domingo, 1 de julio de 2012

Cortito 7 "Fluidity in Chile and USA (when measurement (and time) matters)"

I will begin from the end to end with the beginning.
Torche’s article is rather impressive and furthermore it will become a classical work in Chilean sociology giving that this is the first mobility study regarding Chile published in a journal of high impact. Was it both apish and risky, however, to acknowledge that Chile—even though its level of inequality is rather high—is more fluid than USA and, of course, “more fluid than any of the advanced countries” (Torche, 2005:441), without making any reference to a potential case of structural mobility but also because fluidity depends on both conventions and time?[1] Certainly my question has a rhetoric tone; however, this question raises the issue of whether metrics can, after all, be a representation of reality or normative ideas.
One of the onymous reviewers of Torche’s article was Richard Breen (Torche, 2005:421). Both scholars were already aware of the very weak relationship which can be established between inequality and fluidity. Torche states the foregoing in p. 426, and Breen (and Luijkx) in the following quote informs us that  
“Our model posits that, given the level of income inequality in each country in the 1970s, changes within a country in β follow the same trend as the Gini index. Thus we test for a common effect on b, in all countries, of income inequality: or, in other words, a common slope coefficient (…) In the second model we include income inequality, as measured by the Gini index. Not only is there no significant relationship between fluidity and the Gini index, its coefficient has the wrong sign.” (Breen and Luijkx, 2004:396, my emphasis)
If the relationship was acknowledged to be problematic by Torche, or in more polite terms inconclusive (p.426), why insist in this matter? Or in other words, why to build a work based on a hypothesis that does not have strong empirical support? What is the (theoretical or methodological) value of presenting Chile as a counter case of the indirect relationship of these two variables? In order to falsify any direction of this hypothesis it seems that not only several decades of research are required (as Breen’s book suggests) since a research program of this sort entails the collection of several generational trends that cross sectional cohort analyzes cannot assess—or at least they should have enough large Ns to be representative of each generation, as Blau and Duncan’s work suggested (1967). More particularly, under what criteria was appropriated to assess fluidity in contrast to inequality in Chile from a cross-sectional survey whose N reported was 3,002 (p.432), however the N of each cohort (1964-1973; 1974-1988; 1989-2000) (p.441-442) was not reported. This latter criticism can be unfair because I am using a methodological argument—similarly as Hauser and Logan’s (1992) cross-validation argument to disprove Rytina’s SSCI—to disregard Torche’s findings rather than study her main argument i.e. Chile is an adjudicative case since its level of inequality is one of the highest in the world, yet its fluidity surpasses USA and several other industrialized countries. In sum this observation attacks the inappropriate use of inequality as a variable which can be associated to fluidity.
However behind this observation another element can be discussed i.e. Chile’s fluidity. I am inclined to think that Chile’s high fluidity is capitalizing from its pattern of late urbanization (if compared to urbanization processes which some industrialized countries went through). For instance as we can observe in table 1, there has been an indeed dramatic increase of urbanization in Chile for the 1950-2000 period, and the latter can be associated to the “reverse” spillover of the 444 cases of class VIIb (Farmer workers) and the 316 cases of class IVc (Farmers) (p.433).
Table 1 Chile´s percentage of urbanization (1950-2000)
Percentage of urbanization
 Source: Alfredo Lattes, Urbanizacón, Crecimiento Urbano y Migraciones en América Latina UN-ECLAC (2000)
The massive dropping of both classes from 15% to 8% in class VIIb, and from 11% to 4% in class IVc denote a very fluid country. But interpretation of fluidity, if one accepts the class typology proposed by Erikson and Goldthorpe, must be then done cautiously. More precisely movements from origin to destination might not at all occur due to competition between individuals which fight for scarce resources since what we observe is the expansion of the service sector and rural-urban immigration.
While it is indeed attractive, and actually promising, to think of different directions between these two variables as it is suggested in figure 1 (complementing in fact Friedman’s theory (1962)), pursuing with this enterprise can be nonetheless misleading or at best really expensive. Specifically, to obtain values for the fluidity variable requires the introduction of several ad-hoc effects to standardize the metric for very low Ns (i.e. countries),[2] and therefore it stretches several assumptions which need to account for internal differences. Furthermore, if one variable of a “simple” bivariate analysis presents problems of operationalization, it would be then even more difficult to convey one of the four suggested functions (figure 1).
A last point regarding Torche’s analysis is the metric of distance among classes based on the implicit hierarchical and therefore normative definition of class structure proposed by Erikson and Goldthorpe. While Erikson and Goldthorpe acknowledge that countries have their own economic dynamics, which can be captured by adding the four effects—elements which are very well developed by Torche—the distance ultimately depends on who or better what occupation occupies what class. The problem with this strategy however, as we will see with Rytina’s SSCI, is that these distances are arbitrary because certain consensus, in this case mostly European,  capture a pre-determined social order. Two examples can illustrate the latter, a university professor in Chile compared to a university professor in England, is less likely to have, ceteris paribus, higher income, or an itinerary seller, that is an individual who sell products in the street particularly to car drivers or passengers when the traffic light is red, populates the membership of self-employed category and this occupation is likely to be inexistent in England.
Figure 1 Four theoretical macro-relations between ‘Inequality’ and ‘Fluidity’

