Graphical summary s, with polls weighted by proximity in time and a
logarithmic function of sample size. 95% confidence ribbons represent uncertainty about the trendlines, not the likelihood that actual election results would fall within the intervals.
Notes : In cases when linked poll details distinguish between the margin of error associated with the total sample of respondents (including undecided and non-voters) and that of the subsample of decided/leaning voters, the latter is included in the table. Also not included is the margin of error created by rounding to the nearest whole number or any margin of error from methodological sources. Most online polls—because of their opt-in method of recruiting panellists which results in a non-random sample—cannot have a margin of error. In such cases, shown is what the margin of error would be for a survey using a random probability-based sample of equivalent size. : Refers to the total sample size, including undecided and non-voters. : "Telephone" refers to traditional telephone polls conducted by live interviewers; "IVR" refers to automated
Interactive Voice Response polls conducted by telephone; "online" refers to polls conducted exclusively over the internet; "telephone/online" refers to polls which combine results from both telephone and online surveys, or for which respondents are initially recruited by telephone and then asked to complete an online survey. "Rolling" polls contain overlapping data from one poll to the next, with the fraction in parentheses indicating the proportion of independent data compared to the previous published poll in the series. : Identifies polling firms that explicitly apply a "likely voter" turnout adjustment to their top-line results. ==Pre-campaign period==