Wednesday, December 2, 2020

Is the Phrase "Defund the Police" Hurting the Democrats?

Former President Obama recently had an interview where had this to say about "Defund the Police":

"You lost a big audience the minute you say it [Defund the Police], which makes it a lot less likely that you're actually going to get the changes you want done..."

Obama also called "Defund the Police" a 'snappy slogan'. Now there are a lot of things that you can rightfully quibble with, from the reference of "Defund the Police" as a slogan (kind of belittling), bickering about audience (when actual people have died at the hands of the police), and complaining about losing power (where Obama was fairly derelict in maintaining the legislative branch as President). But if you want criticism, many people have already dragged Obama (including Rep.-elect Cori Bush who ran on a Defund the Police platform). But let's break down the idea a bit further. It boils down to one question: "How can you enact a progressive agenda without alienating the populace?"

The Statistics

Let's get it straight right away. Just because an idea is unpopular does not mean that the idea is inherently wrong. With that said, "Defund the Police" is an unpopular idea right now. According to Pew Research, only 25% of people polled supported reducing the budget for the police department while 31% of people polled supported increasing the budget. 42% supported no change. But let's go to Black Lives Matter, which is the organization tied closest to "Defund the Police". According to Civiqs data, in April 2017, only 38% of registered voters supported Black Lives Matter, a number that wavered throughout 2017 despite the Charlottesville Rally. However, in 2018, more registered voters supported Black Lives Matter than opposed (40% to 39%), but it stayed below a majority until this past summer. In the wake of protests over George Floyd, Ahmaud Arbery, and Breonna Taylor's deaths, Black Lives Matter hit a peak of 53% supporting, which fell to 48% supporting by end of November.

Another corollary you could look at is healthcare. Let's start with Obamacare. In 2013, only 24% of those polled thought it had a net benefit to the country as a whole compared to 38% who said it didn't. In 2017 (7 years after Obamacare was initially passed), that opinion flipped where now 44% of those polled thinks it has a net benefit compared to 35% who said it didn't. That's a 23 point swing in opinion over 4 years!

What did we learn

Yes - Defund the Police is an unpopular phrase right now, and it arguably had some impact on Democrats, who lost very real seats in the House and lost key Senate races (most notably the Maine Senate race). Losing power does make it more difficult to enact these changes, but it also does not mean we should abandon these sentiments. Ideas that are unpopular today can only become more popular in the future if we continue to raise awareness and challenge leaders to do the same. We must also support towns and cities who are willing to try this approach and elevate their stories because people will believe it can be done if they see it for themselves. So sure, Obama may have been right about this election, but for change we can believe in, we cannot stop using these snappy slogans.

Tuesday, December 1, 2020

Irrelevant Questions: Are Professional Teams More Likely to Win a Championship if Their Home State Voted for the President?

In an exercise of correlation not causation, I wanted to see if there were any presidents who acted as lucky charms or as jinxes for the states that voted for them. To narrow the analysis, I focused solely on champions in the National Football League (NFL), National Basketball Association (NBA), National Hockey League (NHL), and Major League Baseball (MLB). I also started with the first Super Bowl and based on it on the winner in a given year rather than for the season. For example, the first Super Bowl was in 1967 as part of the 1966-1967 season, so I assumed that the year in question is 1967. The following were the Presidents during that time:

  • Lyndon B. Johnson
  • Richard Nixon
  • Gerald Ford - Editor's note: Since Nixon resigned in 1974, I decided to say Ford was the active president for the 1975 and 1976 years, but used Nixon's electoral win in 1972 as the baseline. This is despite Ford not being on the Nixon ticket in 1972.
  • Jimmy Carter
  • Ronald Reagan
  • George H.W. Bush
  • Bill Clinton
  • George W. Bush
  • Barack Obama
  • Donald Trump
Here were the results:

