WEBVTT

1
00:00:01.679 --> 00:00:02.960
<v Speaker 1>Welcome back to Beyond the Polls.

2
00:00:03.200 --> 00:00:05.839
<v Speaker 2>This week, I explore the world of modern polling and

3
00:00:05.839 --> 00:00:09.800
<v Speaker 2>it's international dimensions with British pollster James Kennegasaurium.

4
00:00:10.279 --> 00:00:11.160
<v Speaker 3>Let's dive in.

5
00:00:13.919 --> 00:00:17.960
<v Speaker 2>Well, what a night that was. It was in one way,

6
00:00:17.960 --> 00:00:20.480
<v Speaker 2>it is the passing of an era in Texas. And

7
00:00:20.559 --> 00:00:23.559
<v Speaker 2>I don't just mean the end of John Cornan's long

8
00:00:23.879 --> 00:00:28.320
<v Speaker 2>political career and tenure in the Senate. I mean the

9
00:00:28.440 --> 00:00:32.840
<v Speaker 2>passing of an era in the Republican Party. We've been

10
00:00:32.920 --> 00:00:35.880
<v Speaker 2>going through it for a while, We've been seeing it

11
00:00:35.960 --> 00:00:41.359
<v Speaker 2>ever since Donald Trump's storming through the Republican primaries in

12
00:00:41.439 --> 00:00:46.640
<v Speaker 2>twenty sixteen. But the utter stomping of John Cornan in

13
00:00:46.759 --> 00:00:50.759
<v Speaker 2>Texas last night, I think in retrospect we'll be looked

14
00:00:50.799 --> 00:00:54.799
<v Speaker 2>at as perhaps the moment when what's been clear for

15
00:00:54.840 --> 00:00:58.560
<v Speaker 2>a long time has become crystal clear to all who

16
00:00:58.600 --> 00:01:02.320
<v Speaker 2>will be willing to see it, and that is that

17
00:01:02.719 --> 00:01:09.480
<v Speaker 2>even a competent, non scandal free, generally conservative but not

18
00:01:10.840 --> 00:01:14.680
<v Speaker 2>maga conservative, meaning that he's got a disposition or she's

19
00:01:14.719 --> 00:01:17.079
<v Speaker 2>got a disposition that is more in tune with the

20
00:01:17.079 --> 00:01:21.040
<v Speaker 2>pre Trump Party than the current Trump Party simply cannot

21
00:01:21.040 --> 00:01:27.400
<v Speaker 2>win a contested election. If there is a strong MAGA candidate,

22
00:01:27.439 --> 00:01:30.879
<v Speaker 2>regardless of the personal weaknesses of that candidate, there's no

23
00:01:31.000 --> 00:01:34.519
<v Speaker 2>other way to explain what happened. Ken Paxton has clearly

24
00:01:34.920 --> 00:01:39.280
<v Speaker 2>been somebody who has been followed by accusations of scandal

25
00:01:39.359 --> 00:01:43.200
<v Speaker 2>financial His wife is going through a divorce with him

26
00:01:43.840 --> 00:01:47.200
<v Speaker 2>and has alleged says that she's doing the divorce on

27
00:01:47.280 --> 00:01:53.719
<v Speaker 2>biblical grounds, which everyone interprets as significant adultery. And yet

28
00:01:53.719 --> 00:01:59.680
<v Speaker 2>this guy plowed through an incumbent senator by nearly thirty points.

29
00:02:00.079 --> 00:02:05.079
<v Speaker 2>He won every county but two. John Cornyan won Kennedy County,

30
00:02:05.400 --> 00:02:08.080
<v Speaker 2>which cast only eight votes, and he won the state

31
00:02:08.159 --> 00:02:12.000
<v Speaker 2>capitol Travis County, which includes Austin, by a narrow margin.

32
00:02:12.439 --> 00:02:16.240
<v Speaker 2>Every other two hundred and fifty two counties voted for

33
00:02:16.360 --> 00:02:19.159
<v Speaker 2>Ken Paxson. And you take a look at that, and

34
00:02:19.199 --> 00:02:22.759
<v Speaker 2>you add it up with what happened in Louisiana, where

35
00:02:22.800 --> 00:02:26.159
<v Speaker 2>you had a similar breakdown, which is that incumbent Senator

36
00:02:26.520 --> 00:02:31.879
<v Speaker 2>Bill Cassidy only carried some upper income areas and the

37
00:02:31.919 --> 00:02:34.960
<v Speaker 2>county where he got his start, and you take a

38
00:02:34.960 --> 00:02:37.360
<v Speaker 2>look at it with other races over the last couple

39
00:02:37.400 --> 00:02:41.240
<v Speaker 2>of years, or anyone who has the whiff of establishment

40
00:02:41.319 --> 00:02:43.599
<v Speaker 2>about them. And I mean that not just in terms

41
00:02:43.639 --> 00:02:45.879
<v Speaker 2>of what they say they'll do, but the way in

42
00:02:45.879 --> 00:02:49.280
<v Speaker 2>which they carry themselves. The idea that they may say

43
00:02:49.280 --> 00:02:52.319
<v Speaker 2>they're a fighter, but they don't have the demeanor of

44
00:02:53.759 --> 00:02:58.199
<v Speaker 2>an MMF fighter. That person loses, and it's not usually

45
00:02:58.199 --> 00:03:02.080
<v Speaker 2>not close, it's just not close. I think we will

46
00:03:02.120 --> 00:03:06.520
<v Speaker 2>look back on last night as the end of the

47
00:03:06.639 --> 00:03:12.039
<v Speaker 2>consolidation of Trumpism as the defining ethos of the modern

48
00:03:12.080 --> 00:03:16.280
<v Speaker 2>Republican primary electorate. What that means going forward is in

49
00:03:16.360 --> 00:03:19.360
<v Speaker 2>part dependent on what type of trump Ism emerges, and

50
00:03:19.400 --> 00:03:22.039
<v Speaker 2>it's not part of this rant to go into that

51
00:03:22.120 --> 00:03:24.319
<v Speaker 2>in any detail, because I'm not sure what I think

52
00:03:24.360 --> 00:03:28.199
<v Speaker 2>about it yet. But I think it's pretty clear that

53
00:03:28.280 --> 00:03:31.800
<v Speaker 2>if you want to have a future in today's Republican Party,

54
00:03:32.800 --> 00:03:37.800
<v Speaker 2>you have to be some variation of Mega and that

55
00:03:37.840 --> 00:03:40.639
<v Speaker 2>can be a religiously tinged variation, that can be an

56
00:03:40.719 --> 00:03:47.120
<v Speaker 2>unhinged variation, that can be a nationalist variation. There's lots

57
00:03:47.159 --> 00:03:50.560
<v Speaker 2>of variations, and certainly we see examples of all different

58
00:03:50.560 --> 00:03:53.560
<v Speaker 2>types of people trying to latch onto Trump and the

59
00:03:53.599 --> 00:03:57.639
<v Speaker 2>philosophical phenomena that he has unleashed, and you can see

60
00:03:57.680 --> 00:04:01.240
<v Speaker 2>people taking different aspects, kind of like that old saying,

61
00:04:01.280 --> 00:04:03.840
<v Speaker 2>you know the blind man. There's a bunch of blind

62
00:04:03.919 --> 00:04:06.199
<v Speaker 2>men go around an elephant, and somebody feels a tale

63
00:04:06.199 --> 00:04:09.319
<v Speaker 2>and says, oh, it's a snake, and somebody feels the

64
00:04:09.319 --> 00:04:11.759
<v Speaker 2>foot and says, oh, it's this, when in fact it's

65
00:04:11.800 --> 00:04:14.319
<v Speaker 2>all parts of a hole. Most people are still only

66
00:04:14.360 --> 00:04:18.160
<v Speaker 2>doing parts of this. I think the Vice President and

67
00:04:18.199 --> 00:04:20.959
<v Speaker 2>the Secretary of State are two people who are doing

68
00:04:21.079 --> 00:04:23.399
<v Speaker 2>holes of it, which is one reason why they are

69
00:04:23.439 --> 00:04:27.040
<v Speaker 2>there and one reason why they are viewed as potential

70
00:04:27.120 --> 00:04:32.160
<v Speaker 2>national candidates. But how the old Guard reacts to this

71
00:04:32.439 --> 00:04:35.040
<v Speaker 2>will be very telling. The old Guard here in Washington,

72
00:04:35.120 --> 00:04:38.800
<v Speaker 2>d C. Continues to be in denial. They continue to

73
00:04:39.720 --> 00:04:41.600
<v Speaker 2>as I like to say, I think that Trump is

74
00:04:42.079 --> 00:04:46.199
<v Speaker 2>the dream season of Dallas. Where after Bobby leaves the show,

75
00:04:46.240 --> 00:04:49.319
<v Speaker 2>and I'm dating myself with this, we go through a

76
00:04:49.399 --> 00:04:51.519
<v Speaker 2>year where the Ewing family doesn't have Bobby. In the

77
00:04:51.560 --> 00:04:54.000
<v Speaker 2>next morning, Bobby's back in the shower and they explain

78
00:04:54.079 --> 00:04:56.720
<v Speaker 2>away and that's because the actor couldn't find the jobs

79
00:04:56.720 --> 00:04:58.160
<v Speaker 2>that he wanted, so he came back to the hit

80
00:04:58.199 --> 00:05:03.000
<v Speaker 2>show and it was all explained as his wife's dream.

81
00:05:03.639 --> 00:05:07.279
<v Speaker 2>No the last decade has not been the dream season

82
00:05:07.319 --> 00:05:09.639
<v Speaker 2>of the Republican Party. The old guard is not going

83
00:05:09.680 --> 00:05:13.079
<v Speaker 2>to come back. This is a new party, and I

84
00:05:13.120 --> 00:05:19.800
<v Speaker 2>think March, not March, May twenty six, two thousand and

85
00:05:19.879 --> 00:05:23.279
<v Speaker 2>twenty six, is going to be the day when the

86
00:05:23.319 --> 00:05:30.399
<v Speaker 2>old Republican Party is finally decisively buried as a dominant

87
00:05:30.399 --> 00:05:33.160
<v Speaker 2>force within it and becomes has to deal with the

88
00:05:33.160 --> 00:05:37.319
<v Speaker 2>fact that is now a significant minority force within that

89
00:05:37.399 --> 00:05:40.120
<v Speaker 2>party coalition. The other thing I wanted to talk about

90
00:05:40.240 --> 00:05:41.959
<v Speaker 2>is not looking back at the past, but looking at

91
00:05:42.000 --> 00:05:44.079
<v Speaker 2>the future, and that is next week we have six

92
00:05:44.199 --> 00:05:46.720
<v Speaker 2>count of six primaries, and I want to quickly go

93
00:05:46.800 --> 00:05:49.680
<v Speaker 2>over the states and which races I'll be looking at

94
00:05:49.720 --> 00:05:55.399
<v Speaker 2>in them. We have California, Iowa, New Jersey, Montana, South Dakota,

95
00:05:55.439 --> 00:05:58.120
<v Speaker 2>and New Mexico. I don't really think there's a I'm

96
00:05:58.160 --> 00:06:00.319
<v Speaker 2>not looking at any primaries in New Mexico, but I'm

97
00:06:00.319 --> 00:06:02.959
<v Speaker 2>looking at ones in each of the other five states.

98
00:06:03.000 --> 00:06:05.959
<v Speaker 2>The first few I want to talk about her in California,

99
00:06:06.120 --> 00:06:10.360
<v Speaker 2>and California's unique system means you're looking for something different here,

100
00:06:11.120 --> 00:06:14.000
<v Speaker 2>which is to say, it's an all party primary and

101
00:06:14.040 --> 00:06:17.560
<v Speaker 2>the top two regardless of party go on. So part

102
00:06:17.600 --> 00:06:22.279
<v Speaker 2>of what you're seeing is a gamesmanship between either two

103
00:06:22.360 --> 00:06:25.560
<v Speaker 2>people in one party to see who will face get

104
00:06:25.600 --> 00:06:28.199
<v Speaker 2>through the top two to face the person in the

105
00:06:28.279 --> 00:06:33.240
<v Speaker 2>other party, or attempts to game the system so that

106
00:06:33.279 --> 00:06:36.360
<v Speaker 2>you get the final to match up that you want

107
00:06:36.680 --> 00:06:40.439
<v Speaker 2>with respect to finding somebody who is weaker or somebody

108
00:06:40.439 --> 00:06:41.439
<v Speaker 2>who you fear more.

109
00:06:42.120 --> 00:06:43.040
<v Speaker 1>In that race.

110
00:06:44.399 --> 00:06:47.439
<v Speaker 2>California governor, it's pretty clear that the Republican Steve Hilton

111
00:06:47.480 --> 00:06:50.639
<v Speaker 2>will go through. It looks like the other Republican, Chad Bianco,

112
00:06:50.720 --> 00:06:55.079
<v Speaker 2>who had a shot to make one two, he's consistently

113
00:06:55.120 --> 00:06:57.600
<v Speaker 2>pulling in fourth place. Now, I wouldn't put it past

114
00:06:58.240 --> 00:07:00.680
<v Speaker 2>the possibility that there'll be a surprise eyes and the

115
00:07:00.800 --> 00:07:04.879
<v Speaker 2>Democrats worst fear of two Republicans getting through in one

116
00:07:04.920 --> 00:07:07.800
<v Speaker 2>of the most democratic states in the country will occur.

117
00:07:07.920 --> 00:07:10.519
<v Speaker 2>But it's increasingly looking like the battle is going to

118
00:07:10.560 --> 00:07:15.680
<v Speaker 2>between progressive billionaire Tom Steyer and a former h former

119
00:07:15.959 --> 00:07:20.040
<v Speaker 2>cabinet secretary for Biden Chavy or Bishera. Stire, of course,

120
00:07:20.160 --> 00:07:22.639
<v Speaker 2>is much likelier to be drawing from the upper income,

121
00:07:22.720 --> 00:07:28.480
<v Speaker 2>progressive white community. Bishera likely to draw from the lower

122
00:07:28.560 --> 00:07:32.240
<v Speaker 2>to middle income Hispanic community, which means we should expect

123
00:07:32.279 --> 00:07:34.560
<v Speaker 2>Spire to do much better in the Bay Area and

124
00:07:34.600 --> 00:07:37.079
<v Speaker 2>Bishera to do much better in his native Los Angeles.

125
00:07:37.319 --> 00:07:40.360
<v Speaker 2>Central Valley has elements of both, depending on which county here.

126
00:07:40.399 --> 00:07:42.519
<v Speaker 2>And that's the race that I'm looking at for governor

127
00:07:42.560 --> 00:07:45.839
<v Speaker 2>who finishes second to Steve Hilton, because the odds are

128
00:07:45.920 --> 00:07:48.680
<v Speaker 2>that that person becomes the next governor given how democratic

129
00:07:48.720 --> 00:07:51.160
<v Speaker 2>the state is, then I'm taking a look at a

130
00:07:51.199 --> 00:07:53.480
<v Speaker 2>few other things. California one is going to be a

131
00:07:53.519 --> 00:07:56.759
<v Speaker 2>Democratic seat, but there's a strong Republican running. There's a

132
00:07:56.759 --> 00:07:59.560
<v Speaker 2>lot of Republicans in the race, and it's highly likely

133
00:07:59.600 --> 00:08:01.439
<v Speaker 2>that he will one of the top two who will

134
00:08:01.480 --> 00:08:03.519
<v Speaker 2>be the Democrat who faces them. And here we have

135
00:08:03.560 --> 00:08:07.800
<v Speaker 2>an ideological contest between I think it's Aubrey Denny and

136
00:08:07.959 --> 00:08:13.120
<v Speaker 2>State Senator McGuire. Denny is a DSA Our Revolution endorsed candidate.

