Modelling has to be a core skill for any research analyst in any profession. The ability to test and evaluate different hypotheses is crucial to any understanding of a problem. This does not have to be a complex process, although for more taxing problems tools like Microfit or GLIM can be very useful, much of the more mudane construction and use of models takes place using Excel or Lotus, making the tools available for everyone.
Of course, one can only really model data in which the structure is fully understood, harping back again to skills in data analysis.

Throughout my career I have been involved in all levels of modelling from simple space planning using different density assumptions through to relatively complex econometric models using labour market variables to project demand for different types of physical space, then rolling this on to forecast movement in rental values.