Estimation of the Stereotyped Ordered Regression Model 

 While reading Xiaogang Wu and Donald Treiman's paper entitled "Inequality and Equality under Chinese Socialism: The Hukou System and Intergenerational Occupational Mobility" in  American Journal of Sociology  (2007, 113: 415-445) , I was directed to a technical paper written by John Hendrickx (2000), describing how to use "mclgen" and "mclest" in Stata to estimate the Stereotyped Ordered Regression Model (SOR) in social mobility studies. 

 SOR is similar to conventional ordinal Logit models, but with a scaling metric to scale the effects of the independent variables on the dependent variables so that the effects of an independent variable vary by the values of the dependent variable. In addition, SOR does not assume strict ordering among values of the dependent variable, which is perfect for studying occupational mobility as occupation is orderable but without strict order. Another desirable property that SOR has is that it specifies an inheritance parameter measuring intergenerational occupational immobility, i.e., the extent to which father and son have the same occupation.  

 These features make SOR appear to outperform ordinal Logit models in social mobility studies.  

  Click here  to consult Hendrickx' paper for more details of the SOR model and the syntax of using "mclgen" and "mclest" in Stata.