Rytina’s proposal deals with certain scale conventions in occupational mobility which can be regarded as problematic. While of high practical value, these metrics can ultimately be misleading since they only capture a lower part of how occupation mobility is distributed but also this distribution is likely masking the inertia of the occupations and favoring findings of achievement. Both SEI and prestige scales are constructs which have been widely used, however, part of their operationalization in trying to capture normative orders (what is conceived to be high or low in terms of honor) and these cannot be regarded as universal (or at least require a big deal of contextualization). More specifically the imposition of arbitrary scores to nominal categories of occupations by introducing two pieces of information such as income and years of education or by assigning subjective scales ultimately implies that what is being captured as mobile or rigid are only scores which represent constructions, rather than a difference between destination and origin based on more pure information i.e. the occupation. According to Rytina the key is to identify the metric which grasps the distance between one value and another (one occupation versus another) in order to regard these several movements as far or close. This metric is critical because it ultimately reflects mobility or rigidity. The question then is what type of values (scores) should be considered in order to assess differences between two occupations while simultaneously avoid the imposition of arbitrary scores? In order to build an algorithm which can distribute values according to the distance between origin and destination Rytina acknowledges that a similar attempt, based on the technique of canonical scale has been done by several authors, however one potential drawback is that the “scale scores of father’s DOCs and offspring’s DOCs need not be equal for asymmetric tables”. The algorithm proposed by Rytina solves the challenge of finding both a vector  “such the correlation r of father with offspring is a maximum” (1992a:1686) and “” (ibid). The values obtained from this formalization are the Symmetric Scaling of Intergenerational continuity (SSIC) scores.
Rytina tests three (actually five) scales that compete for the understanding of occupational mobility: prestige, SEI and SSIC and through several correlations (1992a:1669, see table 1) we can observe that these scales measure very similar patterns. However, the advantage of the SSCI over the SEI scale is that the latter corrects SEI scores by “repositioning occupations to a closer correspondence with the details of ascent and descent” (1992a:1671). In other words SEI scale is less faithful to grasp the feature of occupational mobility, and therefore conclusions from its application can be misleading. In order to test this last statement Rytina reassesses the status transmission process (father’s occupation (a), father’s occupation on education (b), and education (c)) by using five scales. It can be seen that the direct effect of father’s rank on offspring rank is larger than it was commonly accepted (1992a:1674). Or in other words education is not “the predominant generative source fir the intergenerational stability of occupational rank”. The implications of this finding can be for many uncomfortable because there is a possibility that achievement is a product of more complex processes than individual performance, but also because it entails to change a specific habitus of scaling occupations. Nevertheless, we can see from Hout and DiPrete that “education is [still regarded as] the main factor in the intergenerational reproduction of social standing because the product ac is greater than the direct effect of origins (b)” (Hout and DiPrete, 2005:6).[3]That statement is unfortunate because regardless of how strongly in empirical, methodological and theoretical terms education was proven to be imprecise.
Naturally SSCI should be tested in order to assess its generalizability. Rytina actually observes that one potential source of falsification is the application of this scale to other samples (1992a:1676) and that is actually the route taken by Hauser and Logan (1992). While Rytina observes that one of the main criticism from these authors is the lack of cross-sample validity (1992b:1729), I think another more subtle criticism which these two author also raised is how time is ultimately not captured in SSCI. That is why they use two different times of measurement: 1972-86 and 1987-90 (p.1695 and 1699-1700).
What I mean by time in this case is the theoretical assumption that mobility or fluidity can change in two or more different points of time and that is how we can eventually observe that one the four functions, between fluidity and equality can be conceived. This strategy would allow us to understand whether mobility has increased, remained constant or decreased. Rytina, I think is more explicit about this challenge in 2000, and definitely tackles the issue in 2008.  One elements that Rytina highlights as imprecise in Hauser and Logan’s application and interpretation of SSIC’s scores is that values from one period of time cannot automatically be taken, or imposed to another period of time i.e. an older survey (1992b:1739-1741). In other words SSIC by definition cannot be taken as permanent. What the researcher should ultimately assess is the method of scaling, and therefore avoiding the reification of scales, because there are local elements which also constrain the process of stratification (Rytina, 2008). The risk of taking this approach however is high because a fixed variable seems to protect the illusion of parsimony and by introducing new information to the network of (social stratification) scholars reaction is likely to be of resistance. Lastly, Rytina in 2000 develops the SSIC further and findings are less promising for those who have education to be more decisive than father’s occupation. Furthermore, this methodology also captures massive changes in occupational stratification from 1987 and onwards. The implication of this last finding seems obvious, but it is less precise if the SEI were taken as the proper scale because its values are fixed, or do not depend on distances which change in reference to specific contexts. The development of this study actually is a more solid answer to the core criticism presented by Hauser and Logan because data is taken from different time periods (1972 to 1990)[4] (2000:1236). Perhaps now is time to bring some form of SSIC to another country, I guess Chile can be a good option.

