Table 4.
Wealth Gradient in HIV
| HIV-Positive | Burkina Faso | Cameroon | Ghana | Kenya | Tanzania |
|---|---|---|---|---|---|
| A. Quadratic With Age Group Controls | |||||
| Wealth index | 0.0266 (0.0349) | 0.0861 (0.0493) | 0.0105 (0.0268) | 0.0389 (0.0622) | 0.1613* (0.0486) |
| Wealth index, squared | 0.0132 (0.0450) | −0.0747 (0.0610) | −0.0289 (0.0296) | −0.0378 (0.0785) | −0.1659* (0.0570) |
| F statistic | 4.13 | 2.62 | 2.44 | 0.24 | 5.66 |
| p value | .02 | .07 | .09 | .78 | .00 |
| B. Quadratic With Age Group Controls, Rural Indicator | |||||
| Wealth index | 0.0038 (0.0321) | 0.0258 (0.0504) | −0.0026 (0.0279) | 0.0350 (0.0621) | 0.1196* (0.0486) |
| Wealth index, squared | 0.0072 (0.0453) | −0.0477 (0.0610) | −0.0254 (0.0296) | −0.0790 (0.0800) | −0.2095* (0.0584) |
| F statistic | 0.25 | 0.53 | 3.72 | 0.88 | 8.40 |
| p value | .78 | .59 | .02 | .41 | .00 |
| C. Quadratic With Age Group Controls, Rural Indicator, Region Fixed Effects | |||||
| Wealth index | −0.0014 (0.0345) | −0.0032 (0.0516) | −0.0286 (0.0288) | −0.0177 (0.0689) | 0.0986* (0.0492) |
| Wealth index, squared | 0.0096 (0.0462) | −0.0124 (0.0626) | −0.0033 (0.0306) | −0.0169 (0.0860) | −0.2042* (0.0605) |
| F statistic | 0.09 | 0.28 | 4.48 | 0.64 | 8.76 |
| p value | .91 | .76 | .01 | .53 | .00 |
| Number of Observations | 7,137 | 9,750 | 9,142 | 5,988 | 10,742 |
Source: DHS for Burkina Faso (2003), Cameroon (2004), Ghana (2003), Kenya (2003), and Tanzania (2003).
Notes: All regressions are weighted least squares regressions, weighted using provided HIV sample weights with clustering on the household. The dependent variable is an indicator for whether the respondent is HIV-positive. The F statistic tests the joint significance of the wealth index and the wealth index, squared. The wealth index is the fraction of nine assets or amenities that the respondent’s household has; the assets and amenities are radio, television, refrigerator, bicycle, motorcycle, car, telephone, electricity, and a flush toilet or pit latrine (must have nonmissing values for at least five components to be included). All regressions include a constant and an indicator for whether the respondent is female. Age group controls are five-year age groups. Huber-White standard errors are in parentheses.
p < .05