NORTH CAROLINA POLLS: Cunningham +2.8 Cunningham 45.2, Tillis 41.6 REF: Cunningham +2.4 - Cunningham 44.7, 41.9 RETURN TO SENATE RACES 2016 Exit Poll - 35D/33I/31R without leaners 2018 Exit Poll - Unavailable Pollsters' Averages Chart Political Ref's Averages Chart Poll Date Sample + Party Distribution Result Pollster Avg Ref's Avg Emerson College 10/29-10/31 855LV, 36D/34I/30R Cunningham 50, Tillis 47 x   Political Ref's Take on the above poll Party distribution is reasonable. No adjustment necessary. Included   x Rasmussen Reports 10/28-10/29 800LV, Behind pay wall Cunningham 47, Tillis 44 x   Political Ref's Take on the above poll Party distribution is reasonable. No adjustment necessary. Included   x RMG Research 10/24-10/26 800LV, Not provided Cunningham 49, Tillis 42 x   Political Ref's Take on the above poll Strong GOP turnout model Cunningham 48, Tillis 43   x Harper Polling 10/22-10/25 504LV Cunningham 46, Tillis 43 x   Political Ref's Take on the above poll No adjustment necessary. Included   x Trafalgar Group 10/20-10/22 1098LV, Not provided Cunningham 47, Tillis 49 x   Political Ref's Take on the above poll No adjustment necessary. Included   x Emerson College 9/16-9/18 717LV, 36D/34I/30R Cunningham 49, Tillis 43 >1month   Political Ref's Take on the above poll Party distribution is reasonable. No adjustment necessary. Included   >1month New York Times/Sienna 9/11-9/16 653LV, D30/34I/32R Cunningham 42, Tillis 37 >1month   Political Ref's Take on the above poll Party distribution is reasonable. Cunningham won D's 89-4, I's 38-30 and Tillis won R's 80-5. No adjustment necessary Included   >1month Suffolk Univ 9/11-9/14 500LV, 38D/25I/35R Cunningham 42, Tillis 38 >1month   Political Ref's Take on the above poll Party distribution is reasonable. No adjustment necessary. Included   >1month Kaiser Family Foundation 8/29-9/13 1172RV, 31D/28I/29R Cunningham 41, Tillis 37 >1month   Political Ref's Take on the above poll The party distribution looks good but this a registered voter poll after Labor Day, so a registered voter screen will be applied. Full explanation below. Cunningham 38.8, Tillis 39.2   >1month CNN/SSRS 9/9-9/13 787LV, 30D/42I/28R Cunningham 47, Tillis 46 >1month   Political Ref's Take on the above poll The party distribution is reasonable. No adjustment necessary. Included   >1month Survey USA 9/10-9/13 596LV, 36D/26I/37R Cunningham 47, Tillis 40 >1month   Political Ref's Take on the above poll According to the poll data, Cunningham won D's 90-3, I's 45-29 and Tillis won R's 83-6. No adjustment necessary. Included   >1month Trafalgar Group 9/9-9/11 1048LV, Party distribution not provided Cunningham 46, Tillis 45 >1month   Political Ref's Take on the above poll Neither party distribution nor voter ratios of those parties is provided. Excluded     Benenson Strategy Group/GS Strategy Group 8/30-9/8 1600LV, Party distribution not provided Cunningham 42, Tillis 39 >1month   Political Ref's Take on the above poll Neither party distribution nor voter ratios of those parties is provided. Excluded     Rasmussen Reports 9/7-9/8 1000LV, Party distribution not provided Cunningham 47, Tillis 44 >1month   Political Ref's Take on the above poll Neither party distribution or voter ratios within those parties are publicly provided. Excluded     Change Research 9/4-9/6 442LV, Party distribution not provided Cunningham 51, Tillis 44 >1month   Political Ref's Take on the above poll Neither party distribution nor voter ratios of those parties is provided. Excluded     Monmouth Univ. 8/29-9/1 401LV, 36D/34I/30R Cunningham 46, Tillis 46 >1month   Political Ref's Take on the above poll The D+6 slightly oversamples Dems but the likely voter model corrects for that. Included   >1month Fox News 8/29-9/1 722LV, 46D/9I/45R Cunningham 48, Tillis 42 >1month   Political Ref's Take on the above poll The voter turnout model is a leaners model. I