Table 2.

Survey data and ordered probit regressions: pooled sample.

	(1)	(2)
	SVY: ordered probit	Ordered probit
Variables	Self_Rank_Poverty_2016	Self_Rank_Poverty_2016
ln (Total_inc2016)	0.179*** (0.0167)	0.174*** (0.0168)
BMI2016	−0.0260*** (0.00454)	−0.0257*** (0.00450)
$Cu{t}_1$ ( $1\le \mathit{Self}\_\mathit{Rank}\_\mathit{Poverty}\_2016\le 2$ )	0.249 (0.212)	0.200 (0.216)
$Cu{t}_2$ ( $2\le \mathit{Self}\_\mathit{Rank}\_\mathit{Poverty}\_2016\le 3$ )	0.594*** (0.213)	0.546** (0.216)
$Cu{t}_3$ ( $3\le \mathit{Self}\_\mathit{Rank}\_\mathit{Poverty}\_2016\le 4$ )	0.902*** (0.213)	0.856*** (0.217)
Observations	4,017	4,017
Population	102,900.72	-
Strata	4	-

The estimation is based on the pooled sample of females and males obtained from the ICBS survey. The dependent variable (Self_Rank_Poverty_2016) is the self-ranking of individuals in response to the question: “During the last 15 years, how often did you consider yourself poor?” Possible answers are 1 = “often,” 2 = “sometimes,” 3 = “rarely,” 4 = “never.” The black (grey) line in the corresponding Figure 2 is the projected probability to answer 1 = “often” (4 = “never”) as a function of the BMI (= $\frac{\mathrm{kg}.}{{\mathrm{meter}}^2}$ where 18 ≤ BMI ≤ 25 is normal weight; 25 ≤ BMI < 30 is overweight; BMI ≥ 30 is considered obesity – see the links to the WHO websites in the reference list). Column (1) reports the outcomes of the survey data analysis. The analysis includes four strata based on the four categories of the variable Self_Rank_Poverty_2016. The weight given to each observation is based on the inverse of the BMI variable. Column (2) reports the outcomes where equal weight is given to each observation. In column (1)/(2) The linearized/conventional standard errors are given in parentheses ***p < 0.01.