Bilingual Polling

Most discussion of potential sampling error problems with polling that I've heard in recent cycles has focused on cell phone only households. Time and again, however, pollsters seem to be able to adjust for this correctly. Joshua Tucker rounds up some research on a different potential source of error—monolingual polling in a country where many people primarily speak Spanish. This could account for the way polls seem to have systematically overrated Republican performance in Nevada, Colorado, and California.