Breen, R. and R. Luijkx. 2004. “Conclusions.” Chapter 15 in Social Mobility in Europe, edited by Richard Breen. Oxford, UK: Oxford University Press.

Friedman, M. 1962. Capitalism and Freedom. Chicago: University of Chicago Press
Hauser, R M. and J. A. Logan. 1992. “How Not to Measure IntergenerationalOccupational Persistence.” American Journal of Sociology 97(6):1689-711.

Hout, T. A. DiPrete. 2006. What we have learned: RC28’s contributions to knowledge about social stratification? Research in Social Stratification and Mobility 24, 1–20.

Lattes A. 2000. Urbanización, Crecimiento Urbano y Migraciones en América Latina [Urbanization, Urban Development and Migration]UN-ECLAC.

Rytina S. 2000. Sticky Struggles. . 2008.

________, 2000. Is Occupational Mobility Declining in the U.S.? Social Forces, Jun., vol. 78, no. 4, p. 1227-1276.

________, 1992a.Scaling the Intergenerational Continuity of Occupation: Is Occupational Inheritance Ascriptive After All?” American Journal of Sociology, v97 n6.

________, 1992b. Response to Hauser and Logan and Grusky and Van Rompaey. American Journal of Sociology 97:1729-48.

Torche, F. 2005. Unequal but Fluid: Social Mobility in Chile in Comparative Perspective. American Sociological Review, Jun., vol. 70, no. 3, p. 422-450.

[1] Author’s emphasis.
[2] This indeed could be corrected by introducing years per country, however, as Sorokin’s analysis implicitly suggests capturing this type of variability can take up decades.
[3] I am deliberately using this quote because Hout and DiPrete suggest this finding to be one of the 19 empirical generalizations which about 40 researches arrived to after 55 years of social stratification research.(Hout and
[4] 1979 and 1981 were the only two years where surveys were not carried out (2000:1236).

No hay comentarios:

Publicar un comentario