Skip to main content

Table 7 Conditional response probabilities and latent class sizes for the five-class of subjective social status

From: Discordance between subjective and objective social status in contemporary China

  Respondent type of subjective social status
Manifest variables Upper Upper middle Middle Lower middle Lower
Education      
 Primary school and below 0.034 0.200 0.00 0.379 0.755
 Junior high school 0.024 0.439 0.411 0.510 0.230
 Senior high school 0.135 0.328 0.517 0.106 0.031
 Tertiary 0.806 0.033 0.072 0.005 0.001
Income      
 1st level 0.007 0.065 0.034 0.100 0.500
 2nd level 0.018 0.054 0.067 0.235 0.222
 3rd level 0.057 0.117 0.156 0.293 0.132
 4th level 0.153 0.262 0.267 0.225 0.078
 5th level 0.250 0.269 0.233 0.102 0.040
 6th level 0.515 0.233 0.243 0.044 0.028
ISEI      
 1st level 0.007 0.000 0.052 0.435 0.888
 2nd level 0.016 0.350 0.065 0.228 0.047
 3rd level 0.083 0.309 0.241 0.171 0.027
 4th level 0.266 0.307 0.328 0.116 0.017
 5th level 0.628 0.035 0.313 0.050 0.021
 6th level 0.000 0.000 0.000 0.000 0.000
Class size (assigned) 12.28 % 6.08 % 21.15 % 30.59 % 29.90 %
  1. The conditional probabilities indicate the probability that an individual in a certain latent class will score a particular way on a manifest variable (McCutcheon 1987, p. 33). A respondent belonging to Class 1, for instance, has a chance close to 97 % of turning to personal contact who is “very much” trusted. Then we assign the respondents to the five latent classes to create a new categorical variable representing one’s objective status. According to McCutcheon (1987, p. 36), the assignment of observations to latent classes is probabilistic and therefore involving some error. Although not reported here, in our case, the two measures to estimate the expected proportion of classification errors and the closeness of association between the latent variable and the joint variables shows that no big uncertainty would appear during the process of assignment. As can be seen from the table, C1 has the highest probabilities of reporting the highest level of education, income, and occupations, while C5 has the lowest probabilities of these. We therefore label C1 through C5 upper class (12.28 %), upper middle class (6.08 %), middle class (21.15 %), lower middle class (30.59 %), and lower class (29.90 %), respectively