137
00:08:13.160 --> 00:08:18.399
<v Speaker 2>McGuire is the establishment straight through. This includes some establishment

138
00:08:18.399 --> 00:08:21.480
<v Speaker 2>friendly areas. Of course, as a state senator, McGuire represents

139
00:08:21.480 --> 00:08:23.800
<v Speaker 2>a large part of this district. State senate districts in

140
00:08:23.800 --> 00:08:27.839
<v Speaker 2>California are larger. They're in congressional district. But we know

141
00:08:27.920 --> 00:08:30.680
<v Speaker 2>that ideological candidates can break through, and there are strong

142
00:08:30.720 --> 00:08:34.279
<v Speaker 2>ideological components of the district, you know, for example around

143
00:08:34.279 --> 00:08:38.840
<v Speaker 2>Mendocino or in Humboldt County County, around Humboldt State. It'll

144
00:08:38.840 --> 00:08:42.200
<v Speaker 2>be very interesting to see from the Democratic partisan standpoint,

145
00:08:42.399 --> 00:08:47.639
<v Speaker 2>who wins the ideologue or the lawmaker. In California seven,

146
00:08:47.759 --> 00:08:51.039
<v Speaker 2>we have long time incumbent Doris Matsui who's being challenged

147
00:08:51.039 --> 00:08:55.039
<v Speaker 2>from the left by a woman named Vang and a Republican.

148
00:08:55.080 --> 00:08:58.720
<v Speaker 2>Now this is a seat where it's not entirely pop

149
00:08:59.000 --> 00:09:01.720
<v Speaker 2>sure that a Republican is going to get through. So

150
00:09:01.720 --> 00:09:04.679
<v Speaker 2>what Matt Suey wants to do and is try and

151
00:09:04.720 --> 00:09:07.919
<v Speaker 2>boost the Republican a little bit so that Vang finishes

152
00:09:08.000 --> 00:09:10.759
<v Speaker 2>third and then it's her versus a Republican. Of course,

153
00:09:10.799 --> 00:09:14.519
<v Speaker 2>what Vang wants to do is make sure she finishes second,

154
00:09:14.559 --> 00:09:17.200
<v Speaker 2>so then the Republicans don't have a choice and perhaps

155
00:09:17.200 --> 00:09:19.399
<v Speaker 2>she can just take out the incumbent and a Democrat

156
00:09:19.440 --> 00:09:22.159
<v Speaker 2>On Democrat five, I'll be very interested to see the

157
00:09:22.200 --> 00:09:27.759
<v Speaker 2>strength of a generation change left wing challenge to a

158
00:09:27.879 --> 00:09:32.200
<v Speaker 2>long term establishment Democrat. California eleven is a Speaker Pelosi's

159
00:09:32.279 --> 00:09:34.799
<v Speaker 2>old seat, and this is again a seat where no

160
00:09:35.000 --> 00:09:38.320
<v Speaker 2>Republican is going to make the runoff here. I think

161
00:09:38.360 --> 00:09:40.480
<v Speaker 2>Donald Trump got under ten percent of the vote or

162
00:09:40.600 --> 00:09:42.720
<v Speaker 2>just around ten percent of the vote here. So this

163
00:09:42.840 --> 00:09:45.279
<v Speaker 2>is which two Democrats are going to face one another.

164
00:09:45.360 --> 00:09:47.840
<v Speaker 2>State Senator Scott Wiener is a person who leads in

165
00:09:47.840 --> 00:09:49.919
<v Speaker 2>the polls, and then you have the question is is

166
00:09:49.960 --> 00:09:54.399
<v Speaker 2>it going to be a more establishment friendly person person

167
00:09:54.440 --> 00:09:57.480
<v Speaker 2>polosees endorsed, or is it going to be AOC's former

168
00:09:57.559 --> 00:10:01.039
<v Speaker 2>chief of staff chach Rabarti. That's the race I'm going

169
00:10:01.120 --> 00:10:03.879
<v Speaker 2>to see, because again, it'll be very interested to see

170
00:10:04.279 --> 00:10:08.240
<v Speaker 2>when you have a clear person who is going to

171
00:10:08.279 --> 00:10:14.639
<v Speaker 2>make it through. Is the progressive ideologue strength in San

172
00:10:14.639 --> 00:10:20.320
<v Speaker 2>Francisco sufficiently strong enough and sufficiently coalesced around Chakra Party

173
00:10:20.360 --> 00:10:22.639
<v Speaker 2>that he can get twenty five to thirty percent, because

174
00:10:22.879 --> 00:10:25.639
<v Speaker 2>that's probably what he needs to guarantee or have a

175
00:10:25.720 --> 00:10:28.320
<v Speaker 2>good shot at getting through. And it's the polls that

176
00:10:28.440 --> 00:10:31.320
<v Speaker 2>do exist, which are very few, suggests it's a toss

177
00:10:31.360 --> 00:10:34.840
<v Speaker 2>up there. California twenty two is two for Democrats to

178
00:10:34.840 --> 00:10:36.320
<v Speaker 2>see who's going to take on one of the most

179
00:10:36.399 --> 00:10:41.799
<v Speaker 2>vulnerable Republicans in the House. David Valdeo has won the

180
00:10:41.879 --> 00:10:45.799
<v Speaker 2>various versions of the seat despite heavy odds for many years.

181
00:10:45.840 --> 00:10:48.919
<v Speaker 2>There's a lot of popularity. And here we have another

182
00:10:49.080 --> 00:10:53.440
<v Speaker 2>establishment versus progressive. Viegis who is endorsed by AOC and

183
00:10:53.440 --> 00:10:57.120
<v Speaker 2>Bernie and our Revolution versus Baines. Be very interested in

184
00:10:57.120 --> 00:11:00.799
<v Speaker 2>this highly Hispanic district, and Baines is a state assembly person.

185
00:11:00.840 --> 00:11:03.799
<v Speaker 2>It'll be very interested in this district to see who

186
00:11:03.840 --> 00:11:07.480
<v Speaker 2>gets through against Valaday. Obviously Republicans would probably prefer the

187
00:11:07.559 --> 00:11:11.639
<v Speaker 2>more extreme person. But it's again yet another test in

188
00:11:11.679 --> 00:11:13.840
<v Speaker 2>a very progressive state as to the strength that the

189
00:11:13.840 --> 00:11:18.279
<v Speaker 2>progressive left within the Democratic Party primary. It is reversed

190
00:11:18.320 --> 00:11:21.799
<v Speaker 2>in California forty. This is a vote sinc for Republicans,

191
00:11:21.799 --> 00:11:24.360
<v Speaker 2>and we have two Republican incumbents, Ken Calvert and Young

192
00:11:24.440 --> 00:11:27.320
<v Speaker 2>Kim who are facing off to see who will have

193
00:11:27.360 --> 00:11:30.240
<v Speaker 2>the right to get the Democrat because there is enough

194
00:11:30.240 --> 00:11:32.519
<v Speaker 2>Democratic strength that a Democrat will probably get a third

195
00:11:32.519 --> 00:11:34.039
<v Speaker 2>of the voter. So there and then the question is

196
00:11:34.080 --> 00:11:37.000
<v Speaker 2>which a Republican is going to get through. They've spent

197
00:11:37.200 --> 00:11:41.519
<v Speaker 2>millions of dollars attacking each other. Listeners of this podcast

198
00:11:41.679 --> 00:11:44.000
<v Speaker 2>heard some of those ads a few weeks ago as

199
00:11:44.000 --> 00:11:46.960
<v Speaker 2>they both were trying to compete for basically the same constituency.

200
00:11:47.279 --> 00:11:50.960
<v Speaker 2>Kim represents more of this area has represented more of

201
00:11:50.960 --> 00:11:53.519
<v Speaker 2>this area in the past. That's often a good sign

202
00:11:53.559 --> 00:11:56.360
<v Speaker 2>as to who wins member on member fights. But we

203
00:11:56.440 --> 00:11:58.000
<v Speaker 2>shall see. That's the other one I'm looking at. And

204
00:11:58.039 --> 00:12:00.000
<v Speaker 2>then the last one I'm looking at is California four.

205
00:12:00.799 --> 00:12:03.360
<v Speaker 2>This is a seat that Democrats are trying to pick up.

206
00:12:03.559 --> 00:12:06.519
<v Speaker 2>Darryl Isa the Republican was drawn into the seat that

207
00:12:06.600 --> 00:12:09.120
<v Speaker 2>leans Democrat but is not a safe Democratic seat in

208
00:12:09.159 --> 00:12:14.080
<v Speaker 2>a normal year. There's a strong Republican candidate super San

209
00:12:14.080 --> 00:12:17.399
<v Speaker 2>Diego Supervisor Jim Desmond, and there is a four way

210
00:12:17.519 --> 00:12:21.440
<v Speaker 2>clown car race to take him on, and it's totally

211
00:12:21.559 --> 00:12:25.360
<v Speaker 2>unclear who is going to do it, but one of

212
00:12:25.360 --> 00:12:31.679
<v Speaker 2>the leading candidates is a person camp Nihar, who is

213
00:12:31.679 --> 00:12:34.919
<v Speaker 2>a perennial candidate, has failed in races before, but has

214
00:12:34.960 --> 00:12:39.440
<v Speaker 2>the advantage of being the boyfriend of nearby Congresswoman Sarah Jacobs,

215
00:12:39.480 --> 00:12:42.440
<v Speaker 2>and she has the advantage of being the heiress of

216
00:12:42.480 --> 00:12:46.159
<v Speaker 2>a billionaire fortune. We shall see whether or not the fortune,

217
00:12:46.200 --> 00:12:48.639
<v Speaker 2>the connections, and so forth gets him the fifteen or

218
00:12:48.639 --> 00:12:52.320
<v Speaker 2>twenty percent. He probably needs to face Desmond and hence

219
00:12:52.360 --> 00:12:56.159
<v Speaker 2>become perhaps the favorite in this Democratic leaning seat. Now

220
00:12:56.200 --> 00:12:58.159
<v Speaker 2>let's jump to Iowa. The only race I'm looking at

221
00:12:58.200 --> 00:13:01.159
<v Speaker 2>there is who's going to be the Democratic senatorial nominee.

222
00:13:01.200 --> 00:13:03.519
<v Speaker 2>And this is another one Walls versus Turek, where you

223
00:13:03.559 --> 00:13:06.919
<v Speaker 2>have a little bit of a progressive versus establishment matchup,

224
00:13:07.200 --> 00:13:10.519
<v Speaker 2>Walls being more progressive, Torek being a little bit more establishment.

225
00:13:11.120 --> 00:13:15.720
<v Speaker 2>New Jersey seven, we have a four way Democratic race

226
00:13:15.840 --> 00:13:20.159
<v Speaker 2>to space Representative Thomas Kane. This is another seat that

227
00:13:20.240 --> 00:13:23.960
<v Speaker 2>Democrats have in their targets. It is one that, given

228
00:13:24.000 --> 00:13:26.879
<v Speaker 2>the demographics and the closeness of it, is a prime

229
00:13:27.080 --> 00:13:30.159
<v Speaker 2>likelihood to flip. And that is compounded by the fact

230
00:13:30.159 --> 00:13:32.480
<v Speaker 2>that kin has been absent from Congress for the last

231
00:13:32.480 --> 00:13:36.159
<v Speaker 2>couple of months for an unspecified and undisclosed health issue.

232
00:13:36.679 --> 00:13:40.679
<v Speaker 2>I have no feel who is coming through. There's no

233
00:13:40.840 --> 00:13:44.840
<v Speaker 2>obvious ideological break between the four candidates, all of whom

234
00:13:44.919 --> 00:13:47.799
<v Speaker 2>have raised and spent over a million dollars. But there's

235
00:13:47.799 --> 00:13:49.840
<v Speaker 2>no runoff in New Jersey and this is going to

236
00:13:49.879 --> 00:13:52.919
<v Speaker 2>be an interesting race. Not so in New Jersey twelve.

237
00:13:52.960 --> 00:13:55.440
<v Speaker 2>By not so, what I mean is there is a

238
00:13:55.519 --> 00:13:58.679
<v Speaker 2>clear issue here, and that is whether or not Adam Homaway,

239
00:13:58.759 --> 00:14:04.480
<v Speaker 2>a progressive squad endorsed Bernie Sanders endorse person is going

240
00:14:04.519 --> 00:14:07.840
<v Speaker 2>to win the thirteen way Yes, thirteen way Democratic primary

241
00:14:07.879 --> 00:14:11.600
<v Speaker 2>for a safe Democratic seat made open by the retirement

242
00:14:11.600 --> 00:14:16.759
<v Speaker 2>of Bonnie Watson Coleman. Obviously, this is one particularly given

243
00:14:17.000 --> 00:14:21.279
<v Speaker 2>the candidate's views on Israel, where the Jewish community in

244
00:14:21.320 --> 00:14:23.679
<v Speaker 2>the Democratic Party would like to see anything but a

245
00:14:23.720 --> 00:14:26.440
<v Speaker 2>Hamaway victory. I'll be very interested to see whether or

246
00:14:26.440 --> 00:14:29.879
<v Speaker 2>not Hamaway can negotiate a thirteen way primary to get

247
00:14:30.120 --> 00:14:34.600
<v Speaker 2>a plurality triumph. Then we go to Montana. Both parties

248
00:14:34.639 --> 00:14:38.480
<v Speaker 2>have open multi candidate races to succeed retiring Representative Ryan

249
00:14:38.559 --> 00:14:42.120
<v Speaker 2>Zinke in Montana one. This is one of those seats

250
00:14:42.200 --> 00:14:45.039
<v Speaker 2>that should be won by a Republican, but it is

251
00:14:45.080 --> 00:14:49.240
<v Speaker 2>the more democratic of the two seats. It includes mining

252
00:14:49.279 --> 00:14:52.679
<v Speaker 2>communities and universities in the state capital places, so it

253
00:14:52.679 --> 00:14:54.759
<v Speaker 2>has a much stronger Democratic base. And this is one

254
00:14:54.799 --> 00:14:57.840
<v Speaker 2>that in a wave or quasi wave here which this

255
00:14:57.960 --> 00:15:00.840
<v Speaker 2>may be shaping up to be, Democrats would be looking at.

256
00:15:00.879 --> 00:15:03.360
<v Speaker 2>It'll be very interested to see what the matchup is there.

257
00:15:03.639 --> 00:15:05.720
<v Speaker 2>And then finally, in South Dakota, we have a four

258
00:15:05.799 --> 00:15:09.879
<v Speaker 2>way Republican race to succeed Christino. The appointment the lieutenant

259
00:15:09.879 --> 00:15:12.919
<v Speaker 2>governor who stepped up is not dominating the race, but

260
00:15:13.000 --> 00:15:16.159
<v Speaker 2>neither is state Representative Dusty Johnson, who is leaving his

261
00:15:16.279 --> 00:15:20.200
<v Speaker 2>seat in order to run for governor. The winner needs

262
00:15:20.360 --> 00:15:23.399
<v Speaker 2>thirty five percent in order to avoid a runoff. The

263
00:15:23.440 --> 00:15:26.480
<v Speaker 2>most recent public poll has none of the candidates breaking that.

264
00:15:26.879 --> 00:15:29.720
<v Speaker 2>This may very well be the first time that South

265
00:15:29.799 --> 00:15:34.320
<v Speaker 2>Dakota's runoff provision is triggered in a statewide race. So

266
00:15:34.320 --> 00:15:37.159
<v Speaker 2>those are twelve races that I'm looking at. But let's

267
00:15:37.240 --> 00:15:40.320
<v Speaker 2>return to Texas, and let's return to the historic moment

268
00:15:40.360 --> 00:15:45.279
<v Speaker 2>that befell us. When a high ranking Republican who has

269
00:15:45.320 --> 00:15:49.360
<v Speaker 2>had a successful career, winning statewide, who has been set

270
00:15:49.399 --> 00:15:53.080
<v Speaker 2>by no scandal, loses by nearly thirty points. That's not

271
00:15:53.200 --> 00:15:59.039
<v Speaker 2>an accident, and that's not unimportant. This remarks a sea

272
00:15:59.120 --> 00:16:01.720
<v Speaker 2>change in the Republican Party, and we will continue to

273
00:16:01.799 --> 00:16:05.759
<v Speaker 2>see that sea change work its way through the party system,

274
00:16:05.799 --> 00:16:09.480
<v Speaker 2>both the remainder of this year and going into twenty

275
00:16:09.519 --> 00:16:19.000
<v Speaker 2>twenty eight. Well, in a podcast called Beyond the Polls,

276
00:16:19.399 --> 00:16:21.919
<v Speaker 2>I do spend a lot of time on polling, because

277
00:16:22.039 --> 00:16:25.279
<v Speaker 2>while we want to go beyond that, it's often the

278
00:16:25.320 --> 00:16:29.720
<v Speaker 2>most accessible and sometimes the deepest information that is readily

279
00:16:29.720 --> 00:16:30.840
<v Speaker 2>accessible to everybody.

