This section explores whether the results are sensitive to the selected estimation model (ordered probit, equation (7) ). Tables 10 and 11 present the linear fixed-effects regression model for total consumption and consumption categories, respectively. This corresponds to estimating equation (3). The results are robust to model specification. As before total consumption is significantly associated with LS and H, but not D, and income is only significant for LS. It turns out that the linear specification favors additional significant consumption categories: food & beverage at home and car-related expenses for LS and household products for H. Otherwise, results are widely unaffected.