Table 5.
Multilevel models of the black-white difference in census tract percentage non-Hispanic black: Black and white respondents from the 1981, 1991, and 2001 waves of the Panel Study of Income Dynamics (standard errors in parentheses)
| Independent Variables | Model 1 | Model 2 | Model 3 | Model 4 |
|---|---|---|---|---|
| Intercept | 6.017*** (.477) |
5.439*** (.641) |
7.385*** (1.334) |
9.574*** (1.380) |
| Black | 32.906*** (2.082) |
37.463*** (2.137) |
33.675*** (2.576) |
36.623*** (2.924) |
| Fixed Effects on Black-White Difference | ||||
| Year 1991 (ref. = 1981) | −4.045*** (1.051) |
−2.079† (1.073) |
−3.185 (2.037) |
|
| Year 2001 (ref. = 1991) | −4.570*** (1.098) |
−3.907*** (1.101) |
−4.351* (1.832) |
|
| Individual Characteristics | ||||
| Age 30–39 (ref. = 18–29) | −0.728 (1.208) |
−0.480 (1.200) |
||
| Age 40–49 (ref. = 18–29) | 1.357 (1.390) |
1.446 (1.381) |
||
| Age 50–59 (ref. = 18–29) | 5.997*** (1.596) |
6.191*** (1.584) |
||
| Age >60 (ref. = 18–29) | 3.163† (1.673) |
3.806* (1.661) |
||
| Female (1 = yes) | −0.113 (1.167) |
−0.063 (1.159) |
||
| Married (1 = yes) | −3.714** (1.226) |
−3.603** (1.217) |
||
| Number of children | 0.587 (0.468) |
0.558 (0.464) |
||
| Homeowner (1 = yes) | 2.892** (1.077) |
2.766** (1.069) |
||
| Persons per room | 2.136 (1.433) |
2.520† (1.421) |
||
| Education (in years) | −0.700*** (0.164) |
−0.728*** (0.162) |
||
| Family income ($1,000s) | −0.110*** (0.016) |
−0.107*** (0.016) |
||
| Employed (1 = yes) | 0.051 (1.200) |
0.158 (1.192) |
||
| Metropolitan Characteristics | ||||
| Population size (ln) | 7.887*** (1.843) |
|||
| % non-Hispanic black | 0.684*** (0.179) |
|||
| % Hispanic | −0.627*** (0.158) |
|||
| % Asian | −0.152 (0.279) |
|||
| % in poverty | 0.516 (0.372) |
|||
| % of labor force in manufacturing | 0.148 (0.237) |
|||
| % of labor force in local, state, or federal government |
−0.572* (0.272) |
|||
| % new housing | −0.107 (0.114) |
|||
| % living in suburban area | −0.022 (0.094) |
|||
| Municipal fragmentation | 7.122 (8.101) |
|||
| Variance Components | ||||
| Level 2: τ00 Random intercept (MSA) | 17.87 | 18.16 | 18.53 | 2.41 |
| Level 2: τ11 Random slope (black) | 446.27 | 425.05 | 372.08 | 199.43 |
| Level 2: τ01 Covariance (intercept, slope) |
20.96 | 21.62 | 16.87 | −7.52 |
| Level 1: σ2 Residual variance | 408.51 | 405.02 | 389.15 | 385.76 |
| Pseudo-R2 Level 2 Intercept | .00 | .00 | .87 | |
| Pseudo-R2 Level 2 Slope | .05 | .12 | .46 | |
| Pseudo-R2 Level 1 Residual | .00 | .04 | .00 | |
| Likelihood Ratio Test χ2 (df) | 933.60*** (2)a | 81.426*** (4) | 379.15*** (24) | 267.19*** (20) |
| Log-Likelihood (df) | −39,006 (6) | −38,965 (10) | −38,776 (34) | −38,642 (54) |
| AIC | 78,024 | 77,950 | 77,619 | 77,392 |
| BIC | 78,066 | 78,021 | 77,860 | 77,774 |
| N | 8,767 | 8,767 | 8,767 | 8,767 |
a
The null model for the likelihood ratio test is a random intercept model only. The two additional degrees of freedom in Model 1 are attributed to a random slope (black, τ11) and the covariance (τ01) between the random intercept and random slope.
†
p < .10
*
p < .05
**
p < .01
***
p < .001