280
00:16:31.159 --> 00:16:33.120
<v Speaker 1>And tonight I have.

281
00:16:33.080 --> 00:16:36.279
<v Speaker 2>One of the more interesting posters in the entire world here,

282
00:16:36.559 --> 00:16:39.919
<v Speaker 2>and that is James Kannagasurium, the chief research officer of

283
00:16:40.320 --> 00:16:43.200
<v Speaker 2>Britain's Focal Data. But he's not just a British poster.

284
00:16:43.279 --> 00:16:46.440
<v Speaker 2>We'll get into that. James, welcome to be on the Polls.

285
00:16:46.840 --> 00:16:48.480
<v Speaker 3>Thank you, Henry, delight to be here.

286
00:16:49.960 --> 00:16:51.799
<v Speaker 2>Well, you know, one of the things that's attracted me

287
00:16:52.120 --> 00:16:55.759
<v Speaker 2>to your work is the wide depth of questions, why

288
00:16:55.799 --> 00:16:58.399
<v Speaker 2>breadth of questions that you do, the sort of deep stuff,

289
00:16:59.240 --> 00:17:00.679
<v Speaker 2>and I want to get in that, But I want

290
00:17:00.679 --> 00:17:02.840
<v Speaker 2>to start a little bit with what's it like to

291
00:17:02.879 --> 00:17:06.640
<v Speaker 2>be upholster these days in the world where random samples

292
00:17:06.680 --> 00:17:10.519
<v Speaker 2>are impossible to come by.

293
00:17:10.599 --> 00:17:12.000
<v Speaker 3>Let's start easy, shall I?

294
00:17:13.160 --> 00:17:14.960
<v Speaker 1>So oh yeah, well let's be on the pole.

295
00:17:15.039 --> 00:17:17.599
<v Speaker 2>So we're going to get into the high level calculus

296
00:17:17.640 --> 00:17:21.400
<v Speaker 2>behind those things soon. So prepare your regression analysis and

297
00:17:21.599 --> 00:17:22.400
<v Speaker 2>Greek letter.

298
00:17:23.519 --> 00:17:25.920
<v Speaker 3>If I take a kind of if I go up

299
00:17:26.000 --> 00:17:28.000
<v Speaker 3>kind of a thousand feet in the air and I

300
00:17:28.119 --> 00:17:32.440
<v Speaker 3>think about why, how has polling evolved? I think we've

301
00:17:32.440 --> 00:17:36.799
<v Speaker 3>basically moved in a direction where the underlying samples that

302
00:17:36.880 --> 00:17:39.839
<v Speaker 3>we obtain from people have become harder to obtain, which

303
00:17:39.880 --> 00:17:42.519
<v Speaker 3>I think is the premise of your question exactly, And

304
00:17:42.559 --> 00:17:45.279
<v Speaker 3>therefore what you do with the data we all have

305
00:17:45.359 --> 00:17:47.480
<v Speaker 3>to work harder in order to get a result. So

306
00:17:47.519 --> 00:17:51.599
<v Speaker 3>the modeling that is undertaken has become more intensive and

307
00:17:51.680 --> 00:17:57.319
<v Speaker 3>more technical in order to basically compensate for difficulties around samples.

308
00:17:57.319 --> 00:17:59.279
<v Speaker 3>But I'll take a couple of couple of steps back.

309
00:17:59.359 --> 00:18:05.039
<v Speaker 3>So obviously polling as a US construct, it is, it

310
00:18:05.079 --> 00:18:08.720
<v Speaker 3>is downstream of It is an American invention, very very

311
00:18:08.720 --> 00:18:11.799
<v Speaker 3>clearly in terms of its history, whether that's from straw

312
00:18:11.839 --> 00:18:15.319
<v Speaker 3>polls or the history of gallop and you know, even

313
00:18:15.319 --> 00:18:18.519
<v Speaker 3>from the failure of readers digest and how that was taken,

314
00:18:18.519 --> 00:18:21.440
<v Speaker 3>and the whole concept of what is a representative sample.

315
00:18:21.480 --> 00:18:27.160
<v Speaker 3>We've obviously shifted as an industry away from you know,

316
00:18:27.279 --> 00:18:31.160
<v Speaker 3>taking people's views from magazines and going onto landlines and

317
00:18:31.200 --> 00:18:34.880
<v Speaker 3>sells cell phones and obtaining it that way. And then

318
00:18:34.880 --> 00:18:38.480
<v Speaker 3>we had the advent obviously of internet polling in and

319
00:18:38.599 --> 00:18:42.519
<v Speaker 3>around kind of the early kind of two thousands as

320
00:18:42.519 --> 00:18:45.039
<v Speaker 3>it were, really where that kind of took off. But

321
00:18:45.119 --> 00:18:48.119
<v Speaker 3>the theme that kind of unites everything. The common thread

322
00:18:48.160 --> 00:18:52.119
<v Speaker 3>through it is that basically, as our attention and our

323
00:18:52.200 --> 00:18:55.680
<v Speaker 3>time has contracted because of the number of distractions that

324
00:18:55.759 --> 00:19:00.400
<v Speaker 3>exist now in the world, people's propensity and people's type

325
00:19:00.400 --> 00:19:03.160
<v Speaker 3>to take poles has collapsed. Pham poles have moved.

326
00:19:02.880 --> 00:19:08.319
<v Speaker 4>From a environment where maybe one in three adults would say,

327
00:19:08.799 --> 00:19:12.200
<v Speaker 4>of course, I'll take a poll, I'll you know, I'll

328
00:19:12.599 --> 00:19:16.599
<v Speaker 4>give you my views, to a world where it's far

329
00:19:16.680 --> 00:19:18.160
<v Speaker 4>below zero point five percent.

330
00:19:18.559 --> 00:19:21.920
<v Speaker 3>In terms of so one in every two hundred adults

331
00:19:22.039 --> 00:19:26.160
<v Speaker 3>if we rang them, we're sent to taking a survey.

332
00:19:26.279 --> 00:19:28.599
<v Speaker 2>And the word, of course, one of the things is

333
00:19:28.599 --> 00:19:30.279
<v Speaker 2>that when it's down to one or two hundred, it's

334
00:19:30.319 --> 00:19:33.039
<v Speaker 2>no longer a random assessment of who will take that.

335
00:19:33.119 --> 00:19:37.240
<v Speaker 3>So I think that's right, and the hope I think

336
00:19:37.319 --> 00:19:40.279
<v Speaker 3>with internet polling, which is the idea that people can

337
00:19:40.359 --> 00:19:47.759
<v Speaker 3>sent to taking surveys for polsters on online platforms, online communities,

338
00:19:48.319 --> 00:19:50.160
<v Speaker 3>was the idea that you could then in some way

339
00:19:50.440 --> 00:19:54.559
<v Speaker 3>compensate for that. But the reality is that today to

340
00:19:54.960 --> 00:19:58.839
<v Speaker 3>achieve a representative sample, and so what posters obviously mean

341
00:19:58.880 --> 00:20:01.559
<v Speaker 3>by that is it is the group of people I'm

342
00:20:01.559 --> 00:20:06.319
<v Speaker 3>looking at by state, by nation, by subregion, representative on

343
00:20:06.480 --> 00:20:11.880
<v Speaker 3>things like age and gender, race, ethnicity, education, but also

344
00:20:12.519 --> 00:20:15.599
<v Speaker 3>more complicated demographics, and then the interaction between all of

345
00:20:15.640 --> 00:20:19.400
<v Speaker 3>those that becomes difficult. And I think what you're getting

346
00:20:19.440 --> 00:20:22.759
<v Speaker 3>at is that even in the advent of international and

347
00:20:23.400 --> 00:20:26.839
<v Speaker 3>internet polling from firms like the Yugov and Morning consult

348
00:20:26.880 --> 00:20:30.039
<v Speaker 3>and all these other firms that now operate on an

349
00:20:30.079 --> 00:20:33.880
<v Speaker 3>Internet basis, is that even then it's very challenging. And why,

350
00:20:34.079 --> 00:20:38.319
<v Speaker 3>And it's not necessarily about the demographics, because the act

351
00:20:38.400 --> 00:20:41.880
<v Speaker 3>of being willing to take a poll has embedded and

352
00:20:41.960 --> 00:20:48.799
<v Speaker 3>nested within it behavioral and political and belief kind of

353
00:20:48.880 --> 00:20:53.200
<v Speaker 3>differences basically to do with that correlate very sharply to

354
00:20:53.240 --> 00:20:57.000
<v Speaker 3>things like social trusts and the idea that you trust

355
00:20:57.640 --> 00:21:01.079
<v Speaker 3>the pollster, and that correlates to things like trusting government

356
00:21:01.160 --> 00:21:05.160
<v Speaker 3>and trusting institutions. And that becomes a problem because those

357
00:21:05.279 --> 00:21:08.480
<v Speaker 3>precise things are in fact then related to if you're

358
00:21:08.519 --> 00:21:12.279
<v Speaker 3>a political polster, voting intention. So the very method by

359
00:21:12.279 --> 00:21:17.680
<v Speaker 3>which you are seeking to reach people then becomes problematic

360
00:21:17.759 --> 00:21:20.599
<v Speaker 3>because you're putting through a filter that is not demographic,

361
00:21:21.200 --> 00:21:26.160
<v Speaker 3>but is non demographic, but fundamentally changes beyond just individual

362
00:21:26.279 --> 00:21:30.160
<v Speaker 3>level markers of someone how they might vote. And I

363
00:21:30.160 --> 00:21:34.519
<v Speaker 3>think that's why as a polster, particularly in the US context,

364
00:21:34.559 --> 00:21:37.559
<v Speaker 3>we live in what's called a mixed method mode to

365
00:21:37.640 --> 00:21:41.359
<v Speaker 3>your very geeky listeners, where polsters then will say, okay,

366
00:21:41.480 --> 00:21:43.920
<v Speaker 3>I need to take multiple approaches. I might need to

367
00:21:43.960 --> 00:21:48.039
<v Speaker 3>find sample from internet panels, I might need to use

368
00:21:48.039 --> 00:21:51.480
<v Speaker 3>cell phones. I might need to use what's called texter Web,

369
00:21:51.680 --> 00:21:53.640
<v Speaker 3>so you know, you'll send a survey on your phone

370
00:21:53.759 --> 00:21:56.920
<v Speaker 3>and then it follows a link and it gets more

371
00:21:56.920 --> 00:22:01.160
<v Speaker 3>and more complicated, and the art and because it is

372
00:22:01.440 --> 00:22:05.160
<v Speaker 3>a bit of both drawing together representative samples on a

373
00:22:05.200 --> 00:22:08.640
<v Speaker 3>demographic basis, but also non demographic features, so making sure

374
00:22:08.640 --> 00:22:11.640
<v Speaker 3>that people aren't too interested in politics? Who are booking

375
00:22:11.720 --> 00:22:14.039
<v Speaker 3>your survey? Right? Who think about? Who are the types

376
00:22:14.079 --> 00:22:20.160
<v Speaker 3>of people to respond to the invitation? It's usually it's

377
00:22:20.240 --> 00:22:23.240
<v Speaker 3>usually the opinionated, it's usually the very interested in politics.

378
00:22:23.279 --> 00:22:27.000
<v Speaker 3>It's normally the higher trust. And I think, particularly in

379
00:22:27.039 --> 00:22:31.480
<v Speaker 3>a US context, this is really the architecture in which

380
00:22:31.519 --> 00:22:36.880
<v Speaker 3>sits repeated polling failures and misses the US cycle, usually

381
00:22:36.920 --> 00:22:41.519
<v Speaker 3>in one direction, which is to fail to capture enough

382
00:22:42.039 --> 00:22:47.839
<v Speaker 3>Republicans in voting intention, particularly for federal elections, where turnout

383
00:22:47.920 --> 00:22:51.039
<v Speaker 3>is higher and we have occasional voters, So it's a

384
00:22:51.119 --> 00:22:54.039
<v Speaker 3>challenge everywhere. Your question is what is it like as opolster. Well,

385
00:22:54.720 --> 00:22:59.319
<v Speaker 3>because people are honest, and I hope mostly my colleagues

386
00:22:59.359 --> 00:23:02.240
<v Speaker 3>across different firms and internationally are quite transparent about those

387
00:23:02.240 --> 00:23:06.359
<v Speaker 3>problems where people just don't want to take surveys. There

388
00:23:06.400 --> 00:23:10.160
<v Speaker 3>is then a need to adjust those surveys with quite

389
00:23:10.240 --> 00:23:13.880
<v Speaker 3>intensive data computation. I'm chief research officer. My job is

390
00:23:13.920 --> 00:23:17.279
<v Speaker 3>to interpret pulse. I'm not the person that goes on

391
00:23:17.400 --> 00:23:20.440
<v Speaker 3>and creates these complicated models. I may have built some

392
00:23:20.559 --> 00:23:22.880
<v Speaker 3>very basic stuff ten to fifteen years ago, but what

393
00:23:22.960 --> 00:23:27.319
<v Speaker 3>happens now is of a totally different order of magnitude.

394
00:23:27.519 --> 00:23:29.119
<v Speaker 3>Henry and we can kind of get into it, But

395
00:23:29.920 --> 00:23:37.200
<v Speaker 3>now polsters will They might impute, for example, someone's propensity

396
00:23:37.240 --> 00:23:41.000
<v Speaker 3>to turn out based off a regression model, which means

397
00:23:41.039 --> 00:23:44.400
<v Speaker 3>that we have to work out whether someone really is

398
00:23:44.480 --> 00:23:47.400
<v Speaker 3>likely to turn out based on validated information and then

399
00:23:47.480 --> 00:23:50.640
<v Speaker 3>create models. Or it might be that it is so

400
00:23:50.880 --> 00:23:53.960
<v Speaker 3>hard to get the interaction of all these different variables,

401
00:23:54.599 --> 00:23:56.680
<v Speaker 3>so it's not enough to have the right number of

402
00:23:56.720 --> 00:24:00.039
<v Speaker 3>people by ethnicity that's correct and matches. For example, the

403
00:24:00.119 --> 00:24:02.960
<v Speaker 3>voter file, it might not be enough. Then it matches

404
00:24:03.200 --> 00:24:06.480
<v Speaker 3>by having the correct composition in a state and having

405
00:24:06.839 --> 00:24:09.599
<v Speaker 3>the right numbers of people by congressional districts or county

406
00:24:10.079 --> 00:24:13.680
<v Speaker 3>or precinct. It then gets into what is the interaction

407
00:24:14.119 --> 00:24:18.599
<v Speaker 3>between the two. And that's critical because what we've seen

408
00:24:18.640 --> 00:24:23.440
<v Speaker 3>across the developed world is very complex voter flows, with

409
00:24:23.559 --> 00:24:26.839
<v Speaker 3>people swapping who they're going to vote for between elections

410
00:24:27.000 --> 00:24:30.160
<v Speaker 3>in a way that is very complex that it becomes

411
00:24:30.160 --> 00:24:34.960
<v Speaker 3>doubly hard if you can't get those interactions correct. And

412
00:24:35.000 --> 00:24:39.599
<v Speaker 3>once you've done all of that, Henry, you still have

413
00:24:39.640 --> 00:24:43.880
<v Speaker 3>to ask the right question. And the way I think

414
00:24:43.920 --> 00:24:47.599
<v Speaker 3>about it is that when you're polling a race or

415
00:24:47.759 --> 00:24:51.519
<v Speaker 3>a federal, national, or sub regional election, you'd have thought

416
00:24:51.599 --> 00:24:54.240
<v Speaker 3>that asking what we call the horse race polling, which

417
00:24:54.279 --> 00:24:58.200
<v Speaker 3>is candidate X versus candidate Y for party X and

418
00:24:58.240 --> 00:25:02.640
<v Speaker 3>Party Y. You'd have thought simply asking the exact question

419
00:25:03.279 --> 00:25:05.440
<v Speaker 3>is the right way to go about it. But what

420
00:25:05.559 --> 00:25:08.279
<v Speaker 3>I have found, both from our work at Focal Data

421
00:25:08.279 --> 00:25:12.720
<v Speaker 3>but also other colleagues, is that really polling operates on

422
00:25:13.000 --> 00:25:16.480
<v Speaker 3>much more of a traffic light system, which is about signals,

423
00:25:16.640 --> 00:25:21.119
<v Speaker 3>which is saying my horse race polling might say that

424
00:25:22.119 --> 00:25:25.119
<v Speaker 3>this candidate is four points ahead of this candidate, But

425
00:25:25.200 --> 00:25:28.440
<v Speaker 3>when I look into the issue set, in other words,

426
00:25:29.440 --> 00:25:33.000
<v Speaker 3>which candidate is doing better on issues like healthcare or

427
00:25:33.000 --> 00:25:35.480
<v Speaker 3>immigration or the cost of living and how salient are

428
00:25:35.480 --> 00:25:38.480
<v Speaker 3>those issues, I might get a slightly different signal, and

429
00:25:38.559 --> 00:25:41.119
<v Speaker 3>that's as important all I have. Might, for example, ask

430
00:25:42.559 --> 00:25:45.400
<v Speaker 3>my survey respondents, who do you think is going to

431
00:25:45.480 --> 00:25:50.000
<v Speaker 3>win in your seat? Because sometimes if I haven't captured

432
00:25:50.079 --> 00:25:53.000
<v Speaker 3>exactly white people, I still might be able to get

433
00:25:53.000 --> 00:25:57.240
<v Speaker 3>a better community based prediction, and I would necessarily for

434
00:25:57.400 --> 00:26:01.279
<v Speaker 3>voting intention, or I might ask whole heap of other questions.

