable context, whether by pre-analysis standardisation within a category (such as calculating separate stratification scales for men and women - e.g. Prandy 1986)" (Lambert and Barnett 2021:195)

Like different CAMSIS scores (note on p.195)

"It is often thought that occupation-based measures, specifically, are inflexible -" (Lambert and Barnett 2021:196)

"an important occupation-based social class measure such as ESeC, for instance, is purportedly fixed in time and context, and uses categories that are hard to compare between contexts, yet this doesn't mean that all occupation-based measures work in this way" (Lambert and Barnett 2021:196)