Table 3. Random intercept models predicting property crime rates among all tracts.
| Model 1 | Model 2 | Model 3 | |
|---|---|---|---|
| Tract-level Predictors | |||
| Gentrification stage (ref = vulnerable, never gentrified) | |||
| Gentrified | 0.265** | 0.184 | 0.310** |
| (0.094) | (0.104) | (0.095) | |
| Not vulnerable | -0.020 | -0.109 | -0.010 |
| (0.095) | (0.121) | (0.095) | |
| Year | -0.097*** | -0.128*** | -0.095*** |
| (0.011) | (0.014) | (0.011) | |
| Year2 | 0.007*** | 0.011*** | 0.007*** |
| (0.001) | (0.001) | (0.001) | |
| Gentrified x Year | 0.070** | ||
| (0.022) | |||
| Not vulnerable x Year | 0.087 | ||
| (0.050) | |||
| Gentrified x Year2 | -0.009*** | ||
| (0.002) | |||
| Not vulnerable x Year2 | -0.011* | ||
| (0.005) | |||
| Time-varying predictors | |||
| Gentrified in a given year | -0.082** | ||
| (0.027) | |||
| Constant | 3.960*** | 3.996*** | 3.949*** |
| (0.119) | (0.121) | (0.119) | |
* p<0.
** p<0.01
*** p<0.001. N = 684 tract-years across 76 tracts.
Notes. Standard errors in parentheses. Models control for time invariant and time -variant measures of concentrated disadvantage, residential stability, % foreign born, and % youth.