435
00:26:01.480 --> 00:26:05.000
<v Speaker 3>And generally what pulsters will do and it's not a

436
00:26:05.119 --> 00:26:09.519
<v Speaker 3>defensive or kind of an obtuse way of answering the question,

437
00:26:10.319 --> 00:26:12.960
<v Speaker 3>but many of them will operate, whether it's conscious or not.

438
00:26:13.319 --> 00:26:19.000
<v Speaker 3>A traffic light system, which is on balance, the traffic

439
00:26:19.079 --> 00:26:21.720
<v Speaker 3>lights suggests in full green that this candidate is going

440
00:26:21.759 --> 00:26:25.839
<v Speaker 3>to win with some caveats and amber you know, amber

441
00:26:25.920 --> 00:26:29.279
<v Speaker 3>lights on these types of questions and these types of analyses.

442
00:26:29.720 --> 00:26:33.880
<v Speaker 3>And that is the job of the polster, which is

443
00:26:34.160 --> 00:26:41.720
<v Speaker 3>the constant trade off between signal and conviction, and that

444
00:26:41.720 --> 00:26:44.960
<v Speaker 3>can be very hard, but it is very enjoyable. And

445
00:26:45.920 --> 00:26:48.960
<v Speaker 3>I guess the other aspect of it is that you

446
00:26:49.039 --> 00:26:53.079
<v Speaker 3>then might turn your head to the other half of

447
00:26:53.160 --> 00:26:56.880
<v Speaker 3>the field of market research, which is qualitative work, which

448
00:26:56.920 --> 00:27:00.240
<v Speaker 3>is then listening to people's conversations. Whether that's why are

449
00:27:00.279 --> 00:27:04.880
<v Speaker 3>focus groups or as is now happening across the world

450
00:27:05.559 --> 00:27:10.160
<v Speaker 3>actually AI moderated interviews with real people, which means that

451
00:27:10.200 --> 00:27:12.599
<v Speaker 3>you can do kind of focus groups in scale, and

452
00:27:12.640 --> 00:27:15.839
<v Speaker 3>that again might give you a different signal. And we

453
00:27:15.960 --> 00:27:19.440
<v Speaker 3>have seen in many elections where the qualitative data and

454
00:27:19.480 --> 00:27:21.960
<v Speaker 3>the conversations. You know, not to quote Jeff Bezos, but

455
00:27:22.039 --> 00:27:24.480
<v Speaker 3>you know when he says, when I'm got the anecdotes

456
00:27:24.519 --> 00:27:26.279
<v Speaker 3>and the data, I go with the anecdotes. That's not

457
00:27:26.359 --> 00:27:29.440
<v Speaker 3>quite how we operate. But it is a totally different

458
00:27:29.559 --> 00:27:34.839
<v Speaker 3>and fairly independent signal of polling data. So in the round, Henry,

459
00:27:35.119 --> 00:27:39.440
<v Speaker 3>it's exciting, it's interesting. It's far more interesting understanding the

460
00:27:39.440 --> 00:27:46.440
<v Speaker 3>why than what, because any given poster will call elections incorrectly.

461
00:27:46.559 --> 00:27:50.799
<v Speaker 3>The most important thing is that that there's transparency and

462
00:27:50.839 --> 00:27:53.480
<v Speaker 3>clarity over why you were wrong or right. Actually, you

463
00:27:53.519 --> 00:27:56.200
<v Speaker 3>can be right for the wrong reasons, so that you

464
00:27:56.240 --> 00:28:01.200
<v Speaker 3>can then go and adjust, but it's also about providing

465
00:28:01.720 --> 00:28:06.680
<v Speaker 3>confidence bands and caveats so that people understand exactly what's

466
00:28:06.720 --> 00:28:10.880
<v Speaker 3>being articulated. But that's had in an Internet age, in

467
00:28:10.920 --> 00:28:14.000
<v Speaker 3>an age of headlines, in an age of the need

468
00:28:14.240 --> 00:28:19.039
<v Speaker 3>to be one hundred percent convinced of a particular thing.

469
00:28:19.799 --> 00:28:23.799
<v Speaker 3>That's a hard thing to articulate today. So it is

470
00:28:23.839 --> 00:28:29.880
<v Speaker 3>a real mixture of listening, of probing, of asking none

471
00:28:29.920 --> 00:28:35.519
<v Speaker 3>obvious things, of data extraction, data modeling, and then storytelling

472
00:28:35.799 --> 00:28:39.680
<v Speaker 3>because the why is not that important leading up to

473
00:28:39.680 --> 00:28:41.960
<v Speaker 3>the election, but the why is absolutely critical for the

474
00:28:42.039 --> 00:28:46.599
<v Speaker 3>years after, because there are many circumstances in which election

475
00:28:46.720 --> 00:28:53.839
<v Speaker 3>outcomes are not particularly kind of described or interpreted correctly.

476
00:28:53.880 --> 00:28:58.359
<v Speaker 2>Basically, that happens here. I know it's happened in Britain.

477
00:28:58.400 --> 00:29:01.799
<v Speaker 2>I do want to ask a couple of questions. Polling

478
00:29:01.839 --> 00:29:03.480
<v Speaker 2>failure isn't just an American thing?

479
00:29:03.720 --> 00:29:03.880
<v Speaker 3>You know?

480
00:29:04.039 --> 00:29:07.000
<v Speaker 2>Is that you had the Brexit large polling fail You

481
00:29:07.160 --> 00:29:12.319
<v Speaker 2>had virtually everybody predicted the Labor victory correctly but massively

482
00:29:12.359 --> 00:29:18.039
<v Speaker 2>inflated vote shares for Labor. Both of those instances created,

483
00:29:18.240 --> 00:29:22.400
<v Speaker 2>of course, significant introspection in the British polling industry, and

484
00:29:22.440 --> 00:29:27.200
<v Speaker 2>you have a much more collective British polling council that

485
00:29:27.440 --> 00:29:30.920
<v Speaker 2>encourages that. You know, us being in the wild West,

486
00:29:30.960 --> 00:29:33.359
<v Speaker 2>we don't have an American polling council that kind of

487
00:29:33.400 --> 00:29:36.039
<v Speaker 2>gets all the leading bolsters together and says, hey, how

488
00:29:36.079 --> 00:29:39.079
<v Speaker 2>did we collectively all screw up? Everyone says they have

489
00:29:39.200 --> 00:29:42.880
<v Speaker 2>the secret sauce and tries to sell it. What do

490
00:29:42.920 --> 00:29:49.519
<v Speaker 2>you think British pollsters learned from twenty sixteen and why

491
00:29:49.559 --> 00:29:53.440
<v Speaker 2>did the things they learned from twenty sixteen not lead

492
00:29:53.759 --> 00:29:56.839
<v Speaker 2>to They led to the correct directional outcome, but not

493
00:29:56.920 --> 00:30:01.400
<v Speaker 2>the magnitude outcome in twenty twenty four because that of

494
00:30:01.440 --> 00:30:06.640
<v Speaker 2>course relates to method of how one's sample or questioning

495
00:30:06.720 --> 00:30:10.480
<v Speaker 2>or taking signals in some way messed up in pretty

496
00:30:10.519 --> 00:30:11.319
<v Speaker 2>big ways.

497
00:30:12.960 --> 00:30:18.440
<v Speaker 3>So yeah, I mean UK polling absolutely has is very

498
00:30:18.480 --> 00:30:22.279
<v Speaker 3>transparent endeavor, but there have been a number of misses

499
00:30:22.359 --> 00:30:24.319
<v Speaker 3>and a number of successes, and I kind of i'll

500
00:30:24.359 --> 00:30:28.680
<v Speaker 3>talk through talk through that that kind of history. The

501
00:30:28.680 --> 00:30:32.039
<v Speaker 3>twenty sixteen referendum was actually preceded by the twenty fifteen

502
00:30:32.200 --> 00:30:35.000
<v Speaker 3>polling miss There was we had a general election, right

503
00:30:35.200 --> 00:30:38.839
<v Speaker 3>so where David Cameron, the Conservative leader, want a clear

504
00:30:38.960 --> 00:30:45.200
<v Speaker 3>majority against the Labor Party and the polls had a

505
00:30:45.319 --> 00:30:48.079
<v Speaker 3>very small conservative leader or actually a tie, and I

506
00:30:48.119 --> 00:30:50.519
<v Speaker 3>think he ended up winning by seven points, which in

507
00:30:50.559 --> 00:30:54.960
<v Speaker 3>the first part the post system is definitely and there

508
00:30:54.960 --> 00:30:57.880
<v Speaker 3>was actually a polling inquiry then in twenty fifteen, which

509
00:30:58.039 --> 00:31:01.279
<v Speaker 3>was that it was a it was a data issue

510
00:31:01.279 --> 00:31:04.799
<v Speaker 3>as opposed to a modeling issue. People. You know, it

511
00:31:04.839 --> 00:31:08.599
<v Speaker 3>sounds very tautological, which is that people's underlying samples because

512
00:31:08.599 --> 00:31:12.160
<v Speaker 3>we had to send them in and there was a convening,

513
00:31:12.599 --> 00:31:16.880
<v Speaker 3>and people's underlying samples did not have enough what we

514
00:31:17.039 --> 00:31:21.240
<v Speaker 3>call time poor center right voters as opposed to populace

515
00:31:21.319 --> 00:31:24.039
<v Speaker 3>right voters, which then is a problem that comes is

516
00:31:24.079 --> 00:31:28.200
<v Speaker 3>a problem that appears the next year in twenty sixteen.

517
00:31:28.880 --> 00:31:32.160
<v Speaker 3>Is actually it was less of appollingness because there were

518
00:31:32.279 --> 00:31:35.960
<v Speaker 3>posters who got it fundamentally correct. You go was the

519
00:31:36.039 --> 00:31:41.640
<v Speaker 3>UK firm that actually very quietly unveiled what's called MRP.

520
00:31:42.160 --> 00:31:45.039
<v Speaker 1>Which I do want to get to m RPS.

521
00:31:45.039 --> 00:31:49.920
<v Speaker 3>Right, which we're going to get to MRPs later and

522
00:31:49.960 --> 00:31:52.759
<v Speaker 3>we don't need to talk for it now, but you know,

523
00:31:52.799 --> 00:31:56.400
<v Speaker 3>it's a technique that fundamentally allows it's called multi level

524
00:31:56.400 --> 00:31:58.839
<v Speaker 3>reggression and post gratification. As modeling technique where you can

525
00:31:58.880 --> 00:32:02.720
<v Speaker 3>bine national sample to get to local level estimates. Basically,

526
00:32:02.799 --> 00:32:06.319
<v Speaker 3>it's an American technique. It was created by Andrew Gelman,

527
00:32:06.319 --> 00:32:09.039
<v Speaker 3>who's a professor of political science at Columbia. Again another

528
00:32:09.160 --> 00:32:14.279
<v Speaker 3>US innovation, but that very quietly got the election and

529
00:32:14.319 --> 00:32:19.759
<v Speaker 3>the referendum in twenty sixteen correct and even the range correct.

530
00:32:19.799 --> 00:32:22.119
<v Speaker 3>But of course you have also released a traditional poll

531
00:32:22.640 --> 00:32:27.079
<v Speaker 3>at the referendum which was not correct, and there that

532
00:32:27.240 --> 00:32:29.799
<v Speaker 3>was very much a case of a little bit of

533
00:32:29.799 --> 00:32:32.039
<v Speaker 3>a case of data input, but actually there was also

534
00:32:32.079 --> 00:32:35.960
<v Speaker 3>a case of maybe there's a future here of modeling

535
00:32:36.000 --> 00:32:38.319
<v Speaker 3>technique that needs to be looked at, and that was

536
00:32:38.440 --> 00:32:41.839
<v Speaker 3>very much verified at the then twenty seventeen general election. Man,

537
00:32:42.000 --> 00:32:44.359
<v Speaker 3>your listeners at this stage may have forgotten that we

538
00:32:44.480 --> 00:32:48.559
<v Speaker 3>basically held general elections and referendums as annual events for

539
00:32:48.640 --> 00:32:53.759
<v Speaker 3>a period of British political history. And again the polls

540
00:32:54.440 --> 00:32:57.720
<v Speaker 3>kind of failed and that was again an issue of

541
00:32:57.759 --> 00:32:59.920
<v Speaker 3>time because but you got it.

542
00:33:00.079 --> 00:33:03.480
<v Speaker 1>MRP did not. That's where the MRP made it did

543
00:33:03.559 --> 00:33:04.119
<v Speaker 1>not fail.

544
00:33:04.240 --> 00:33:09.799
<v Speaker 3>It captured very accurately a resurgent populace left Labor Party

545
00:33:09.880 --> 00:33:13.720
<v Speaker 3>led by Jeremy Corbyn, and they very specifically didn't just

546
00:33:13.799 --> 00:33:16.799
<v Speaker 3>call a national picture, but they forecast the six hundred

547
00:33:16.839 --> 00:33:21.599
<v Speaker 3>and thirty two seats across Great Britain in our Parliament,

548
00:33:22.160 --> 00:33:26.799
<v Speaker 3>including some very surprising changes, including my home seat of Canterbury,

549
00:33:27.440 --> 00:33:30.400
<v Speaker 3>which had been a kind of traditional Conservative vote with

550
00:33:30.920 --> 00:33:34.000
<v Speaker 3>you know, has a cathedral city flipped labor and.

551
00:33:36.240 --> 00:33:41.559
<v Speaker 1>Field one. Sorry that one was exactly that's right.

552
00:33:41.599 --> 00:33:43.559
<v Speaker 3>Who now sits? I think is an independent MP? But

553
00:33:43.640 --> 00:33:50.039
<v Speaker 3>she she she won that. And again the different problem

554
00:33:50.480 --> 00:33:54.559
<v Speaker 3>that came about was of late swing, where the twenty

555
00:33:54.599 --> 00:33:59.920
<v Speaker 3>seventeen election genuinely had huge vote shift changes during the

556
00:34:00.039 --> 00:34:02.880
<v Speaker 3>election cycle and the campaign, and I was polling it

557
00:34:03.039 --> 00:34:07.079
<v Speaker 3>very very closely, and Jeremy Corbyn put on double digits

558
00:34:07.759 --> 00:34:10.760
<v Speaker 3>in vote share. If you think you think about the

559
00:34:10.840 --> 00:34:12.679
<v Speaker 3>last time that might have happened in a US context,

560
00:34:12.679 --> 00:34:15.800
<v Speaker 3>it's quite rare for that to happen. But Yugov's model

561
00:34:15.880 --> 00:34:18.519
<v Speaker 3>was able to kind of adjust for that and from

562
00:34:18.559 --> 00:34:21.800
<v Speaker 3>a time period perspective, so we really all as an

563
00:34:21.800 --> 00:34:26.000
<v Speaker 3>industry there was a sigh of relief of the twenty

564
00:34:26.119 --> 00:34:32.039
<v Speaker 3>nineteen get Brexit done. Boris Johnson's not landslide but very comfortable,

565
00:34:32.800 --> 00:34:35.639
<v Speaker 3>large majority victory where the Conservative Party I think got

566
00:34:35.679 --> 00:34:39.440
<v Speaker 3>three hundred and sixty five seats, having declined previously in

567
00:34:39.480 --> 00:34:45.840
<v Speaker 3>twenty seventeen, and everyone was directionally correct. Everyone forecast that

568
00:34:45.880 --> 00:34:49.039
<v Speaker 3>there was going to be a Conservative minority, but underneath

569
00:34:49.079 --> 00:34:52.960
<v Speaker 3>the surface, actually the scale of the Conservative victory was

570
00:34:53.000 --> 00:34:58.760
<v Speaker 3>slightly underpredicted and wasn't wasn't that fully understood or known.

