WEBVTT 00:00:00.000 --> 00:00:02.480 Welcome to the Deep Dive. We are so glad you 00:00:02.480 --> 00:00:06.040 can join us today because we are taking a really 00:00:06.040 --> 00:00:09.480 fascinating statistical journey. through just 00:00:09.480 --> 00:00:12.839 a single document. Yeah, literally one single 00:00:12.839 --> 00:00:15.400 Wikipedia list. Right. It is a comprehensive 00:00:15.400 --> 00:00:18.239 Wikipedia list of United States Senate elections 00:00:18.239 --> 00:00:22.339 in Alabama. And today we are going to look past 00:00:22.339 --> 00:00:25.760 the dry tables and uncover the dramatic story 00:00:25.760 --> 00:00:29.320 of how voting pools expanded, you know, how political 00:00:29.320 --> 00:00:32.289 monopolies shifted. And we'll look at the statistical 00:00:32.289 --> 00:00:35.490 anomalies hidden in over a century of state election 00:00:35.490 --> 00:00:37.909 data. It is a wild ride when you actually look 00:00:37.909 --> 00:00:40.490 at the math. It really is. But before we get 00:00:40.490 --> 00:00:42.289 into it, because our source material spans over 00:00:42.289 --> 00:00:44.810 a century of political data covering both Democratic 00:00:44.810 --> 00:00:46.829 and Republican victories, we want to be absolutely 00:00:46.829 --> 00:00:49.369 clear with you. We're not taking any sides here. 00:00:49.429 --> 00:00:51.170 Right. Exactly. We're not endorsing any of the 00:00:51.170 --> 00:00:53.130 left wing or right wing viewpoints or candidates 00:00:53.130 --> 00:00:55.429 mentioned in these historical records. Our goal 00:00:55.429 --> 00:00:57.789 is strictly to impartially report on the factual 00:00:57.789 --> 00:01:00.380 data provided in the source material. And convey 00:01:00.380 --> 00:01:02.859 the statistical shifts as they occurred. Yeah, 00:01:02.979 --> 00:01:05.280 we are just looking at the numbers. And to lay 00:01:05.280 --> 00:01:07.200 the foundation for those numbers, we should explain 00:01:07.200 --> 00:01:09.319 how the Senate works based on the text here. 00:01:09.420 --> 00:01:12.719 Go for it. So each state gets two senators. They 00:01:12.719 --> 00:01:15.659 serve staggered six -year terms. And those seats 00:01:15.659 --> 00:01:17.900 are divided into three classes. Right, which 00:01:17.900 --> 00:01:20.849 creates a really unique data set. Exactly. When 00:01:20.849 --> 00:01:24.469 Alabama joined the union back in 1819, it was 00:01:24.469 --> 00:01:27.510 assigned class two and class three seats. So 00:01:27.510 --> 00:01:30.069 the data we are looking at tracks those two specific 00:01:30.069 --> 00:01:33.989 overlapping electoral cycles. But there is a 00:01:33.989 --> 00:01:35.629 massive systemic change we have to highlight 00:01:35.629 --> 00:01:38.650 right up front. The 17th Amendment. Yes. Before 00:01:38.650 --> 00:01:42.030 1913, senators were selected by state legislatures. 00:01:42.170 --> 00:01:44.810 You don't see direct voter turnout numbers in 00:01:44.810 --> 00:01:46.730 the early tables because the voters weren't directly 00:01:46.730 --> 00:01:49.290 making the choice. It was the 17th. Amendment 00:01:49.290 --> 00:01:51.670 that finally gave the power directly to the voters. 