Previous: What do I do

Up: Frequently Asked Questions

Next: How do I test

Why does eiread report some precinct-level estimates as missing values?

When $X_i=0$ (in the running example, this indicates no blacks in the precinct), $\beta_i^b$ (the fraction of blacks who vote) is undefined, and its estimate is set at missing. Similarly, if $X_i=1$, $\beta_i^w$ is set to missing. In ei2, if $T_i=0$ from the previous stage then $V_i$ should be undefined and both $\beta_i^b$ and $\beta_i^w$ should be missing (although it is best to remove observations for which $T_i=0$ from the ei2 run.)