571
00:34:59.320 --> 00:35:02.559
<v Speaker 3>So and so we get to the twenty twenty four election,

572
00:35:02.599 --> 00:35:04.360
<v Speaker 3>which is I think what your question was at, which

573
00:35:04.599 --> 00:35:10.440
<v Speaker 3>was by far and away the kind of biggest failure

574
00:35:10.519 --> 00:35:13.719
<v Speaker 3>of the UK polling industry, hidden by the fact that

575
00:35:13.760 --> 00:35:16.559
<v Speaker 3>everyone was directionally correct. There was such a large difference

576
00:35:17.280 --> 00:35:20.440
<v Speaker 3>between the fortunes of the Conservative Party, which was back

577
00:35:20.480 --> 00:35:23.840
<v Speaker 3>then hemorrhaging votes to reform UK, which went from kind

578
00:35:23.840 --> 00:35:27.599
<v Speaker 3>of low double digits to fourteen percent. You know, the

579
00:35:27.679 --> 00:35:30.639
<v Speaker 3>gap at a seat level was very very clear. So

580
00:35:30.639 --> 00:35:32.719
<v Speaker 3>it was very clear from about a year a year

581
00:35:32.760 --> 00:35:34.760
<v Speaker 3>and a half that the Labor Party were going to

582
00:35:34.800 --> 00:35:39.760
<v Speaker 3>win a majority. But again in terms of vote shares,

583
00:35:39.800 --> 00:35:41.880
<v Speaker 3>the industry was five or six points out, which was

584
00:35:42.039 --> 00:35:44.840
<v Speaker 3>much bigger polling failure than any of the previous elections

585
00:35:44.960 --> 00:35:48.840
<v Speaker 3>referendums that I've just described, and it was specifically the

586
00:35:48.920 --> 00:35:54.519
<v Speaker 3>Labor Party vote where which was overscored by literally everyone

587
00:35:54.679 --> 00:35:57.559
<v Speaker 3>bar a few. And we can get into the very

588
00:35:57.559 --> 00:36:00.920
<v Speaker 3>specifics because there were a few holding companies that have

589
00:36:01.039 --> 00:36:03.719
<v Speaker 3>slightly different approaches that got it right. But interestingly, Henry,

590
00:36:04.280 --> 00:36:07.880
<v Speaker 3>the models were broadly okay. So the models told the

591
00:36:07.960 --> 00:36:10.880
<v Speaker 3>story that Labour were going to win between four hundred

592
00:36:11.039 --> 00:36:13.840
<v Speaker 3>four point fifty seats they won towards the lower end

593
00:36:13.880 --> 00:36:17.679
<v Speaker 3>and around four point fifteen that somehow the vote shares

594
00:36:17.719 --> 00:36:21.880
<v Speaker 3>were totally out And that's because in many, many safe

595
00:36:22.440 --> 00:36:26.679
<v Speaker 3>labor areas, the vote share shifted dramatically down and they

596
00:36:26.679 --> 00:36:29.960
<v Speaker 3>won many many seats on a hugely marginal vote share,

597
00:36:30.760 --> 00:36:34.599
<v Speaker 3>which is the story of how British politics has continued

598
00:36:34.639 --> 00:36:38.960
<v Speaker 3>to be volatile, which is the ability to win quite

599
00:36:38.960 --> 00:36:43.400
<v Speaker 3>big mandates on very small slices of public opinion. You know,

600
00:36:43.480 --> 00:36:47.280
<v Speaker 3>we talk about we might get into the president's low

601
00:36:47.320 --> 00:36:49.320
<v Speaker 3>approval ratings, but I was looking at it. He's still

602
00:36:49.360 --> 00:36:54.119
<v Speaker 3>liked by thirty seven percent of the American public. Most

603
00:36:54.239 --> 00:36:59.199
<v Speaker 3>European leaders would be very happy with thirty seven percent

604
00:36:59.239 --> 00:37:01.000
<v Speaker 3>of the public. Now, some of that as a function

605
00:37:01.039 --> 00:37:05.840
<v Speaker 3>of a two party system, but it is it is structurally,

606
00:37:06.000 --> 00:37:08.920
<v Speaker 3>very very different. So that's the story of UK polling.

607
00:37:09.280 --> 00:37:11.639
<v Speaker 3>But again there's been some fantastic work that's gone on

608
00:37:11.840 --> 00:37:14.159
<v Speaker 3>in the industry around the modeling, which I think more

609
00:37:14.199 --> 00:37:18.039
<v Speaker 3>people are feeling comfortable with. There are challenges around our

610
00:37:18.559 --> 00:37:25.039
<v Speaker 3>rapidly diversifying society, so much of the pollingness was concentrated

611
00:37:25.119 --> 00:37:29.960
<v Speaker 3>in ethnically diverse areas and non white voters. We now

612
00:37:29.960 --> 00:37:33.360
<v Speaker 3>have a multi party system in the UK, which is

613
00:37:33.920 --> 00:37:35.960
<v Speaker 3>a very obvious point and not I think the feature

614
00:37:36.000 --> 00:37:37.880
<v Speaker 3>of this discussion. But we have a Green Party, we

615
00:37:37.960 --> 00:37:42.000
<v Speaker 3>have a populist right party called Reform, We have independents

616
00:37:42.199 --> 00:37:46.880
<v Speaker 3>who vote very much in certain areas along ethnic blocks

617
00:37:46.920 --> 00:37:50.599
<v Speaker 3>based on foreign policy. We then have a Liberal Democrat party,

618
00:37:50.719 --> 00:37:54.519
<v Speaker 3>and then we have two nationalist party parties. Now this

619
00:37:54.639 --> 00:38:00.280
<v Speaker 3>creates a massively complex picture in which to get it right.

620
00:38:00.360 --> 00:38:04.960
<v Speaker 3>The bar is very very high. Most countries with a

621
00:38:05.119 --> 00:38:08.960
<v Speaker 3>system that is that complex with that many parties five

622
00:38:09.000 --> 00:38:16.960
<v Speaker 3>effected parties, but in many areas seven means most countries

623
00:38:16.960 --> 00:38:19.800
<v Speaker 3>that have that many number of parties have a proportional system,

624
00:38:20.400 --> 00:38:23.639
<v Speaker 3>like Germany, which you know doesn't have overhang seats anymore,

625
00:38:23.639 --> 00:38:27.280
<v Speaker 3>but it's still proportional. Beyond the threshold. So that's a

626
00:38:27.360 --> 00:38:29.559
<v Speaker 3>unique challenge, which means we all have to work very

627
00:38:29.639 --> 00:38:36.239
<v Speaker 3>very hard. But each country presents very very different challenges.

628
00:38:36.360 --> 00:38:39.639
<v Speaker 3>They're not the same in terms of the difficulties in

629
00:38:39.679 --> 00:38:43.440
<v Speaker 3>getting data, and the modeling is never the same either,

630
00:38:43.559 --> 00:38:46.559
<v Speaker 3>even if the kind of toolkit can have some resonance

631
00:38:46.559 --> 00:38:47.559
<v Speaker 3>between different countries.

632
00:38:50.480 --> 00:38:52.199
<v Speaker 2>Well, that actually got to is going to get to

633
00:38:52.199 --> 00:38:54.519
<v Speaker 2>my next question, which is you do polling in another country,

634
00:38:54.559 --> 00:38:56.719
<v Speaker 2>but let's focus on the United States because you have

635
00:38:56.840 --> 00:39:01.679
<v Speaker 2>started to do polled. I featured your mrp of America's

636
00:39:02.079 --> 00:39:06.280
<v Speaker 2>poll and my Washington Post column and some people said,

637
00:39:06.960 --> 00:39:09.559
<v Speaker 2>I'll bet you money that that's not going to be

638
00:39:09.639 --> 00:39:14.119
<v Speaker 2>right because it's too good for Republicans. We can come

639
00:39:14.159 --> 00:39:15.840
<v Speaker 2>back to that in a minute. But what are the

640
00:39:15.920 --> 00:39:22.559
<v Speaker 2>methodological challenges that you that are different here than you face,

641
00:39:22.920 --> 00:39:25.559
<v Speaker 2>say in your native country of Great Britain or are

642
00:39:25.559 --> 00:39:30.000
<v Speaker 2>they pretty much very similar? Is that you have similar

643
00:39:30.039 --> 00:39:34.400
<v Speaker 2>non response rates and similar technological barriers with different types

644
00:39:34.400 --> 00:39:38.280
<v Speaker 2>of voters getting through them. How do you approach an

645
00:39:38.320 --> 00:39:40.920
<v Speaker 2>American pole that might be a little bit different than

646
00:39:41.440 --> 00:39:42.960
<v Speaker 2>approaching a British pole.

647
00:39:43.159 --> 00:39:45.320
<v Speaker 3>It's a great question, and so the demand for polling

648
00:39:45.320 --> 00:39:47.280
<v Speaker 3>in the US, and the structure of the system is

649
00:39:47.360 --> 00:39:49.800
<v Speaker 3>very different. It depends on what you're pulling. So let's

650
00:39:49.800 --> 00:39:53.599
<v Speaker 3>say you did a congressional districts poll or for example,

651
00:39:53.639 --> 00:39:57.000
<v Speaker 3>a county level poll or a state pole. The expectation

652
00:39:58.239 --> 00:40:01.280
<v Speaker 3>is that you don't model those out comes. You actually

653
00:40:01.400 --> 00:40:04.760
<v Speaker 3>do a survey of people who literally live in those areas,

654
00:40:06.199 --> 00:40:09.079
<v Speaker 3>and that means that you approach it very differently. It

655
00:40:09.119 --> 00:40:11.880
<v Speaker 3>means that you have to find sample from wherever you can,

656
00:40:11.960 --> 00:40:16.679
<v Speaker 3>whether that's online, whether it's by cell, whether that's texta web.

657
00:40:17.880 --> 00:40:22.480
<v Speaker 3>You know, the demand for modeled answers by state is

658
00:40:22.599 --> 00:40:24.920
<v Speaker 3>much less because people just say, well, you can just

659
00:40:25.079 --> 00:40:27.920
<v Speaker 3>do a pole of Texas of a thousand people, which

660
00:40:27.920 --> 00:40:31.599
<v Speaker 3>is what we did last week, instead of modeling what

661
00:40:31.920 --> 00:40:37.639
<v Speaker 3>the state would hypothetically do. If the demographics of Texas

662
00:40:37.679 --> 00:40:40.000
<v Speaker 3>and the demographics of the country look like X, then

663
00:40:40.039 --> 00:40:42.480
<v Speaker 3>you kind of impute what Texas should look like, which

664
00:40:42.519 --> 00:40:44.920
<v Speaker 3>is what a model does. People just say, we'll just

665
00:40:45.360 --> 00:40:47.440
<v Speaker 3>run a pole, just ask people what it is. So

666
00:40:48.039 --> 00:40:50.679
<v Speaker 3>there is that challenge in the US where there is

667
00:40:50.719 --> 00:40:53.360
<v Speaker 3>just a broad expectation and if you're going to make

668
00:40:53.400 --> 00:40:56.039
<v Speaker 3>a horse race call or do a pole of an area.

669
00:40:56.159 --> 00:40:58.960
<v Speaker 3>It is a pole and not a model. Now that's

670
00:40:59.039 --> 00:41:03.079
<v Speaker 3>interesting because what we found at the twenty twenty four

671
00:41:03.199 --> 00:41:07.039
<v Speaker 3>cycle is that our modeling of how states would vote

672
00:41:07.079 --> 00:41:10.239
<v Speaker 3>was in many ways more stable and more accurate than

673
00:41:10.440 --> 00:41:15.400
<v Speaker 3>literally doing a pole of the state, which is you know,

674
00:41:15.639 --> 00:41:17.360
<v Speaker 3>I think we talked about it at the time a

675
00:41:17.400 --> 00:41:21.519
<v Speaker 3>couple of years ago, and as we did the MRP's

676
00:41:21.960 --> 00:41:26.679
<v Speaker 3>multi level regression and post ratification. You know, these big

677
00:41:26.840 --> 00:41:30.599
<v Speaker 3>computationally intensive models that split the public into millions of groups,

678
00:41:30.679 --> 00:41:34.440
<v Speaker 3>and you take a national survey that then basically works

679
00:41:34.480 --> 00:41:39.880
<v Speaker 3>out what demographics were, which way in which area all

680
00:41:39.880 --> 00:41:42.440
<v Speaker 3>interacted with each other, and then you basically work out

681
00:41:42.440 --> 00:41:44.280
<v Speaker 3>what's happening at a national level, and you impute what's

682
00:41:44.280 --> 00:41:46.960
<v Speaker 3>happening at a state level. Those seem to be more

683
00:41:47.000 --> 00:41:50.000
<v Speaker 3>stable and more accurate than going in and doing a

684
00:41:50.039 --> 00:41:54.239
<v Speaker 3>state pole. That's one challenge in the US. The second

685
00:41:54.280 --> 00:41:56.440
<v Speaker 3>is obviously the resources are different. You guys have a

686
00:41:56.519 --> 00:41:59.679
<v Speaker 3>voter file right where you can literally see who's on there.

687
00:41:59.719 --> 00:42:02.559
<v Speaker 3>We have very different laws in the UK regarding GDPR.

688
00:42:02.639 --> 00:42:05.280
<v Speaker 3>We don't have things like a voter file that polsters

689
00:42:05.360 --> 00:42:08.159
<v Speaker 3>and can use, and the laws were even stricter in

690
00:42:08.199 --> 00:42:12.400
<v Speaker 3>Europe as well, so the challenges are very different in

691
00:42:12.480 --> 00:42:15.719
<v Speaker 3>terms of who doesn't want to take polls. There are similarities, Henry,

692
00:42:16.800 --> 00:42:20.320
<v Speaker 3>there are. We do see groups of people who are

693
00:42:20.360 --> 00:42:26.280
<v Speaker 3>low trust. By that I mean less trusting of institutions,

694
00:42:26.360 --> 00:42:30.480
<v Speaker 3>less trusting upholsters, less willing to take surveys. They do

695
00:42:30.519 --> 00:42:34.480
<v Speaker 3>appear to be groups that are harder to survey, particularly

696
00:42:34.480 --> 00:42:37.559
<v Speaker 3>across the English speaking world. I wouldn't necessarily say Europe,

697
00:42:37.559 --> 00:42:41.800
<v Speaker 3>for example, French polls don't underestimate Bardella all A Penn.

698
00:42:41.840 --> 00:42:44.360
<v Speaker 3>We saw no evidence of that at the Assembly elections,

699
00:42:44.960 --> 00:42:47.719
<v Speaker 3>but that does appear to be a particularly acute problem

700
00:42:47.960 --> 00:42:52.119
<v Speaker 3>in the English speaking world, I think where and again,

701
00:42:52.239 --> 00:42:56.360
<v Speaker 3>I would also say that some of the failures at

702
00:42:56.360 --> 00:42:59.840
<v Speaker 3>a twenty to twenty four election, which were captured correctly

703
00:43:00.079 --> 00:43:03.559
<v Speaker 3>by another British friend called Jail Partners, run by James Johnson,

704
00:43:04.719 --> 00:43:09.960
<v Speaker 3>a colleague of mine who's actually America based for British.

705
00:43:08.840 --> 00:43:11.360
<v Speaker 1>At the whole, was on this podcast two years ago

706
00:43:11.440 --> 00:43:12.159
<v Speaker 1>to talk about.