have no basis to correct it but it seems reasonable. No adjustment necessary. Included   >1month East Carolina University 8/29-8/30 1101LV, 37D/31I/33R Cunningham 44, Tillis 44 x   Political Ref's Take on the above poll The party distribution is reasonable. No adjustment necessary Included   x Civitas 8/6-8/10 600LV, 37D/31I/30R Cunningham 41, Tillis 38 >1month   Political Ref's Take on the above poll The D+7 sample oversamples Democrats. Adjusting to the more likely D+4, 35D/33I/31R. Full explanation below. Cunningham 45.9, Tillis 46.8   >1month All polls above are included in the average. Full explanations for all adjustments are included below this line. Emerson College 9/16-9/18 717LV, 36D/34I/30R Cunningham 49, Tillis 43 x   Political Ref's Take on the above poll Party distribution is reasonable. No adjustment necessary. Included   x New York Times/Sienna 9/11-9/16 653LV, D30/34I/32R Cunningham 42, Tillis 37 x   Political Ref's Take on the above poll Party distribution is reasonable. Cunningham won D's 89-4, I's 38-30 and Tillis won R's 80-5. No adjustment necessary Included   x Suffolk Univ 9/11-9/14 500LV, 38D/25I/35R Cunningham 42, Tillis 38 x   Political Ref's Take on the above poll Party distribution is reasonable. No adjustment necessary. Included   x Kaiser Family Foundation 8/29-9/13 1172RV, 31D/28I/29R Cunningham 41, Tillis 37 x   Political Ref's Take on the above poll The party distribution looks good but this a registered voter poll after Labor Day, so a registered voter screen will be applied. Adjusts to 24/24.3. Applied that ratio to the 78% of respondents who chose one of the candidates, Cunningham receives 38.8 to Tillis 39.2. Cunningham 38.8, Tillis 39.2   x CNN/SSRS 9/9-9/13 787LV, 30D/42I/28R Cunningham 47, Tillis 46 x   Political Ref's Take on the above poll The party distribution is reasonable. No adjustment necessary. Included   x Survey USA 9/10-9/13 596LV, 36D/26I/37R Cunningham 47, Tillis 40 x   Political Ref's Take on the above poll According to the poll data, Cunningham won D's 90-3, I's 45-29 and Tillis won R's 83-6. No adjustment necessary. Included   x Trafalgar Group 9/9-9/11 1048LV, Party distribution not provided Cunningham 46, Tillis 45 x   Political Ref's Take on the above poll Neither party distribution nor voter ratios of those parties is provided. Excluded     Benenson Strategy Group/GS Strategy Group 8/30-9/8 1600LV, Party distribution not provided Cunningham 42, Tillis 39 x   Political Ref's Take on the above poll Neither party distribution nor voter ratios of those parties is provided. Excluded     Rasmussen Reports 9/7-9/8 1000LV, Party distribution not provided Cunningham 47, Tillis 44 x   Political Ref's Take on the above poll Neither party distribution or voter ratios within those parties are publicly provided. Excluded     Change Research 9/4-9/6 442LV, Party distribution not provided Cunningham 51, Tillis 44 x   Political Ref's Take on the above poll Neither party distribution nor voter ratios of those parties is provided. Excluded     Monmouth Univ. 8/29-9/1 401LV, 36D/34I/30R Cunningham 46, Tillis 46 x   Political Ref's Take on the above poll The D+6 slightly oversamples Dems but the likely voter model corrects for that. Included   x Fox News 8/29-9/1 722LV, 46D/9I/45R Cunningham 48, Tillis 42 x   Political Ref's Take on the above poll The voter turnout model is a leaners model. I have no basis to correct it but it seems reasonable. No adjustment necessary. Included   x East Carolina University 8/29-8/30 1101LV, 37D/31I/33R Cunningham 44, Tillis 44 x   Political Ref's Take on the above poll The party distribution is reasonable. No adjustment necessary Included   x YouGov 7/28-7/31 1121LV, 37D/30I/33R Cunningham 48, Tillis 39 >1month   Political Ref's Take on the above poll The turnout assumption is reasonable. No adjustment necessary. Included   >1month Marist College 7/14-7/22 882RV, 