00:01:51.790 --> 00:01:53.870 That is the moment the tables switch from appointments 00:01:53.870 --> 00:01:57.650 to actual raw voter data. Exactly. Okay, let's 00:01:57.650 --> 00:01:59.750 unpack this. Because the numbers are crazy. They 00:01:59.750 --> 00:02:02.189 really are. When we look at the earliest data 00:02:02.189 --> 00:02:05.069 on the list, the margins are shocking. You know, 00:02:05.109 --> 00:02:07.209 you and I are used to modern elections where 00:02:07.209 --> 00:02:08.949 someone wins by a couple of percentage points, 00:02:09.009 --> 00:02:11.930 right? Yeah, 55 % win today is a huge landslide. 00:02:12.169 --> 00:02:15.210 But look at the data from 1918. We have the class 00:02:15.210 --> 00:02:18.840 two Democrat, John H. Bankhead. He won his election 00:02:18.840 --> 00:02:21.689 with literally 100 % of the vote. One hundred 00:02:21.689 --> 00:02:24.009 percent. One hundred percent. The table says 00:02:24.009 --> 00:02:26.810 he got exactly fifty four thousand eight hundred 00:02:26.810 --> 00:02:30.129 eight hundred eighty votes. Zero opposition on 00:02:30.129 --> 00:02:32.469 the board. Just a total shutout. And it wasn't 00:02:32.469 --> 00:02:34.449 a one time fluke either. If you scroll down the 00:02:34.449 --> 00:02:37.330 class two table to 1942, you see John H. Bankhead 00:02:37.330 --> 00:02:39.189 the second. The second. Right. Yeah. And he also 00:02:39.189 --> 00:02:41.409 won with one hundred percent. It is just an unbroken 00:02:41.409 --> 00:02:44.509 wall of single party dominance in that era. And 00:02:44.509 --> 00:02:46.229 you see it in class three too. Jay Lister Hill. 00:02:46.349 --> 00:02:48.469 Right. Jay Lister Hill won with one hundred percent 00:02:48.469 --> 00:02:51.050 in a 1938 special election. And then he does 00:02:51.050 --> 00:02:54.389 it again in the 1956 regular election. He pulled 00:02:54.389 --> 00:02:58.569 in 330 ,182 votes and again captured 100 percent 00:02:58.569 --> 00:03:00.949 of the electorate. Which is just staggering to 00:03:00.949 --> 00:03:03.830 think about. What's fascinating here is the absolute 00:03:03.830 --> 00:03:06.389 dominance of the Democratic Party during this 00:03:06.389 --> 00:03:09.530 era. Republicans were barely making a dent in 00:03:09.530 --> 00:03:12.770 the data. For instance, in 1936, you have Republican 00:03:12.770 --> 00:03:15.050 H .E. Burke's stressor running against Bankhead 00:03:15.050 --> 00:03:19.569 II. And what did he get? Only 12 .24%. Wow. Yeah, 00:03:19.669 --> 00:03:22.090 that was the entire footprint of the opposition. 00:03:22.530 --> 00:03:24.229 You know, when we talk about these percentages, 00:03:24.490 --> 00:03:26.889 I want to direct your attention to the sheer 00:03:26.889 --> 00:03:30.509 volume of the voting pool. The raw numbers are 00:03:30.509 --> 00:03:33.310 just as important as the percentages. The denominator. 00:03:33.650 --> 00:03:37.939 Right, the denominator. In 1918, a 100 % victory 00:03:37.939 --> 00:03:41.979 meant capturing just 54 ,880 votes. Which is 00:03:41.979 --> 00:03:44.360 tiny. It's incredibly tiny. Think about that 00:03:44.360 --> 00:03:46.400 from a modern perspective. Fifty four thousand 00:03:46.400 --> 00:03:48.879 people is essentially the capacity of a mid -sized 00:03:48.879 --> 00:03:51.199 college football stadium. Yeah. You just had 00:03:51.199 --> 00:03:53.300 to convince one stadium's worth of voters and 00:03:53.300 --> 00:03:55.780 you secured a 100 percent monopoly on the United 00:03:55.780 --> 00:03:58.620 States Senate seat. The voting base was so extremely 00:03:58.620 --> 00:04:00.819 small compared to the state's population today. 