Most of our public polling has the same methodology.

Beginning in February 2014, we use Address Based Sampling:

• The sample frame is residential addresses of all households in the continental United States, states, or districts we poll.
• The residential addresses randomly selected are matched to landline and cell phone telephone numbers.
• Landline and cell phone telephone interviews are conducted by live interviewers.
• Respondents are randomly selected among all adults (18 years or older) in the households.
• For election surveys, respondents are screened for likely voting twice (we use a 1-10 scale - 1 is definitely not vote and 10 is definitely vote, as an entry screen, and a definitely vote as an exit screen).
• Because we capture demographics from all households, we weight, if necessary, based on the total sample using current demographic estimates for the sample area.
  
• We do not weight by political party. Political party is determined by a party registration question in party-registration states and "do you consider yourself to be" in non party-registration states.
  
• Residential addresses not matched to telephone numbers are mailed survey packages. The survey packages allow respondents to identify their telephone numbers or to complete the surveys by mail, toll-free telephone number, or on the Internet using desktop or mobile devices.