707
00:43:12.920 --> 00:43:15.119
<v Speaker 3>A great guy, and I think, you know, they very

708
00:43:15.239 --> 00:43:18.880
<v Speaker 3>accurately captured how what was going on amongst the low

709
00:43:19.000 --> 00:43:23.559
<v Speaker 3>trust non white vote across the seven Swing states which

710
00:43:23.639 --> 00:43:27.800
<v Speaker 3>was absolutely critical if you at least that's the thesis

711
00:43:27.800 --> 00:43:30.800
<v Speaker 3>of David Shaw, the polster who runs Blue Rose, a

712
00:43:30.840 --> 00:43:34.199
<v Speaker 3>Democrat who's a bit different, you know. He was very

713
00:43:34.280 --> 00:43:36.599
<v Speaker 3>much in the view that the seven swing states were

714
00:43:36.679 --> 00:43:40.440
<v Speaker 3>very much delivered by this kind of low trust, ethnically diverse,

715
00:43:40.920 --> 00:43:45.440
<v Speaker 3>slightly younger group of people who basically came out voted

716
00:43:45.480 --> 00:43:48.039
<v Speaker 3>for Trump and then didn't bother voting down ballot, which

717
00:43:48.079 --> 00:43:52.400
<v Speaker 3>slightly explains the number of split states at the election. Again,

718
00:43:53.239 --> 00:43:55.800
<v Speaker 3>I don't think that's a million miles away from what

719
00:43:55.880 --> 00:43:58.800
<v Speaker 3>happened with the overestimation of the labor vote in the

720
00:43:58.920 --> 00:44:03.599
<v Speaker 3>UK election, So there are resonances. What I have to

721
00:44:03.639 --> 00:44:06.880
<v Speaker 3>say is the reason why things have converged is partly

722
00:44:06.880 --> 00:44:10.119
<v Speaker 3>because the US is no longer a low turnout democracy,

723
00:44:10.880 --> 00:44:14.320
<v Speaker 3>and turnout in the UK has actually declined quite sharply

724
00:44:14.800 --> 00:44:16.679
<v Speaker 3>over the last ten years. So we've gone from around

725
00:44:16.719 --> 00:44:20.719
<v Speaker 3>seventy two percent turnout at the twenty seventeen election. I

726
00:44:20.760 --> 00:44:22.639
<v Speaker 3>think it was about fifty nine to fifty eight percent

727
00:44:22.760 --> 00:44:26.000
<v Speaker 3>at the last election, which I believe is actually lower

728
00:44:26.000 --> 00:44:29.000
<v Speaker 3>than the last president the US presidential.

729
00:44:28.719 --> 00:44:30.960
<v Speaker 2>It is, we we at about sixty two percent, and

730
00:44:31.039 --> 00:44:35.079
<v Speaker 2>by that for my listeners. We're not talking about registered voters.

731
00:44:35.119 --> 00:44:38.719
<v Speaker 2>We're talking about what we were called the voter eligible population,

732
00:44:38.960 --> 00:44:42.320
<v Speaker 2>the citizens who are adults of voting age and not

733
00:44:42.440 --> 00:44:46.639
<v Speaker 2>disqualified by things like legal inhibitions or insanity.

734
00:44:47.559 --> 00:44:51.360
<v Speaker 3>Yeah, so there are there are, There are big residences,

735
00:44:51.840 --> 00:44:55.960
<v Speaker 3>but the challenges are always slightly different. So I mean,

736
00:44:56.119 --> 00:44:59.599
<v Speaker 3>that's why you need domain experts. That's why most of

737
00:44:59.639 --> 00:45:03.880
<v Speaker 3>the uk as that operate in America and do forecasting America,

738
00:45:04.000 --> 00:45:06.599
<v Speaker 3>we have teams of people here, some of whom are American. Right.

739
00:45:06.679 --> 00:45:10.480
<v Speaker 3>You need that domain knowledge, that expertise, You need that

740
00:45:10.559 --> 00:45:12.519
<v Speaker 3>kind of deep, deep level knowledge. Really.

741
00:45:13.559 --> 00:45:14.880
<v Speaker 2>But you know, one of the things I do want

742
00:45:14.920 --> 00:45:16.639
<v Speaker 2>to talk about is, you know, when I'm listening to this,

743
00:45:16.679 --> 00:45:19.199
<v Speaker 2>I'm thinking, gee, this sounds a lot like wine making.

744
00:45:19.440 --> 00:45:21.559
<v Speaker 2>Is that the grapes are the same, but the terr

745
00:45:21.559 --> 00:45:24.360
<v Speaker 2>are and then the skill of the wine maker produces

746
00:45:24.760 --> 00:45:27.880
<v Speaker 2>you know, the difference between a ten dollars bottle of

747
00:45:27.920 --> 00:45:31.199
<v Speaker 2>cable wine and a five hundred dollars bottle of petrus.

748
00:45:32.559 --> 00:45:36.760
<v Speaker 2>But I got the race largely right because I dove

749
00:45:36.840 --> 00:45:38.760
<v Speaker 2>into what I do is I look at when I

750
00:45:38.800 --> 00:45:42.000
<v Speaker 2>make my biennial predictions, which you know, I owe it

751
00:45:42.039 --> 00:45:45.599
<v Speaker 2>to my readers or followers to you know, make the

752
00:45:45.639 --> 00:45:47.920
<v Speaker 2>hard call. Instead of saying it's too hard, it's too

753
00:45:47.920 --> 00:45:49.719
<v Speaker 2>close to call, I said, now, this is what I think.

754
00:45:49.760 --> 00:45:53.480
<v Speaker 1>I'm usually directionally right. Once I was very very wrong.

755
00:45:55.119 --> 00:45:58.000
<v Speaker 2>But I looked at all the average and what struck

756
00:45:58.000 --> 00:46:00.639
<v Speaker 2>me the reason why they thought Kamala Harris would narrowly

757
00:46:00.639 --> 00:46:02.360
<v Speaker 2>win and I thought Trump would win and might win

758
00:46:02.400 --> 00:46:05.679
<v Speaker 2>the popular vote was because they systematically underestimated the share

759
00:46:05.679 --> 00:46:10.239
<v Speaker 2>of Republicans in the elector, which is a way you

760
00:46:10.280 --> 00:46:13.920
<v Speaker 2>could have teased out of the aggregate cross tab data.

761
00:46:14.000 --> 00:46:17.960
<v Speaker 2>Exactly the methodological problem. But there was also, I think

762
00:46:18.000 --> 00:46:20.480
<v Speaker 2>for some people a modeling question, which was there was

763
00:46:20.519 --> 00:46:23.039
<v Speaker 2>one firm, which we can discuss offline. I don't want

764
00:46:23.079 --> 00:46:25.920
<v Speaker 2>to name it here. They produced a very detailed Friday

765
00:46:26.000 --> 00:46:28.440
<v Speaker 2>before the election, Here's how every state is going to vote,

766
00:46:28.480 --> 00:46:31.119
<v Speaker 2>and they got every swing state wrong, because every swing

767
00:46:31.920 --> 00:46:35.840
<v Speaker 2>with nationally and through every swing state, they over projected

768
00:46:35.840 --> 00:46:39.079
<v Speaker 2>the Democratic vote. And that's clearly a modeling question, is

769
00:46:39.119 --> 00:46:42.320
<v Speaker 2>that their model of how they were interpreting the data

770
00:46:43.119 --> 00:46:48.119
<v Speaker 2>was simply wrong. Do we run that risk if we

771
00:46:48.159 --> 00:46:53.440
<v Speaker 2>look at individual polsters that the Winemaker maybe getting too

772
00:46:53.480 --> 00:47:00.760
<v Speaker 2>much yeast in the must too early, and that if

773
00:47:00.800 --> 00:47:05.639
<v Speaker 2>we're not looking at averages or doing multi polling analysis,

774
00:47:05.639 --> 00:47:10.960
<v Speaker 2>we're just looking at one person that garbage in, garbage out.

775
00:47:10.960 --> 00:47:14.280
<v Speaker 2>If your model is wrong, even if it's been right before,

776
00:47:14.400 --> 00:47:17.159
<v Speaker 2>you will be horribly off in an important race.

777
00:47:17.679 --> 00:47:20.119
<v Speaker 3>I know the difficulty there is. Of course, you have

778
00:47:20.159 --> 00:47:23.119
<v Speaker 3>off cycle midterms as well, which have a totally different,

779
00:47:23.920 --> 00:47:26.079
<v Speaker 3>totally different elector. You know a lot of the story.

780
00:47:27.119 --> 00:47:29.079
<v Speaker 2>And one of the things we're seeing is people saying, well,

781
00:47:29.119 --> 00:47:32.800
<v Speaker 2>Democrats are more educated, therefore they'll have a larger overturnout

782
00:47:32.840 --> 00:47:36.719
<v Speaker 2>share of the election. And presumably that assumption, which has

783
00:47:36.800 --> 00:47:41.039
<v Speaker 2>been born out historically, may be correct.

784
00:47:40.639 --> 00:47:41.679
<v Speaker 1>But it may not be.

785
00:47:41.960 --> 00:47:45.800
<v Speaker 2>And we have the Virginia redistricting referendum where that should

786
00:47:45.880 --> 00:47:49.480
<v Speaker 2>have been the case, and in fact Republicans out voted Democrats,

787
00:47:49.840 --> 00:47:53.039
<v Speaker 2>and it was the lower educated Republican areas that had

788
00:47:53.119 --> 00:47:57.360
<v Speaker 2>a higher share of the twenty twenty four presidential vote

789
00:47:57.400 --> 00:48:03.000
<v Speaker 2>turnout for this April off cycle only theoretical redistrict thing initiative,

790
00:48:03.000 --> 00:48:06.360
<v Speaker 2>which if I were Polster, I would say, warning, warning,

791
00:48:06.440 --> 00:48:10.119
<v Speaker 2>Will Robinson, danger, You know that my assumption of education

792
00:48:10.320 --> 00:48:14.679
<v Speaker 2>driving turnout and low motivated Trump only voters may not

793
00:48:14.880 --> 00:48:15.320
<v Speaker 2>be right.

794
00:48:16.440 --> 00:48:21.119
<v Speaker 3>And I think the wine maker analogy is a short one,

795
00:48:21.400 --> 00:48:25.320
<v Speaker 3>maybe also because of the stretched the metaphor, Henry, all

796
00:48:25.320 --> 00:48:30.320
<v Speaker 3>the way through. Vintages are different, like firms are not

797
00:48:30.400 --> 00:48:33.559
<v Speaker 3>static things like you can have different teams working at

798
00:48:33.559 --> 00:48:37.079
<v Speaker 3>different times. You might be missing that one analyst who

799
00:48:37.119 --> 00:48:40.039
<v Speaker 3>looks at things from a completely different perspective. You might

800
00:48:40.039 --> 00:48:42.400
<v Speaker 3>have got a model that was carefully calibrated to one

801
00:48:42.440 --> 00:48:46.360
<v Speaker 3>election that the electric fundamentally changes at the next. You

802
00:48:46.440 --> 00:48:49.800
<v Speaker 3>might not feel the need to reinvent from first principles

803
00:48:49.800 --> 00:48:53.159
<v Speaker 3>what you've done because you've been successful. It's often my failure.

804
00:48:53.920 --> 00:48:56.480
<v Speaker 3>I'm very interested in firms, you know, and sometimes including

805
00:48:56.519 --> 00:48:58.920
<v Speaker 3>when I and other firms I've been involved with have

806
00:48:59.000 --> 00:49:02.239
<v Speaker 3>got it wrong. Often I'm very interested in their first

807
00:49:02.280 --> 00:49:05.480
<v Speaker 3>call after the event, because very often it could be right,

808
00:49:05.599 --> 00:49:07.920
<v Speaker 3>because they basically have to go back to the drawing

809
00:49:07.960 --> 00:49:11.119
<v Speaker 3>board and redo it. And for a pole watchers yourself,

810
00:49:11.159 --> 00:49:13.400
<v Speaker 3>I don't know of anyone who knows as much as

811
00:49:13.480 --> 00:49:16.840
<v Speaker 3>international polling as you, Henry Enough spoken to a few

812
00:49:16.880 --> 00:49:21.199
<v Speaker 3>people about this, but you know that that creates challenges

813
00:49:21.239 --> 00:49:21.679
<v Speaker 3>for observer.

814
00:49:21.760 --> 00:49:25.239
<v Speaker 2>I mean not everyone runs German language crosstabs through a

815
00:49:25.280 --> 00:49:27.320
<v Speaker 2>Google translation I understand.

816
00:49:29.360 --> 00:49:32.480
<v Speaker 3>Which I know you do. But that creates problems because

817
00:49:32.960 --> 00:49:35.480
<v Speaker 3>people who were good before can suddenly get it wrong,

818
00:49:35.880 --> 00:49:38.000
<v Speaker 3>and people who got it totally wrong before, who no

819
00:49:38.000 --> 00:49:44.559
<v Speaker 3>one's paying attention to get it fundamentally completely correct and

820
00:49:44.639 --> 00:49:49.079
<v Speaker 3>have completely and that creates a kind of a kind

821
00:49:49.119 --> 00:49:54.000
<v Speaker 3>of problem of continuity of trust. And that's very important

822
00:49:54.000 --> 00:49:56.280
<v Speaker 3>action to take a step back and to use I

823
00:49:56.280 --> 00:49:59.079
<v Speaker 3>guess the traffic light analogy, even if you're pole watching,

824
00:50:00.119 --> 00:50:02.599
<v Speaker 3>to be like, here are the signals, and I'm going

825
00:50:02.679 --> 00:50:04.800
<v Speaker 3>to get them from everywhere and everything. And that's why

826
00:50:05.480 --> 00:50:08.719
<v Speaker 3>really delving into the corners of an industry yields a

827
00:50:08.719 --> 00:50:11.519
<v Speaker 3>lot of games, because there might be a new model

828
00:50:11.800 --> 00:50:15.920
<v Speaker 3>or a new approach that is very very interesting that

829
00:50:16.039 --> 00:50:17.320
<v Speaker 3>might be part of the future.

830
00:50:17.360 --> 00:50:17.559
<v Speaker 1>You know.

831
00:50:17.719 --> 00:50:23.280
<v Speaker 3>MRP was announced with no fanfare, got a huge referendum correct,

832
00:50:23.800 --> 00:50:25.960
<v Speaker 3>and basically no one paid attention for about a year

833
00:50:26.559 --> 00:50:30.679
<v Speaker 3>until they had been validated again. So from my perspective,

834
00:50:30.960 --> 00:50:32.199
<v Speaker 3>it is very much.

835
00:50:32.159 --> 00:50:35.400
<v Speaker 2>Validated against the grain because I was following twenty seventeen

836
00:50:35.480 --> 00:50:38.800
<v Speaker 2>and pretty much you had Yugov saying, hey, there's a

837
00:50:38.840 --> 00:50:41.360
<v Speaker 2>really good chance. May it's not going to win her majority,

838
00:50:41.440 --> 00:50:44.079
<v Speaker 2>even if she wins the popular vote by X number

839
00:50:44.119 --> 00:50:47.280
<v Speaker 2>of points, because of these demographics that are moving, that

840
00:50:47.360 --> 00:50:49.719
<v Speaker 2>have effect in your six hundred and thirty two non

841
00:50:49.760 --> 00:50:52.280
<v Speaker 2>northern Irish seats, And kind of like the reaction up

842
00:50:52.360 --> 00:50:54.440
<v Speaker 2>until election day, as far as I could tell from

843
00:50:54.760 --> 00:50:58.320
<v Speaker 2>the British public, not necessarily the private calls on the

844
00:50:58.320 --> 00:51:02.400
<v Speaker 2>pollsters was yeah, yeah, yeah, okay, thanks, and then of

845
00:51:02.400 --> 00:51:05.199
<v Speaker 2>course you proved to be right inward. Yeah yeah yeah,

846
00:51:05.280 --> 00:51:07.960
<v Speaker 2>isn't this great? And then now everybody does MRPs it

847
00:51:08.000 --> 00:51:11.119
<v Speaker 2>seems before your election, some of them better than others.

848
00:51:11.159 --> 00:51:12.920
<v Speaker 3>Some of them better than others. I mean, what I

849
00:51:12.960 --> 00:51:16.639
<v Speaker 3>would say, Henry, I think in defense of the polling

850
00:51:17.679 --> 00:51:21.559
<v Speaker 3>industry is that a lot of the experts like yourself

851
00:51:21.639 --> 00:51:25.480
<v Speaker 3>who are making calls are doing so from an advantaged position,

852
00:51:25.559 --> 00:51:28.239
<v Speaker 3>because a polster can only project what their own projection is.

