TY - JOUR
T1 - Class careers as sequences
T2 - An optimal matching analysis of work-life histories
AU - Halpin, Brendan
AU - Chan, Tak Wing
PY - 1998
Y1 - 1998
N2 - We apply optimal matching techniques to class careers from age 15 to age 35 for two moderately large samples, as a means of exploring the utility of this sequence-oriented approach for the analysis of work-life social mobility. We first apply multi-dimensional scaling techniques to the inter-sequence distances generated by the optimal matching algorithm in order to test whether the technique locates sequences in a coherent and interpretable space. We find the space to be highly patterned and reasonably interpretable. Next we run the two moderately large samples (each approximately 1500 sequences) through the analysis and examine the nature of the set of clusters that emerges. We find the clusters to be distinct and an intuitively attractive grouping of the sequences. Finally, we consider how the clusters are distributed across cohorts: the distributions change markedly, though this is largely due to changes in the distribution of classes over time. We briefly discuss means of separating 'pure sequence'change from change in the gross 'class time-budget'of cohorts, and consider means of applying statistical models to the problem. We conclude by endorsing Optimal Matching Analysis, especially as a means of exploratory analysis of longitudinal data.
AB - We apply optimal matching techniques to class careers from age 15 to age 35 for two moderately large samples, as a means of exploring the utility of this sequence-oriented approach for the analysis of work-life social mobility. We first apply multi-dimensional scaling techniques to the inter-sequence distances generated by the optimal matching algorithm in order to test whether the technique locates sequences in a coherent and interpretable space. We find the space to be highly patterned and reasonably interpretable. Next we run the two moderately large samples (each approximately 1500 sequences) through the analysis and examine the nature of the set of clusters that emerges. We find the clusters to be distinct and an intuitively attractive grouping of the sequences. Finally, we consider how the clusters are distributed across cohorts: the distributions change markedly, though this is largely due to changes in the distribution of classes over time. We briefly discuss means of separating 'pure sequence'change from change in the gross 'class time-budget'of cohorts, and consider means of applying statistical models to the problem. We conclude by endorsing Optimal Matching Analysis, especially as a means of exploratory analysis of longitudinal data.
UR - http://www.scopus.com/inward/record.url?scp=0039992458&partnerID=8YFLogxK
U2 - 10.1093/oxfordjournals.esr.a018230
DO - 10.1093/oxfordjournals.esr.a018230
M3 - Article
AN - SCOPUS:0039992458
SN - 0266-7215
VL - 14
SP - 111
EP - 130
JO - European Sociological Review
JF - European Sociological Review
IS - 2
ER -