Telephone polls still provide accurate data on a wide range of social, demographic, and political variables, but some weaknesses persist. After decades of decline, the response rates for telephone polls like those conducted for Pew Research Center have stabilized in recent years to around 9% (see table below).
Telephone poll estimates for party affiliation, political ideology, and religious affiliation continue to track well with estimates from high response rate surveys conducted in-person, e.g. General Social Survey. So even at low response rates, telephone surveys that include interviews via landlines and cellphones, and that are adjusted to match the demographic profile of the U.S., can produce accurate estimates for political attitudes.
The fact is that national polls (in 2016) were actually quite accurate. Collectively, they indicated that Clinton had about a 3 percentage point lead nationally and she ultimately won the popular vote by 2 points. Furthermore, according to a new report, there are clear reasons why national polls as a group fared better than state polls. For instance, national polls were much more likely than state polls to adjust for respondent education level in their weighting, which proved critically important in the 2016 election.
Live interviewer phone polls now represent a minority share of all polling conducted in the U.S. Online polls and automated (Interactive Voice Response) polls, or combinations of the two, are collectively more common and tend to have significantly lower response rates. Response rates to online opt-in surveys are so low as to be incomputable because far more internet users are invited to join opt-in survey panels (or to take one-off online surveys) than actually do so.