00:04:00.960 --> 00:04:03.139 It makes every single individual vote carry so 00:04:03.139 --> 00:04:05.960 much mathematical weight. It does. But here's 00:04:05.960 --> 00:04:08.370 where it gets really interesting. Because even 00:04:08.370 --> 00:04:12.030 in this era of small voting pools and 100 % sweeps, 00:04:12.030 --> 00:04:14.430 there are these totally unexpected statistical 00:04:14.430 --> 00:04:17.850 anomalies. Oh, the third party intruders? Yes. 00:04:18.350 --> 00:04:20.089 They just pop up in the tables and completely 00:04:20.089 --> 00:04:22.870 disrupt the map. The 1930 class two race is the 00:04:22.870 --> 00:04:25.329 perfect example of this. You track James Thomas 00:04:25.329 --> 00:04:28.089 Heflin in the tables. The data shows he won the 00:04:28.089 --> 00:04:30.649 seat as a Democrat in 1920. He wins it again 00:04:30.649 --> 00:04:33.610 as a Democrat in 1924. He is part of that dominant 00:04:33.610 --> 00:04:36.269 monopoly. Exactly. But then you look at the 1930 00:04:36.269 --> 00:04:38.970 data and Heflin suddenly appears as an inattentive 00:04:38.970 --> 00:04:40.939 runner up. And he doesn't just get a handful 00:04:40.939 --> 00:04:44.420 of votes. No, he pulls in a massive 40 .07 % 00:04:44.420 --> 00:04:49.339 of the vote. That is 100 ,952 votes. He ran against 00:04:49.339 --> 00:04:51.980 Democrat John H. Bankhead II, and an independent 00:04:51.980 --> 00:04:54.480 just casually walks away with 40 % of the electorate. 00:04:54.519 --> 00:04:56.779 A huge fracture in that 100 % wall. Totally. 00:04:56.839 --> 00:04:59.339 And you see it again in the 1950 Class III race. 00:04:59.519 --> 00:05:02.019 You have independent John G. Cromlin. He captured 00:05:02.019 --> 00:05:05.279 23 .46 % of the vote against Jay Lister Hill. 00:05:05.439 --> 00:05:07.759 So even when one party is sweeping the board, 00:05:08.160 --> 00:05:10.180 There are these massive chunks of the voter base 00:05:10.180 --> 00:05:13.300 willing to cast a protest vote. Right. It proves 00:05:13.300 --> 00:05:16.259 the electorate was never a pure monolith. But 00:05:16.259 --> 00:05:19.279 my absolute favorite anomaly in this data set, 00:05:19.319 --> 00:05:21.759 and I know you love this one too, is the Prohibition 00:05:21.759 --> 00:05:25.040 Party. It is so bizarre to see them there. Right. 00:05:25.319 --> 00:05:27.500 You think of the Prohibition Party as a movement 00:05:27.500 --> 00:05:31.149 from the 1920s. But look at the data in the 1970s. 00:05:31.170 --> 00:05:33.009 They are still on the board. Very much on the 00:05:33.009 --> 00:05:37.089 board. In the 1974 Class 3 election, a candidate 00:05:37.089 --> 00:05:41.009 named Alvin Abercrombie took 4 .16 % of the vote 00:05:41.009 --> 00:05:44.310 running for the Prohibition Party. In 1974. Yes. 00:05:44.410 --> 00:05:47.209 And then four years later, in the 1978 Class 00:05:47.209 --> 00:05:50.449 2 race, Jerome B. Couch ran for the Prohibition 00:05:50.449 --> 00:05:54.610 Party and secured 6 .01 % against Howell Heflin. 00:05:54.779 --> 00:05:57.120 When you look closely at that raw data, you realize 00:05:57.120 --> 00:05:59.259 what that percentage actually means. It's a lot 00:05:59.259 --> 00:06:02.459 of people. It is. A 6 % poll in that 1978 race 00:06:02.459 --> 00:06:04.920 means tens of thousands of individual voters 00:06:04.920 --> 00:06:07.779 went to the polls and actively voted for a prohibitionist. 