More science, less exegesis, please: a rejoinder

Key takeaways

content: "@chanMoreScienceLess2019" -file:@chanMoreScienceLess2019

Reading notes

Imported on 2024-06-10 13:11

⭐ Important

• & I note that Reeves and de Vries (2019) have recently provided some answers to Karademir-Hazir’s question. Using Understanding Society data, they show that, net of socio-demographic background, education, social class, and so on, taking part in one more type of cultural activity is associated with an earnings increase of £18 to £25 per month (Reeves and de Vries 2019: 228, Table II). This return to cultural consumption seems pretty modest to me. Moreover, it is probably an overestimate as Reeves and de Vries’s model is vulnerable to omitted variable bias. (p. 915)
• & For example, ‘rock, pop and jazz’ is a very broad and heterogeneous indicator. I would love to have more fine-grained measures, such as separate indicators for ‘jazz’ and for ‘rock and pop’. But, imperfect as this indicator is, I would maintain that most people do not see ‘rock, pop and jazz’ as being on a par with classical music or opera. (p. 915)
• & And I would be very surprised if video games, rap music and street art are widely regarded as forms of highbrow culture. (p. 916)
• & Lambert also questions my LCA. Because he sees omnivorousness as ‘more naturally a continuous dimension’, he argues that my tripartite division of the respondents into omnivores, paucivores, and univores ‘might diminish the accuracy of the relevant estimated associations’. (p. 916)
• & There are indeed different ways to think about and measure omnivorousness. I use LCA for two reasons. First, scholars in the field often refer to distinct types of cultural consumers. Secondly, the conditional probabilities of the LCA solution allow you to check, quite directly, whether there are people who are likely to consume both highbrow and popular culture. This speaks to a key insight of Peterson, namely the boundary-crossing nature of omnivores’ cultural consumption. Having said that, it is possible to think of everyone as being a little more or less omnivorous, rather than of one type or another. One way to capture this idea is to count just how many types of cultural consumption people do. Let’s call this the volume measure of omnivorousness. Whether the two measures (i.e., latent class and volume) make any difference to the results is an empirical matter. (p. 916)
• & There is, for example, no overlap at all in their interquartile range. Given this, I would argue that thinking in terms of three discrete latent classes is not unreasonable. I have re-run some (but not all) regressions reported in the paper using the volume measure of omnivorousness, and have obtained very similar results. I strongly suspect that using the volume measure that is favoured by Lambert will not change the overall conclusion of my paper. (Perhaps this could be a dissertation project for an undergraduate or an MSc student with some basic quantitative skills. Any taker?). (p. 916)
• & Let me quickly deal with some other points raised by Lambert. All regression models reported in the paper take age into account. So the confounding between age and education is not as problematic as he suggests. (p. 916)
• & Baumann is very complimentary of both my paper and FJR’s:‘tremendous data sets’, ‘careful, highly expert approach to data analysis’, ‘significant advances in our understanding’, and so on. But he flatly rejects the conclusion that I draw: ‘it is not clear that we can draw the conclusion that if omnivorousness is about cosmopolitanism and openness, it is therefore not about status distinction ... omnivorous cultural consumption simultaneously expresses egalitarian and democratic impulses and status distinction’. Baumann is, of course, free to define distinction however he wants. It is hard for me to imagine a form of distinction that draws symbolic boundaries on the one hand and ‘expresses egalitarian and democratic impulses’ on the other. But perhaps that just reflects a lack of imagination on my part. That said, if distinction of this ‘egalitarian and democratic’ type exists, it would be quite different from how Bourdieu sees it. A fair reading of the following passages would suggest that distinction, as articulated by Bourdieu, involves a good deal of condescending attitudes. (p. 917)
• & Let us turn to Lizardo’s comments which mainly address my reply to FJR, ‘Understanding Social Status’ (Chan 2019b). Lizardo cannot make up his mind about my work on class and status. ‘A lot of important results came out of this work’, he says. But he also thinks that the class–status debate is ‘moribund’. (p. 919)
• & The methodological contrast that Lambert, Lizardo and Wuggenig refer to is way over-drawn.This is because, as Chan and Goldthorpe (2007) observe, multiple correspondence analysis is the categorical counterpart of principal component analysis (see Gower and Hand 1996: ch. 4). Indeed, Goodman (1996: 408) points out that many tools for analysing the association in contingency tables, including log-linear models and correspondence analysis ‘can be viewed as special cases of [a] general method of analysis’. (p. 920)
• & More substantively, Lizardo’s criticism of the class–status distinction relates to his adoption of David Grusky’s microclass proposal. (p. 920)
• & It is beyond the scope of this rejoinder to assess Grusky’s microclass approach in any detail (but see Erikson, Goldthorpe and Hällsten 2012). However, consider this extraordinary claim of Lizardo: ‘the worst thing American stratification researchers ever did was to lump together these really nice, socially and phenomenologically valid classes ... to create little 5 × 5 tables featuring such meaningless lumps as “Managerial/Professional” all so they could fit a loglinear model and get an L-squared statistic’. Can it possibly be true that ‘managerial/professional’ is a meaningless category, or that fitting loglinear models to 5 × 5 mobility tables was the worst thing that stratification researchers (American or otherwise) ever did? (p. 921)