35D/36D/29R Cunningham 50, Tillis 41 >1month   Political Ref's Take on the above poll The D+6 sample oversamples Democrats. Adjusting to the more likely 35D/33I/31R. According to the poll data, Cunningham won D's 91-6, I's 46-40 and Tillis won R's 85-10. Applying these ratios to the more likely voter turnout results in Cunningham with 50.1 to Tillis 41.6. Cunningham 50.1, Tillis 41.6   >1month PPP 7/30-7/31 934RV, 37D/32I/31R Cunningham 48, Tillis 44 x   Political Ref's Take on the above poll The D+6 sample slightly oversampled Dems. Will adjust to D+4, or 35D/33I/31R. Voter ratios within parties are not provided but can be inferred. Cunningham was up 4 in the poll with a 6 point turnout advantage suggesting Tillis did slightly better with Independents and Dems than Cunningham did with Independents and Republicans. Assuming Tillis won Independents 16-14 with 2% voting third party, or 50-44%. Assuming each candidate won his party 90-5, Cunningham receives 47.6 to Tillis 46.2. Cunningham 47.6, Tillis 46.2   x PPP 7/22-7/23 939RV, 37D/32I/31R Cunningham 48, Tillis 40 Repeat   Political Ref's Take on the above poll The D+6 sample oversamples Democrats. Will adjust to the more likely D+4, 35D/33I/31R from 2016. Voter ratios within parties are not provided but can be inferred. Cunningham had a +6 turnout advantage but leads by 8, so he must have done slightly better with Independents than Tillis. Assuming Cunningham won Independents 17-15, or 53% to 47% and also that each candidate won his party 90-5. Applying these voter ratios to the more likely turnout, Cunningham receives 49.6 to Tillis 45.1. Cunningham 49.6, Tillis 45.1   Repeat Spry Strategies 7/12-7/16 700LV, 38D/26I/35R 40.4 Cunningham, 39.8 Tillis >1month   Political Ref's Take on the above poll The D+3 turnout assumption nearly matches the 2016 D+4. No adjustment necessary. Included   >1month Cardinal Point Analytics 7/13-7/15 547LV, Party Distribution Not Provided Cunningham 47.3, Tillis 44.2 >1month   Political Ref's Take on the above poll Party distribution was not provided by voter ratios within those parties were. Cunningham won D's 75-18, Tillis won I's 49-43 and Rs 80-11. Reallocating those voter ratios to the 2016 turnout 35D/33I/31R, Cunningham receives 43.9 to Tillis 47.3. Cunningham 43.9, Tillis 47.3   >1month Change Research 7/10-7/12 655LV, Party Distribution Not available Cunningham 49, Tillis 42 Repeat   Political Ref's Take on the above poll Neither party distribution or voter ratios within those parties are provided. Adjustment not possible. Excluded from our average. Adjustment not possible. Excluded from our average.     PPP 7/7-7/8 818RV, 38D/30I/32R Cunningham 47, Tillis 39 Repeat   Political Ref's Take on the above poll The D+6 turnout is a bit high on the Democrat side. Adjusting the sample to the 2016 voter turnout of 35D/31I/31R. The Cunningham/Tillis vote ratios were not provided by the pollster. Cunningham leads by 8 in the poll while the sample had 6% more Democrats than Republicans. So Cunningham likely won Independents by 2, which was 30% of the sample, suggesting Cunningham won Independents 16-14 or 53% to 47%. Assuming Cunningham won D's 90-5, I's 53-47 and Tillis won R's 90-5, Cunningham receives 48.6 to Tillis 44.2. Cunningham 48.6, Tillis 44.2   Repeat Change Research 6/26-6/28 468LV, The overall sample of all battleground states was 39D/27I/34R. Cunningham 51, Tillis 41 Repeat   Political Ref's Take on the above poll The +5 D sample in battleground states is an oversample of Democrats. Adjustment is impossible because we don't know the party distribution of each state. Excluded from our averages.       