There are obvious limitations to our research, which one should take account of when interpreting the results. Although the CCES sample is large, the non-citizen portion of the sample is modest, with the attendant uncertainty associated with sampling error. We analyze only 828 self-reported non-citizens. Self-reports of citizen status might also be a source of error, although the appendix of our paper shows that the racial, geographic, and attitudinal characteristics of non-citizens (and non-citizen voters) are consistent with their self-reported status.
Another possible limitation is the matching process conducted by Catalist to verify registration and turnout drops many non-citizen respondents who cannot be matched. Our adjusted estimate assumes the implication of a “registered” or “voted” response among those who Catalist could not match is the same as for those whom it could. If one questions this assumption, one might focus only on those non-citizens with a reported and validated vote. This is the second line of the table.
Finally, extrapolation to specific state-level or district-level election outcomes is fraught with substantial uncertainty.
Faux News!! LOL really?
You better seek help.