853
00:51:28.840 --> 00:51:33.320
<v Speaker 3>But you get to sit around taking ten firms worth

854
00:51:33.360 --> 00:51:36.519
<v Speaker 3>of polling data and then look at the qualitative information,

855
00:51:37.239 --> 00:51:40.159
<v Speaker 3>and then get on the ground and then have conversations

856
00:51:40.239 --> 00:51:42.719
<v Speaker 3>and then in the round be able to take a

857
00:51:42.840 --> 00:51:44.840
<v Speaker 3>very detailed look. So we've just had a series of

858
00:51:45.639 --> 00:51:48.599
<v Speaker 3>local elections here in the UK, and it was very

859
00:51:48.639 --> 00:51:51.079
<v Speaker 3>clear that the expert projections were better than the MLP.

860
00:51:51.280 --> 00:51:53.920
<v Speaker 3>But my question goes, what do we think the experts

861
00:51:53.920 --> 00:51:58.880
<v Speaker 3>were looking at, which was everyone's models inaggregate and working

862
00:51:58.880 --> 00:52:01.159
<v Speaker 3>out where they convert to where they differed, and then

863
00:52:01.280 --> 00:52:05.760
<v Speaker 3>using multiple methods and filling in the gaps and therefore

864
00:52:05.760 --> 00:52:07.679
<v Speaker 3>be able to give a bird's eye view. So my

865
00:52:07.880 --> 00:52:11.400
<v Speaker 3>view as ever is that this kind of analysis layer,

866
00:52:11.480 --> 00:52:14.480
<v Speaker 3>this whole ecosystem that sits on top of poling, is

867
00:52:14.519 --> 00:52:17.199
<v Speaker 3>only able to be as good as it is precisely

868
00:52:17.239 --> 00:52:19.360
<v Speaker 3>because it sits on the bed of all these people

869
00:52:19.559 --> 00:52:25.920
<v Speaker 3>making singular specific calls about elections, and so it kind

870
00:52:25.920 --> 00:52:28.280
<v Speaker 3>of works both ways in my opinion. That's why we

871
00:52:28.360 --> 00:52:32.760
<v Speaker 3>have people like yourself, because you get the privileged position

872
00:52:32.960 --> 00:52:34.840
<v Speaker 3>very often. You know the posters will be briefing you

873
00:52:34.880 --> 00:52:38.679
<v Speaker 3>in detail. You're the only person. No one else in

874
00:52:38.719 --> 00:52:41.920
<v Speaker 3>the US or the UK would be able to extract

875
00:52:41.960 --> 00:52:44.480
<v Speaker 3>all about information and have a bird's eye view of

876
00:52:44.519 --> 00:52:47.320
<v Speaker 3>everything because I don't know what another firm is doing

877
00:52:47.360 --> 00:52:49.480
<v Speaker 3>and what that firm is doing. But you're the lucky

878
00:52:49.559 --> 00:52:53.280
<v Speaker 3>person at least, and so your listeners to be able

879
00:52:53.320 --> 00:52:56.800
<v Speaker 3>to say, actually, well, Henry's seen this signal here, and

880
00:52:56.840 --> 00:52:59.360
<v Speaker 3>he's seen this signal here, and he's read this bit

881
00:52:59.400 --> 00:53:01.519
<v Speaker 3>of information here, and then he's visited here and now

882
00:53:01.559 --> 00:53:04.320
<v Speaker 3>here's my kalates. So there is this. You know, a

883
00:53:04.360 --> 00:53:08.400
<v Speaker 3>lot of the costs are born by the industry for

884
00:53:08.480 --> 00:53:11.599
<v Speaker 3>others projections, and I think you know, you're very good

885
00:53:11.599 --> 00:53:14.360
<v Speaker 3>at kind of acknowledging that. But I think that's really

886
00:53:14.360 --> 00:53:15.800
<v Speaker 3>really important going forward.

887
00:53:18.199 --> 00:53:20.960
<v Speaker 2>Well, let me ask you a penultimate question then, and

888
00:53:21.039 --> 00:53:22.840
<v Speaker 2>that is what work are you going to be doing

889
00:53:22.840 --> 00:53:24.840
<v Speaker 2>in the United States? I know you've just come over here.

890
00:53:24.880 --> 00:53:28.280
<v Speaker 2>You mentioned the Texas Paul. Obviously I was privileged enough

891
00:53:28.320 --> 00:53:33.599
<v Speaker 2>to release your mrp of the congressional races. What other

892
00:53:33.639 --> 00:53:35.679
<v Speaker 2>work are you doing here or are looking at doing

893
00:53:35.719 --> 00:53:37.079
<v Speaker 2>in the United States during the interim.

894
00:53:37.079 --> 00:53:40.519
<v Speaker 3>Second, so, we have a media partnership, the Financial Times,

895
00:53:41.280 --> 00:53:45.760
<v Speaker 3>and we're basically partnering with them to provide not just

896
00:53:45.840 --> 00:53:48.800
<v Speaker 3>horse race polling, because that is not super interesting. You

897
00:53:48.800 --> 00:53:51.199
<v Speaker 3>can go on polling aggregators and look at it, but

898
00:53:51.519 --> 00:53:54.039
<v Speaker 3>slightly going in underneath the surface, you might say, beyond

899
00:53:54.079 --> 00:53:59.480
<v Speaker 3>the polls, providing information on what are the top issues?

900
00:53:59.760 --> 00:54:02.880
<v Speaker 3>Where have we moved, what they're basically getting in the

901
00:54:02.920 --> 00:54:07.679
<v Speaker 3>why before the what you know. And in a US context,

902
00:54:07.719 --> 00:54:11.800
<v Speaker 3>it's very clear to me that the current low presidential

903
00:54:11.840 --> 00:54:15.440
<v Speaker 3>approval ratings have very much driven off pole ratings to

904
00:54:15.480 --> 00:54:19.360
<v Speaker 3>do with the economy, because prior to the election, the

905
00:54:19.400 --> 00:54:22.880
<v Speaker 3>statistical modeling showed that that was one of the greatest

906
00:54:22.920 --> 00:54:28.599
<v Speaker 3>singular factors behind the president's victory, particularly in the swing states,

907
00:54:28.599 --> 00:54:30.079
<v Speaker 3>and that is also one of the issues on which

908
00:54:30.119 --> 00:54:33.079
<v Speaker 3>has moved the most away from the Republicans to the Democrats.

909
00:54:33.119 --> 00:54:35.880
<v Speaker 3>And it's looking into those kind of issues and trying

910
00:54:35.880 --> 00:54:39.559
<v Speaker 3>to really unpick the why before the what happens. So

911
00:54:39.599 --> 00:54:43.039
<v Speaker 3>that's why we're interested in looking at the Senate races.

912
00:54:43.079 --> 00:54:46.320
<v Speaker 3>And it's so there are so many, you know, was

913
00:54:46.360 --> 00:54:49.480
<v Speaker 3>it thirty five races, thirty three plus the two specials

914
00:54:50.159 --> 00:54:54.480
<v Speaker 3>that my count, we've got four Democrat defends that you

915
00:54:54.519 --> 00:54:57.840
<v Speaker 3>know in New Hampshire, Minnesota, Georgia, Michigan. Two of those

916
00:54:57.840 --> 00:55:02.920
<v Speaker 3>look hypermarginal, and then we've got six potential flips, all

917
00:55:03.039 --> 00:55:07.719
<v Speaker 3>of but one look hyper margin. You know, North Carolina

918
00:55:08.159 --> 00:55:11.599
<v Speaker 3>to me feels a bit more kind of Roy Cooper asked,

919
00:55:12.239 --> 00:55:15.480
<v Speaker 3>But at least Senate Poland we've been looking at in it. Maine.

920
00:55:15.840 --> 00:55:18.639
<v Speaker 3>Maine is a very hard state to poll. Whenever anyone

921
00:55:18.679 --> 00:55:21.280
<v Speaker 3>asked me what are the hardest states to poll? I

922
00:55:21.280 --> 00:55:23.960
<v Speaker 3>will always say, other than all of the swing states,

923
00:55:25.320 --> 00:55:31.480
<v Speaker 3>Maine and Alaska. And you know, it's unfortunate that those

924
00:55:31.519 --> 00:55:33.400
<v Speaker 3>two now are to the sixth.

925
00:55:33.679 --> 00:55:34.880
<v Speaker 1>May I just jump in here.

926
00:55:35.639 --> 00:55:37.880
<v Speaker 2>That is something that has been observed in the United

927
00:55:37.920 --> 00:55:41.000
<v Speaker 2>States by American pulsters that in twenty twenty, literally every

928
00:55:41.039 --> 00:55:43.840
<v Speaker 2>single public poll said that Susan Collins who's going to lose,

929
00:55:43.880 --> 00:55:45.679
<v Speaker 2>and it wasn't going to be close, and she.

930
00:55:45.760 --> 00:55:47.119
<v Speaker 3>Won seventeen points.

931
00:55:47.119 --> 00:55:48.679
<v Speaker 1>They were seventeen points.

932
00:55:50.039 --> 00:55:52.039
<v Speaker 2>I happen to be talking with the senator friend of

933
00:55:52.079 --> 00:55:54.800
<v Speaker 2>mine beforehand, and he told me, Susan says, it's fine.

934
00:55:55.159 --> 00:55:58.320
<v Speaker 1>So Susan's polster was getting it right. She was lined

935
00:55:58.320 --> 00:55:58.960
<v Speaker 1>to her colleagues.

936
00:56:00.400 --> 00:56:04.039
<v Speaker 2>Is another one that doesn't have quite that ignomious example

937
00:56:04.079 --> 00:56:06.760
<v Speaker 2>of the twenty twenty error, but it too is known

938
00:56:06.880 --> 00:56:11.719
<v Speaker 2>as a very difficult place to pull. And if I

939
00:56:11.840 --> 00:56:15.199
<v Speaker 2>look at say, I think, what do the two states

940
00:56:15.239 --> 00:56:18.599
<v Speaker 2>have in common. They have a significant white and non

941
00:56:19.320 --> 00:56:22.239
<v Speaker 2>Alaska has a significant non white, non college but it

942
00:56:22.280 --> 00:56:27.519
<v Speaker 2>has a very significant rural rocky terrain, people living in

943
00:56:27.559 --> 00:56:32.280
<v Speaker 2>different places. Does that help contribute to a lack of

944
00:56:32.360 --> 00:56:35.400
<v Speaker 2>trust that if you're in a fifty person town and

945
00:56:36.000 --> 00:56:39.559
<v Speaker 2>you're connected to the world by you know, by a

946
00:56:39.679 --> 00:56:44.239
<v Speaker 2>road and intermittent internet service, that that or is that

947
00:56:44.519 --> 00:56:45.719
<v Speaker 2>just a spurious factor.

948
00:56:45.960 --> 00:56:49.239
<v Speaker 3>I also think the partisanship is a bit weaker in

949
00:56:49.280 --> 00:56:55.360
<v Speaker 3>those states like May and main In particular, the demographic

950
00:56:55.400 --> 00:56:59.679
<v Speaker 3>trends of who votes Republican and who votes Democrat is

951
00:56:59.679 --> 00:57:02.760
<v Speaker 3>actually very complex and very very different to the rest

952
00:57:02.800 --> 00:57:06.440
<v Speaker 3>of the US, and that's slightly hidden by the county well.

953
00:57:06.480 --> 00:57:10.480
<v Speaker 3>Actually the small level maps, you know, including the ones

954
00:57:10.480 --> 00:57:14.760
<v Speaker 3>where you're showing counties with nobody living in there, they

955
00:57:14.880 --> 00:57:17.519
<v Speaker 3>make it look like there's a clear geographic division, which

956
00:57:17.599 --> 00:57:20.239
<v Speaker 3>might make you think, actually the coalitions are quite similar

957
00:57:20.239 --> 00:57:21.679
<v Speaker 3>to the rest of the US. They're not at all,

958
00:57:22.760 --> 00:57:24.639
<v Speaker 3>which makes it really hard. If you're opposed, you like,

959
00:57:24.679 --> 00:57:27.880
<v Speaker 3>I'm going to get three hundred Democrats and three hundred Publicans,

960
00:57:27.880 --> 00:57:31.880
<v Speaker 3>and it's like, well, which Democrats and which Republicans. But

961
00:57:31.920 --> 00:57:35.239
<v Speaker 3>if they're very very complex, My instinct on Alaska is

962
00:57:35.280 --> 00:57:40.559
<v Speaker 3>that's also the same as well. I do think if

963
00:57:40.559 --> 00:57:43.599
<v Speaker 3>you think about it, one of the things I've noticed

964
00:57:43.639 --> 00:57:46.159
<v Speaker 3>as an outsider, because of course there are Americans with

965
00:57:46.199 --> 00:57:49.039
<v Speaker 3>deep expertise who actually do really get it right all

966
00:57:49.079 --> 00:57:52.280
<v Speaker 3>the time, is that many of the states that aren't

967
00:57:52.360 --> 00:57:54.800
<v Speaker 3>quite right on ones that where there is still some

968
00:57:54.920 --> 00:57:59.639
<v Speaker 3>degree of split ticket voting, and also where the ideological

969
00:57:59.679 --> 00:58:05.480
<v Speaker 3>part of zian Ship preferences are incredibly complicated, where the

970
00:58:06.159 --> 00:58:08.679
<v Speaker 3>singular definition is not quite there. So I was very interested,

971
00:58:08.719 --> 00:58:12.079
<v Speaker 3>for example, that polsters didn't really get Pennsylvania that wrong

972
00:58:13.039 --> 00:58:16.360
<v Speaker 3>at the election versus you know, there was still a

973
00:58:16.400 --> 00:58:20.679
<v Speaker 3>systemic bias towards Harris that shouldn't have been there. But

974
00:58:20.800 --> 00:58:23.320
<v Speaker 3>for me that as a polster, you're looking through that

975
00:58:23.400 --> 00:58:26.719
<v Speaker 3>data and you're thinking, well, you know, Pennsylvania feels like

976
00:58:26.760 --> 00:58:29.079
<v Speaker 3>a state where as you said, I think you said

977
00:58:29.119 --> 00:58:33.760
<v Speaker 3>it three regions melded into one, right, where you can

978
00:58:33.840 --> 00:58:37.079
<v Speaker 3>have a stronger handle on Well, that's a Republican group

979
00:58:37.079 --> 00:58:38.960
<v Speaker 3>of people, and they are very much Republican. And that's

980
00:58:38.960 --> 00:58:40.760
<v Speaker 3>a Democrat people and they are very much Democrat. And

981
00:58:40.840 --> 00:58:42.719
<v Speaker 3>that's a group of swing voters, and they are very

982
00:58:42.800 --> 00:58:45.599
<v Speaker 3>much a swing And that maybe makes it slightly easier

983
00:58:45.599 --> 00:58:49.519
<v Speaker 3>to poll than some of these other states which are

984
00:58:49.519 --> 00:58:52.960
<v Speaker 3>a bit more nonconformist, a bit more complex and a

985
00:58:52.960 --> 00:58:55.280
<v Speaker 3>bit hard. And I think maybe Iowa is one of

986
00:58:55.280 --> 00:58:58.920
<v Speaker 3>those states that also falls into that. I guess if

987
00:58:58.920 --> 00:59:00.119
<v Speaker 3>you look at the polling.

988
00:59:00.280 --> 00:59:03.480
<v Speaker 2>Is another one that had a significant twenty twenty polling

989
00:59:04.639 --> 00:59:07.639
<v Speaker 2>directional or margin error, you know, which is that there

990
00:59:07.679 --> 00:59:09.880
<v Speaker 2>was a lot of polls suggesting Joany Ernst might be

991
00:59:09.960 --> 00:59:13.519
<v Speaker 2>in trouble. In the twenty eighteen there were poles suggesting

992
00:59:13.519 --> 00:59:16.159
<v Speaker 2>Democrats might win, and of course Republicans won and both

993
00:59:16.159 --> 00:59:17.679
<v Speaker 2>those races a very comfortable.