00:06:07.939 --> 00:06:10.720 You find entire movements still registering massive 00:06:10.720 --> 00:06:13.199 data points decades after you expect them to 00:06:13.199 --> 00:06:15.680 vanish entirely from the tables. They just linger 00:06:15.680 --> 00:06:18.139 as statistical ghosts. They really do. But as 00:06:18.139 --> 00:06:20.339 we keep scrolling down the Wikipedia list into 00:06:20.339 --> 00:06:24.060 the 1960s and 1980s, you start to notice a creeping... 00:06:24.110 --> 00:06:27.009 change in the core numbers, the era of the 100 00:06:27.009 --> 00:06:29.970 % shutout starts to dissolve. The great party 00:06:29.970 --> 00:06:32.910 shift. Exactly. The margins just start compressing. 00:06:32.930 --> 00:06:35.449 Look at the 1962 class three race. This is a 00:06:35.449 --> 00:06:37.730 huge tipping point in the data. It really is. 00:06:37.810 --> 00:06:40.410 You have Jay Lister Hill, the same guy who won 00:06:40.410 --> 00:06:44.769 with 100 % in 1956. But in 1962, Republican James 00:06:44.769 --> 00:06:47.420 D. Martin nearly unseats him. The gap practically 00:06:47.420 --> 00:06:50.959 vanishes. Martin captures 49 .14 % of the vote. 00:06:51.060 --> 00:06:55.180 That's 195 ,134 votes. Hill barely holds on with 00:06:55.180 --> 00:06:59.240 50 .86%. Out of nearly 400 ,000 votes cast, they 00:06:59.240 --> 00:07:02.040 are separated by roughly 6 ,800 votes. Going 00:07:02.040 --> 00:07:04.680 from 100 % sweet to a 1 .something percent margin 00:07:04.680 --> 00:07:07.519 in just a few cycles is incredibly rapid statistical 00:07:07.519 --> 00:07:09.579 decay. And the pressure keeps building until 00:07:09.579 --> 00:07:12.139 the dam finally breaks in 1980. Right, the 1980 00:07:12.139 --> 00:07:14.779 Class III election. Republican Jeremiah Denton 00:07:14.779 --> 00:07:18.189 wins the seat. gets 50 .15 % of the vote against 00:07:18.189 --> 00:07:20.470 Democrat Jim Folsom Jr. If we connect this to 00:07:20.470 --> 00:07:22.990 the bigger picture, Denton's win is the catalyst 00:07:22.990 --> 00:07:25.410 for one of the most fascinating electoral sequences 00:07:25.410 --> 00:07:27.639 we have in this entire document. The Richard 00:07:27.639 --> 00:07:29.439 Shelby phenomenon. The Richard Shelby phenomenon. 00:07:29.720 --> 00:07:32.339 If you want to see how fluid party labels can 00:07:32.339 --> 00:07:35.459 be while maintaining absolute statistical stability, 00:07:35.800 --> 00:07:39.439 you just trace Shelby's data points. Let's do 00:07:39.439 --> 00:07:41.139 it. Let's trace it chronologically. Okay, so 00:07:41.139 --> 00:07:44.639 start in 1986. Richard Shelby runs as a Democrat 00:07:44.639 --> 00:07:47.019 for that class three seat against the Republican 00:07:47.019 --> 00:07:49.579 incumbent, Jeremiah Denton. And Shelby wins, 00:07:49.759 --> 00:07:52.740 but it is incredibly close. Razor thin. Shelby 00:07:52.740 --> 00:07:58.000 beats Denton 50 .28 % to 49 .7%. That is a difference 00:07:58.000 --> 00:08:01.060 of less than 7 ,000 votes. Very fragile margin. 