East Carolina Univ 6/22-6/25 1149RV, 35D/26I/35R Cunningham 41, Tillis 41 >1month   Political Ref's Take on the above poll North Carolina Independents tend to break Republican. For this reason, it's hard to pin down a party distribution model. As long as the parties are close or the Dem's a little ahead in turnout assumptions, no adjustment is necessary. No adjustment necessary   >1month Fox News 6/20-6/23 1012RV, 44D/13I/43R Cunningham 39, Tillis 37 >1month   Political Ref's Take on the above poll This party distribution includes leaners making it hard to compare to 2016. NC usually breaks down fairly evenly when leaners are included, however, so no adjustment necessary. No adjustment necessary   >1month NYT/Sienna College 6/8-6/18 653RV, 33D/33I/31R Cunningham 42, Tillis 39 >1month   Political Ref's Take on the above poll There is a strong chance that Sienna failed to pick up some of the Trump vote among Independents as noted by other poll analysts. The results among Independents are far out of step with other pollsters like YouGov showing a dead heat in the group. Also, the NYT/Sienna College poll includes many respondents it should not. It includes 9% of Independents, 6% of Democrats and 4% of Republicans who self-label as somewhat likely or not at all likely to vote. These voters should not be included in a poll and pollsters will exclude these respondents when filtering for likely voters. Pollsters know that inclusion of these respondents changes the numbers substantially, so why include them? Not only will these voters not vote, but many voters who self-report as likely to vote, and even some who say they are certain to vote, will also not vote. But this category of somewhat likely to vote or will not vote at all is low hanging fruit for a pollster seeking likely voters (accuracy). I will consider this poll's results once it imposes a likely voter screen. The results are too out of the mainstream to ignore the refusal to impose a likely voter screen. Excluded from this average. Excluded from average.     Gravis Marketing 6/17 631RV, 37D/30I/33R Cunningham 45, Tillis 46 >1month   Political Ref's Take on the above poll The D+4 sample matches closes the 2016 exit poll. No adjustment necessary.   >1month PPP 6/2-6/3 949RV, 39D/28I/33R Cunningham 43, Tillis 41 Repeat   Political Ref's Take on the above poll The D+6 turnout is a bit high on the Democrat side. Adjusting the sample to the 2016 voter turnout of 35D/31I/31R. The Cunningham/Tillis vote ratios were not provided by the pollster. Cunningham leads by 2 in the poll while the sample had 6% more Democrats than Republicans. So tillis made up 4 points among Independents which was 28% of the sample, suggesting Tillis won Independents 16-12 or 57% to 43%. Assuming Cunningham won D's 90-5, Tillis won I's 57-43 and R's 90-5, Cunningham receives 46.4 to Tillis 50.4. Cunningham 46.4, Tillis 50.4   Repeat Harper Polling/Civitas Institute 5/26-5/28 500LV, 37D/29I/31R Cunningham 36, Tillis 38 Repeat   Political Ref's Take on the above poll The D+6 sample slightly oversamples Democrats. Adjusting to the more likely 35D/33I/31R voter turnout. Also, the Cunningham/Tillis vote ratio within each party is not provided, but can be inferred. Tillis trails by 6 in the poll turnout model of D+6 yet is leading by 2 overall. This suggests he made up 8 points with Independents or that he is doing very well with Dems. I will assume Cunningham won D's 90-5 and Tillis won R's 90-5, which leaves only the Independent column for Tillis to make up that 7 or 8 points. So I assume Tillis won I's 18-11, or 62% to 38%. While this is a fiction that probably overstates Cunningham's support among D's and overstates Tillis's support among I's, it nevertheless accomplishes the task of giving a good estimate of what would actually happen. This is necessary because there is no reliable way to assume a voter ratio among Democrats that shows Tillis receiving a large share. There is, however, a long history of Independents in North Carolina heavily favoring Republicans. Final adjusted result, Cunningham 45.6 to Tillis 50.1. Cunningham 45.6, Tillis 50.1   Repeat