Forty-Years Change of Dimensionality of Occupational Evaluation in Japan
Ryuhei Tsuji, Namie NagamatsuProblem:
Subjective evaluations of occupations are known to be stable over time (Hodge et al., 1964; Nakao and Treas, 1994). For example, people have evaluated medical doctors, lawyers, or college professors higher than any other occupations over the decades. Japanese evaluations of occupations in 1975 and 1995 are correlated as high as r>.95 (Hara, 1999). However, a high correlation does not necessarily mean that the evaluations are constructed uni-dimensionally. Tarohmaru (1998) tackled this problem using SSM Survey data (see below) collected in 1995 and found that the evaluations were not uni-dimensional. However, his method to distinguish whether the evaluation of a pair of occupations were different was problematic in that setting the value of cut-off was ad hoc.
This paper addresses the problem of whether the subjective evaluations of occupations (OEs) are comprised uni-dimensionally.
We use Occupational Prestige Survey (OPS) data that is a part of SSM Survey data collected in 1975 and 1995, and OPS data in the fiscal year 2016. SSM stands for Social Stratification and Social Mobility, and the survey has been conducted every ten years since 1955, which is known as one of the largest sociological surveys in Japan. OPS had been conducted as a part of SSM survey in 1975 and 1995. However, OPS was not included in SSM 2015 and was undertaken separately in the fiscal year 2016.
The occupational prestige in OPS was measured as follows. Respondents were asked to evaluate the magnitude of occupation in five-point scales, from 1 as highest to 5 as lowest. The five-point scales were then recoded as 100 for 1, 75 for 2, down to 0 for 5. The mean of each occupation is its occupational prestige (whose range is between 0 and 100). However, although it has been traditionally called occupational prestige, it is argued that it is just a general evaluation of each occupation. Therefore, we call it an occupational evaluation (OE).
We applied Wilcoxon's signed-rank test to all pairs of occupations measured in OPS, and created the adjacent matrix of occupational evaluations E. If an occupation i is evaluated higher than the other j, Eij=1 and Eji=0. If the two occupations are evaluated about the same, both Eij=0 and Eji=0.
Then, we draw a Hasse diagram to show the partial order structure of occupational evaluations. If the evaluations are uni-dimensional, the Hasse diagram will be a straight line; if the evaluations are not uni-dimensional, the Hasse diagram will be a more complicated figure.
We analyzed SSM75, SSM95, and OPS16 data. Since OPS16 data were collected using three different questionnaires that contain different sets of occupations, we analyze three data separately.
We found that Hasse diagrams of SSM75 and SSM95 were straight line; i.e., the occupational evaluations at those times were uni-dimensional. However, all of the three OPS16 Hasse diagrams were not straight lines; i.e., the occupational evaluations in recent years are not uni-dimensional.
It may indicate that the Japanese criteria of occupational evaluation have become more diverse in these 20 years.