00:08:01.399 --> 00:08:03.779 But then look at his next row in the table. In 00:08:03.779 --> 00:08:06.480 1992, Shelby runs again as a Democrat and his 00:08:06.480 --> 00:08:10.720 margin explodes. He wins with 64 .82%. So the 00:08:10.720 --> 00:08:13.100 data is telling us, OK, the Democrats have solidly 00:08:13.100 --> 00:08:15.000 reclaimed this class three seat. Right. The math 00:08:15.000 --> 00:08:16.839 looks totally settled. But then you drop down 00:08:16.839 --> 00:08:18.740 to the very next election cycle for that seat 00:08:18.740 --> 00:08:22.399 in 1998. The data shows a wild twist. A complete 00:08:22.399 --> 00:08:24.550 anomaly. Richard Shelby is listed as winning 00:08:24.550 --> 00:08:27.110 the exact same seat, retaining almost the exact 00:08:27.110 --> 00:08:29.750 same percentage of the voter pool, but he is 00:08:29.750 --> 00:08:33.350 listed as a Republican. He wins with 63 .24 % 00:08:33.350 --> 00:08:36.409 of the vote. As a Republican, the primary variable, 00:08:36.629 --> 00:08:39.409 the party label, completely flips, but his share 00:08:39.409 --> 00:08:41.610 of the electorate barely moves. And once that 00:08:41.610 --> 00:08:44.210 label flips, the table shows him... dominating 00:08:44.210 --> 00:08:47.009 that seat as a Republican for decades. He just 00:08:47.009 --> 00:08:50.190 locks it down. He wins in 2004, 2010 and 2016, 00:08:50.409 --> 00:08:53.289 consistently pulling in those mid to high 60 00:08:53.289 --> 00:08:56.769 percent margins. The label changed, but the math 00:08:56.769 --> 00:08:59.730 stayed relentlessly stable. That really ushers 00:08:59.730 --> 00:09:02.990 the data set into its modern era. We are looking 00:09:02.990 --> 00:09:05.850 at the 1990s through the 2020s now, and it becomes 00:09:05.850 --> 00:09:08.950 a period of. Deep Republican dominance. Definitely. 00:09:09.169 --> 00:09:11.029 Look at Jeff Sessions in the class two seat. 00:09:11.149 --> 00:09:13.690 He establishes absolute dominance. He wins in 00:09:13.690 --> 00:09:18.210 96, 2002 and 2008. By 2014, his hold on the numbers 00:09:18.210 --> 00:09:21.149 is so strong that he runs unopposed. Which takes 00:09:21.149 --> 00:09:22.950 us right back to the beginning of the list. Exactly. 00:09:23.149 --> 00:09:26.690 He wins that 2014 race with 97 .25 percent against 00:09:26.690 --> 00:09:29.230 just scattered write ins. It is the closest thing 00:09:29.230 --> 00:09:31.690 to those 1920s bank head margins we've seen in 00:09:31.690 --> 00:09:34.029 the modern era. The table projects this totally 00:09:34.029 --> 00:09:36.820 unshakable Republican stability. But then. You 00:09:36.820 --> 00:09:38.620 get these sudden jolts of volatility. The special 00:09:38.620 --> 00:09:42.279 elections. Yes. Look at the 2017 special election 00:09:42.279 --> 00:09:45.039 for that exact same class two seat. Out of nowhere, 00:09:45.139 --> 00:09:48.100 we see a razor thin Democratic victory. It completely 00:09:48.100 --> 00:09:50.799 disrupts the trend. Democrat Doug Jones edges 00:09:50.799 --> 00:09:53.590 out Republican Roy Moore. The distribution is 00:09:53.590 --> 00:09:58.389 incredibly tight. Jones gets 49 .97 % with 673 00:09:58.389 --> 00:10:05.870 ,896 votes. Moore gets 48 .34 % with 651 ,972 00:10:05.870 --> 00:10:08.570 votes. A difference of barely 20 ,000 votes out 00:10:08.570 --> 00:10:11.330 of over 1 .3 million cast. But the data also 00:10:11.330 --> 00:10:13.330 shows how incredibly brief that disruption was. 00:10:13.610 --> 00:10:16.629 By the 2020 regular election, the math aggressively 00:10:16.629 --> 00:10:19.049 snaps back to the established trend line. Republican 00:10:19.049 --> 00:10:21.830 Tommy Tuberville reclaims the seat. With a massive... 