Page 309 - Contributed Paper Session (CPS) - Volume 6

CPS1937 Xu Sun et al.
3. Data, modeling and results
We use data which the GSS collected information on father’s occupational
status-1988-2010, and thehe sample space was restricted to respondents who
were ages 25-64 and were in the labor force. Melamed (2015) used this data,
and 13746 respondents (include 6864 female respondents and 6919 male
respondents). Respondent and parental occupational classifications were
recorded in 1980 census codes by the GSS, and were then transformed into
social classes using a standard measure (Erickson & Goldthorpe, 1992).

Table 1. Social class positions used in the analyses
___________________________________________________________________________
I Professionals, administrators, officials, and managers, higher level
II Professionals, administrators, officials, and managers, lower level
III Routine non-manual and service workers, higher and lower levels
IV Self-Employed, with or without employees
V Technical specialists and supervisors of manual workers, skilled
manual workers
VI Semiskilled and unskilled manual workers, nonfarm and farm
______________________________________________________________________________
Table 1. illustrates the six class positions, and Table 2. presents the social
mobility tables that are analyzed in this paper. Below models and community
structures will be presented for female, male, and all respondents.

Beginning with the female respondents, the community structure analysis
proceeds as follows: first, Afrom above is defined as the observed mobility
table, and then P is defined as the fitted values from an “independence” model
that are estimated. The matrix B is defined as A-P, then B is compiled into the
block-off-diagonal matrix Z. Following Newman (2006a), taking the
eigenspectrum decomposition of Z sheds light on the community structure of
A net of P. Specifically, the signs of the entries in the eigenvector associated
with the largest eigenvalue indicate the optimal two community split, and
subsequent eigenvectors may similarly be used to determine more than two
communities. Using the quality function modularity (Q), the locally defined

298 | I S I W S C 2 0 1 9