994
00:59:17.400 --> 00:59:20.400
<v Speaker 3>Very important. It's a state that Republicans win, can win

995
00:59:20.440 --> 00:59:22.880
<v Speaker 3>by double digits, sometimes by single digits, but then it

996
00:59:22.920 --> 00:59:26.719
<v Speaker 3>does have a predilection to various times at the gubernatorial

997
00:59:26.760 --> 00:59:28.119
<v Speaker 3>level go the other way. If you look at the

998
00:59:28.159 --> 00:59:30.679
<v Speaker 3>kind of senate Senate polling, which I know it's still

999
00:59:30.760 --> 00:59:32.880
<v Speaker 3>Republicans are still favored to take Iowa. We look at

1000
00:59:32.880 --> 00:59:35.960
<v Speaker 3>the kind of governorship polling, I think that's rob Sam

1001
00:59:36.039 --> 00:59:39.199
<v Speaker 3>so standing there. Again, it's complex, you know. It's many

1002
00:59:39.199 --> 00:59:41.199
<v Speaker 3>of these states that we think of as not complex

1003
00:59:41.280 --> 00:59:43.960
<v Speaker 3>that are in fact much more complex than everyone realizes.

1004
00:59:45.880 --> 00:59:50.760
<v Speaker 3>And I think also the degree to which Utah went

1005
00:59:50.840 --> 00:59:54.840
<v Speaker 3>from megaskeptic to not It was also an underrated phenomena

1006
00:59:54.960 --> 00:59:57.960
<v Speaker 3>because how much polling expertise is there of the Mormon

1007
00:59:58.039 --> 01:00:06.000
<v Speaker 3>community and the shifting sands around there. So, I, you know, polsters,

1008
01:00:06.239 --> 01:00:09.480
<v Speaker 3>I guess to round it off, our people who love

1009
01:00:10.440 --> 01:00:12.760
<v Speaker 3>typically they're quite a patriotic lot. You don't want to

1010
01:00:12.760 --> 01:00:16.199
<v Speaker 3>stare at your own country and analyze it to death

1011
01:00:16.280 --> 01:00:20.079
<v Speaker 3>for multiple decades if you really care and you're interested

1012
01:00:20.079 --> 01:00:23.880
<v Speaker 3>in it. But also you're you're basically a deployed anthropologist,

1013
01:00:23.960 --> 01:00:28.960
<v Speaker 3>right You're looking for the stories, the data, the modeling

1014
01:00:29.000 --> 01:00:31.239
<v Speaker 3>to line up. Very often it doesn't, and you have

1015
01:00:31.280 --> 01:00:34.880
<v Speaker 3>to make a judgment call. But for me, you know,

1016
01:00:35.119 --> 01:00:38.079
<v Speaker 3>that's kind of what we'll be looking at, Henry. We're

1017
01:00:38.119 --> 01:00:40.320
<v Speaker 3>interested in the why, but we're also interested in the

1018
01:00:40.320 --> 01:00:45.239
<v Speaker 3>what because of how wide the landscape is in terms

1019
01:00:45.239 --> 01:00:47.679
<v Speaker 3>of both the Senate. You know, I would say the

1020
01:00:47.679 --> 01:00:51.119
<v Speaker 3>House of Representative has far fewer marginal seats.

1021
01:00:51.519 --> 01:00:53.719
<v Speaker 1>Thank you, Jerry, Yeah, thank you.

1022
01:00:53.800 --> 01:00:55.880
<v Speaker 3>But but I think you know, we were speaking about

1023
01:00:55.920 --> 01:00:59.480
<v Speaker 3>hungry and you very carefully handled our poll that we

1024
01:00:59.519 --> 01:01:03.840
<v Speaker 3>handed to suggesting the kind of strong victory for Magyar.

1025
01:01:04.800 --> 01:01:07.559
<v Speaker 3>The funny thing about redistricting and jerry mandering is it

1026
01:01:07.599 --> 01:01:14.000
<v Speaker 3>works until it doesn't, and it works within certain bounds,

1027
01:01:14.280 --> 01:01:18.440
<v Speaker 3>and the bounds that it works in are ones of

1028
01:01:18.840 --> 01:01:25.599
<v Speaker 3>basically temperate swing where maybe Republicans R plus seven, D

1029
01:01:25.679 --> 01:01:28.760
<v Speaker 3>plus seven, whatever it is, whatever the electric looks like

1030
01:01:28.800 --> 01:01:32.119
<v Speaker 3>in any given year. But where it really breaks down

1031
01:01:32.480 --> 01:01:35.920
<v Speaker 3>is when you have an election outcome that where one

1032
01:01:35.960 --> 01:01:38.960
<v Speaker 3>party is leading the other by double digits, because then

1033
01:01:39.000 --> 01:01:43.760
<v Speaker 3>you've created a cardret of districts or seats that would

1034
01:01:43.760 --> 01:01:46.800
<v Speaker 3>never have been marginal normally, but now suddenly they all are.

1035
01:01:47.760 --> 01:01:53.079
<v Speaker 3>So it can actually exacerbate a landslide the wrong way,

1036
01:01:53.159 --> 01:01:55.840
<v Speaker 3>depending on who's doing it in a special environment.

1037
01:01:56.239 --> 01:02:00.440
<v Speaker 2>One of the reasons why Tissa Magyor's party has a

1038
01:02:00.599 --> 01:02:05.920
<v Speaker 2>constitutional super majority is because they won the native Hungarian

1039
01:02:06.000 --> 01:02:10.239
<v Speaker 2>vote by fifteen to seventeen points, and that hand and

1040
01:02:10.320 --> 01:02:13.079
<v Speaker 2>effect across a wide variety of seats that had they

1041
01:02:13.079 --> 01:02:15.400
<v Speaker 2>won by only twelve points, so they would not have picked.

1042
01:02:15.199 --> 01:02:19.760
<v Speaker 3>Up's And that's the thing. So I'm interested to see

1043
01:02:20.039 --> 01:02:21.880
<v Speaker 3>that there's a story that's going to emerge at some

1044
01:02:21.920 --> 01:02:24.800
<v Speaker 3>point in the next two decades where that kind of

1045
01:02:24.880 --> 01:02:28.800
<v Speaker 3>pack of cards all comes down. But again, that a

1046
01:02:28.800 --> 01:02:31.559
<v Speaker 3>big difference. You know, the UK, we have the Boundary Commission.

1047
01:02:31.639 --> 01:02:35.079
<v Speaker 3>This is all done in an almost census like fashion,

1048
01:02:35.400 --> 01:02:40.159
<v Speaker 3>in an incredibly equal way basically, so again another difference

1049
01:02:40.199 --> 01:02:42.760
<v Speaker 3>to get a hold of. But every country is different.

1050
01:02:43.360 --> 01:02:45.119
<v Speaker 3>You know, we are a UK based firm, but some

1051
01:02:45.159 --> 01:02:48.679
<v Speaker 3>of these techniques are very applicable, whether that's France, whether

1052
01:02:48.679 --> 01:02:52.440
<v Speaker 3>that's Germany. But the key thing is to be very

1053
01:02:52.440 --> 01:02:55.679
<v Speaker 3>clear about what your projection is and why, and explain

1054
01:02:55.760 --> 01:02:58.119
<v Speaker 3>what you've done so that if it's if it's wrong,

1055
01:02:58.360 --> 01:03:00.800
<v Speaker 3>you have some sense of whites, and if it's right,

1056
01:03:01.320 --> 01:03:03.840
<v Speaker 3>instead of saying we're right, you're able to point to

1057
01:03:03.960 --> 01:03:06.280
<v Speaker 3>why you might have been right so that others can learn.

1058
01:03:08.599 --> 01:03:13.159
<v Speaker 2>One last question, looking at America in twenty twenty four,

1059
01:03:13.920 --> 01:03:16.320
<v Speaker 2>you talked about the why you know, which is you know,

1060
01:03:16.519 --> 01:03:20.400
<v Speaker 2>issue salience, issue viewpoints and other things, and the what,

1061
01:03:20.840 --> 01:03:24.400
<v Speaker 2>the partisan outcome or the job approval. Do you think

1062
01:03:24.440 --> 01:03:27.800
<v Speaker 2>you will see movement in the why before movement in

1063
01:03:27.880 --> 01:03:30.400
<v Speaker 2>the what? In other words, that let's say you take

1064
01:03:30.440 --> 01:03:34.239
<v Speaker 2>a poll in June and you start to see movements

1065
01:03:34.280 --> 01:03:37.599
<v Speaker 2>in saliens among issues, or you start to see, hey,

1066
01:03:37.679 --> 01:03:40.840
<v Speaker 2>I think the economy is getting better, and do you

1067
01:03:40.840 --> 01:03:44.639
<v Speaker 2>think that will predict a subsequent movement in like the

1068
01:03:44.719 --> 01:03:48.559
<v Speaker 2>job approval or the congressional generic ballot or is just

1069
01:03:49.039 --> 01:03:51.559
<v Speaker 2>another signal that you wait with all the others. In

1070
01:03:51.599 --> 01:04:01.760
<v Speaker 2>this qualitative art form that is modern polling slash political analysis.

1071
01:04:02.639 --> 01:04:06.840
<v Speaker 3>The why is very often it's correct a lot more

1072
01:04:06.840 --> 01:04:13.199
<v Speaker 3>times than the what. So the explanation of an event

1073
01:04:13.239 --> 01:04:16.000
<v Speaker 3>can often the polling that feeds into that can is

1074
01:04:16.039 --> 01:04:19.880
<v Speaker 3>often much more evergreen and robust than these specifics around

1075
01:04:19.880 --> 01:04:22.480
<v Speaker 3>the what, because of what can only it only has

1076
01:04:22.480 --> 01:04:26.440
<v Speaker 3>to be wrong by three points to be out, whereas

1077
01:04:26.519 --> 01:04:30.480
<v Speaker 3>the why is directional, and very often the why comes

1078
01:04:30.519 --> 01:04:33.519
<v Speaker 3>screaming out from the data, like the cost of living

1079
01:04:33.639 --> 01:04:36.840
<v Speaker 3>is hugely important in the Southern States, immigration control is

1080
01:04:36.880 --> 01:04:40.159
<v Speaker 3>incredibly important. Right, These are truths that have a massive

1081
01:04:40.280 --> 01:04:43.480
<v Speaker 3>order of magnitude that won't be wrong even with the

1082
01:04:43.519 --> 01:04:48.159
<v Speaker 3>polling error. But the shift in the why can be

1083
01:04:48.360 --> 01:04:52.079
<v Speaker 3>very predictive of outcomes. So, you know, should those numbers

1084
01:04:52.119 --> 01:04:56.199
<v Speaker 3>improve on the perception of the economy, you know, it's

1085
01:04:56.199 --> 01:04:59.920
<v Speaker 3>not just perception. Gas prices have a very strong corarel

1086
01:05:00.239 --> 01:05:04.519
<v Speaker 3>relationship to swing against the Republicans, So that's real data showing,

1087
01:05:04.639 --> 01:05:06.960
<v Speaker 3>like having real political consequences.

1088
01:05:07.440 --> 01:05:10.679
<v Speaker 2>So they did with Biden in the summer of twenty

1089
01:05:10.719 --> 01:05:15.519
<v Speaker 2>twenty two. Hence why he started getting licenses in Venezuela

1090
01:05:15.559 --> 01:05:18.800
<v Speaker 2>to ship and changing his stands on Saudi human rights

1091
01:05:18.800 --> 01:05:22.199
<v Speaker 2>to get more production and releasing this because he could

1092
01:05:22.280 --> 01:05:24.440
<v Speaker 2>see the same thing happening to him as guests broke

1093
01:05:24.480 --> 01:05:26.360
<v Speaker 2>five dollars a gallon in most of the country.

1094
01:05:26.719 --> 01:05:29.800
<v Speaker 3>Yeah, and I think so that's what we're tracking with

1095
01:05:29.840 --> 01:05:33.159
<v Speaker 3>the ft. That's what we're tracking here at Focal Data

1096
01:05:33.360 --> 01:05:35.960
<v Speaker 3>is are the explanations of what's happening makes sense? You know,

1097
01:05:36.000 --> 01:05:39.679
<v Speaker 3>We've seen the salience, for example, of the handling over

1098
01:05:39.840 --> 01:05:44.519
<v Speaker 3>the Iran situation become important out of nowhere. That's now

1099
01:05:44.559 --> 01:05:47.840
<v Speaker 3>a thing. We're now seeing that in fact, independence sit

1100
01:05:48.000 --> 01:05:52.039
<v Speaker 3>much closer to Democrat voters on that issue than versus

1101
01:05:52.079 --> 01:05:54.960
<v Speaker 3>a kind of the base of the Republican Party. To me,

1102
01:05:55.079 --> 01:05:57.880
<v Speaker 3>that's really important because I'm going to track the salience

1103
01:05:57.920 --> 01:05:59.719
<v Speaker 3>of that issue as much as we are. Who's doing

1104
01:05:59.719 --> 01:06:00.559
<v Speaker 3>well badly on it?

1105
01:06:02.159 --> 01:06:07.199
<v Speaker 2>Well, James, it's always fascinating chatting with you. I whish

1106
01:06:07.239 --> 01:06:10.079
<v Speaker 2>I could jump across the pond and continue this conversation,

1107
01:06:10.159 --> 01:06:12.360
<v Speaker 2>but all good things must come to an end, and

1108
01:06:12.440 --> 01:06:14.960
<v Speaker 2>so mucht are on air discussion.

1109
01:06:15.320 --> 01:06:17.440
<v Speaker 1>Where can my listeners follow your work?

1110
01:06:18.920 --> 01:06:23.960
<v Speaker 3>Three places. The first is focal Data dot com focl

1111
01:06:24.119 --> 01:06:27.119
<v Speaker 3>data dot com, where we have a blog with all

1112
01:06:27.159 --> 01:06:30.239
<v Speaker 3>of our work from not just myself but a team

1113
01:06:30.280 --> 01:06:32.840
<v Speaker 3>of twenty five researchers, some of which is very technical,

1114
01:06:32.840 --> 01:06:35.719
<v Speaker 3>some of which is not. The second is my substack,

1115
01:06:35.760 --> 01:06:39.199
<v Speaker 3>which is the Political Whiteboard, which is where the super

1116
01:06:39.199 --> 01:06:42.119
<v Speaker 3>super early stuff goes on big three four thousand word

1117
01:06:42.239 --> 01:06:46.599
<v Speaker 3>essays with modeling. And then I write occasionally four times

1118
01:06:46.639 --> 01:06:49.920
<v Speaker 3>a year for The London Times, where I set out

1119
01:06:50.039 --> 01:06:52.199
<v Speaker 3>at a much more high level some of my ideas

1120
01:06:52.239 --> 01:06:54.920
<v Speaker 3>that may or may not be interesting to your readers.

1121
01:06:55.159 --> 01:06:58.599
<v Speaker 3>The latest Times column London Times column is on the

1122
01:06:58.599 --> 01:07:04.400
<v Speaker 3>Democratic and its potential shift to the left of its base.

1123
01:07:05.719 --> 01:07:09.239
<v Speaker 2>I can recommend all of those personally, but for anyone

1124
01:07:09.239 --> 01:07:11.880
<v Speaker 2>who wants to get scared about the future of AI,

1125
01:07:12.159 --> 01:07:16.039
<v Speaker 2>I suggest you jump on Political Whiteboard and read what

1126
01:07:16.159 --> 01:07:18.360
<v Speaker 2>James had to say about that a few months ago.

1127
01:07:18.480 --> 01:07:22.519
<v Speaker 2>It will open your eyes. I'm not talking about dystopian

1128
01:07:22.880 --> 01:07:28.440
<v Speaker 2>terminator stype circumstances, but the political impacts that may be

1129
01:07:28.559 --> 01:07:31.920
<v Speaker 2>happening to the supposedly secure managerial class.

1130
01:07:32.599 --> 01:07:35.559
<v Speaker 1>Well read it and maybe weep.

1131
01:07:35.920 --> 01:07:37.960
<v Speaker 2>James, thank you and I look forward to having you

1132
01:07:38.000 --> 01:07:39.039
<v Speaker 2>back on Beyond the Polls.

1133
01:07:39.360 --> 01:07:43.559
<v Speaker 1>Thank you, Henry, pleasure to be here. That's it for

1134
01:07:43.599 --> 01:07:44.039
<v Speaker 1>this week.

1135
01:07:44.320 --> 01:07:47.519
<v Speaker 2>Next week I'll talk about redistricting, past, present, and future

1136
01:07:47.679 --> 01:07:51.159
<v Speaker 2>with Sean Trendy. Until then, let's reach for the stars

1137
01:07:51.159 --> 01:07:54.880
<v Speaker 2>together as we journey Beyond the Polls.