00:10:21.840 --> 00:10:27.179 60 .9 % majority. He pulls in 1 ,392 ,076 votes 00:10:27.179 --> 00:10:31.059 over Jones's 39 .74%. So what does this all mean? 00:10:31.320 --> 00:10:32.980 When you zoom out and look at the entire arc 00:10:32.980 --> 00:10:34.820 of this table, the most shocking thing isn't 00:10:34.820 --> 00:10:36.600 just the party shifts. It is the sheer scale 00:10:36.600 --> 00:10:38.799 of the math. The expansion of the pool. Yes. 00:10:39.360 --> 00:10:42.720 Think back to that 1918 row. 100 % victory required 00:10:42.720 --> 00:10:46.539 exactly 54 ,880 votes. That stadium -sized pool. 00:10:46.740 --> 00:10:49.120 Right. Now contrast that with the 2020 Tuberville 00:10:49.120 --> 00:10:51.600 versus Jones race. That single election featured 00:10:51.600 --> 00:10:54.460 over 2 .3 million combined votes. The effort 00:10:54.460 --> 00:10:56.720 required to capture a majority has just scaled 00:10:56.720 --> 00:10:59.779 exponentially. And the 2022 data confirms that 00:10:59.779 --> 00:11:03.330 massive pool is the new normal. Republican Katie 00:11:03.330 --> 00:11:08.029 Britt won the Class 3 seat with 66 .62%. That 00:11:08.029 --> 00:11:13.429 percentage translated to 942 ,154 raw votes for 00:11:13.429 --> 00:11:16.179 a single candidate. Confirming that massive modern 00:11:16.179 --> 00:11:18.200 voter pool and current Republican dominance, 00:11:18.460 --> 00:11:21.039 we really just traced a 100 -year arc straight 00:11:21.039 --> 00:11:24.139 through the numbers. From 54 ,000 votes to almost 00:11:24.139 --> 00:11:26.659 a million for a single candidate. We started 00:11:26.659 --> 00:11:29.299 in a time of single -party Democratic sweeps. 00:11:29.340 --> 00:11:31.539 We watched the emergence of independents and 00:11:31.539 --> 00:11:33.860 prohibitionists fracturing the data. We saw razor 00:11:33.860 --> 00:11:36.860 -thin transitions of power. And we finally arrived 00:11:36.860 --> 00:11:40.220 at these massive multi -million vote modern Republican 00:11:40.220 --> 00:11:42.179 majorities. It is an incredible mathematical 00:11:42.179 --> 00:11:44.120 journey. This raises an important question for 00:11:44.120 --> 00:11:46.679 you to think about. How quickly can data that 00:11:46.679 --> 00:11:48.919 feels totally permanent just flip over a few 00:11:48.919 --> 00:11:51.419 cycles? It really makes you wonder. We've seen 00:11:51.419 --> 00:11:53.519 the 17th Amendment change senators from being 00:11:53.519 --> 00:11:56.080 appointed by a legislature to being elected directly 00:11:56.080 --> 00:11:59.100 by the people. We've watched total votes surge 00:11:59.100 --> 00:12:03.309 from 50 ,000 to over 2 .3 million. So with technology 00:12:03.309 --> 00:12:05.850 and demographics changing faster than ever, what 00:12:05.850 --> 00:12:07.750 were the voting mechanism or even the definition 00:12:07.750 --> 00:12:10.690 of a majority look like in the statistical tables 00:12:10.690 --> 00:12:14.129 of 2126? It is definitely something to ponder. 00:12:14.269 --> 00:12:16.009 Thank you so much for joining us as we unpack 00:12:16.009 --> 00:12:18.509 the numbers today. We will see you on the next 00:12:18.509 --> 00:12:19.029 Deep Dive.