In other words, the analysis
neglects structural uncertainty about the adequacy of the assumed linear model, and the parameter uncertainty the analysis does take into account is strongly reduced by models that are «bad» by this model - data mismatch metric.