# of Champs in States that voted for the President of the given year

Before we highlight some clear gaps in this analysis, let's look at what we can find:
  • Barack Obama seemed to be a lucky charm for states that voted for him, as teams in those states won 27 out of the 32 possible titles (or 84.38% success rate). He was followed by Ford, Johnson (both 75%), and Reagan (71.88%). Obama is also the only president where teams across all sports wont more than 50% of the time.
  • Donald Trump seemed to be a jinx for states that voted for him, as teams in those states won 6 out of the possible 16 titles (or 37.25% success rate). He was followed by GW Bush (38.54%) and Nixon (41.67%)
  • Baseball teams seemed to benefit most from the boost, winning 35 out of the 53 possible titles (66.04%). Hockey seemed to benefit the least, winning only 28 out of the 53 possible titles (52.83%)
  • The Democrat vs. Republican spread is most pronounced in baseball (spread of 32.59%), where baseball teams in states that voted for the Democrat president won 18 out of 21 years (85.71%) vs. baseball teams in states that voted for the Republican president won 17 out of 32 years (53.13%). It is followed by hockey (27.56%), football (15.06%), and basketball (1.14%).
There are some obvious gaps:
  1. The number of teams, the team locations, and the state's voting results change, so there's no clear normalization in this analysis.
  2. Especially in more recent years (after 2000), there has become a clear urban vs. rural divide when it comes to voting. This means more states with major league teams would tend to vote for Democrats.
  3. There are a significant number of Canadian teams in the NHL as well as smattering of Canadian teams in the NBA and MLB. Provinces in Canada obviously do not vote for Presidents.
I tried to normalize the results further by looking at the number of teams in any given year that are in a state that voted for the President vs. the total number of teams, giving everyone an equal chance of winning the championship. I also looked at the average number of votes received from the electoral college. Using these, I was tried to normalize the "luck" vs. the "charm".

% of champions vs. % of teams that voted for the President

This margin analysis provides some interesting insight.
  • Obama is the only president where the % of champions was higher than the % of teams (with a fairly significant margin of ~16%).
  • All other presidents saw underperformances, but GW Bush actually had the best underperformance (margin of -3%) as there were fewer teams in states that voted for him compared to other presidents. This likely coincides with the fact that GW Bush had the lowest average electoral college totals (excluding Ford) for the Presidents listed.
  • Nixon saw the greatest underperformance (margin of -17%).
Percentage points per Electoral College vote

Looking at the percentage points / electoral college votes, we see a similar trend. Obama had the highest performance for each Electoral College vote (24.2 points per electoral college vote) while Nixon had the worst (10.2 points per electoral college vote). Interestingly, Republicans had fewer electoral college points compared to the Democrats. 

What does this mean under a Biden presidency?

Using the average margin vs. % of teams (and no changes in teams in the leagues) as well as the average percentage points per Electoral College vote for Democratic presidents, we would expect 9 of the eventual 16 champions to come from states that voted for Biden. Using Obama's margin (which seems fitting as Biden was Obama's VP) and his percentage points per Electoral College vote for Democratic presidents, that goes up to 12 out of 16 teams. Since baseball seems to be the Democrats strong suit, I'd expect a Democratic state to win the World Series next year (bad news Braves and Rays fans, good news Dodgers, Yankees, Padres, and White Sox fans!). Generally, the first year for Democrat presidents is the worst (both Carter and Clinton went 1-4 in their first year as President), but I'll give Biden a bit of an Obama boost and say that the NBA champ will also be from a Democratic state. I think the Super Bowl champ and the Stanley Cup champ in 2021 will be from a Republican state. Does this mean anything and should you gamble using this information? No! But if it works out...who knows.

Wednesday, November 25, 2020

A Tale of Two Emails Part II: Analyzing Joe Biden's Emails to His Supporters

It's been a bit more than 2 weeks since news outlets called the presidential race for President-Elect Joe Biden, which as a Biden supporter and donor to his campaign, is a huge sigh of relief. While other members of the government have yet to accept the results of the election, many states are certifying results, and it's only a matter of time before President-Elect Biden becomes President Biden. But how did we get here? Similar to last week's post, we're going to analyze the Biden team's emails to its supporters to see what changed between the final Presidential Debate, Election Day, and a week after the race was called.

Before we get into it, just a couple of editor's notes:

  • I did donate to the Biden campaign, which may or may not have affected the number and type of emails I received
  • The Word Clouds were done using Power BI, and I didn't really learn much in the last week
  • I copied the data manually from my emails (110 from October 23 to November 15). So there is some inconsistency and potential human error.
Analyzing the Subject Lines

Here's a word cloud of the subjects of Biden emails from October 23 to November 3 (final presidential debate to Election Day):

Word Cloud from subjects of Biden emails: Oct 23 to Nov 3

Over these 11 days, Biden sent only 87 emails. The top 6 words in the subjects (excluding common words) are:
  • Last: 13 occurrences
  • Need: 8 occurrences
  • Ask and Time: 7 occurrences
  • Please and Today: 6 occurrences
Other than the irony of "Last" occurring the most times, there's nothing exception about the wording. There is a more soft and kind tone which is struck, maybe due to the fact that I've donated to the campaign previously.

Here's a word cloud of the subjects of Biden emails from November 4 to November 15 (same timing as the Trump emails analysis):

Word Cloud from subjects of Biden emails: Nov 4 to Nov 15

Over these 12 days, Biden sent only 33 emails (< 20% the number of Trump emails, which makes sense given that Biden was announced the winner in this span). Only 14 emails were sent after Nov 7, which was when nearly all news outlets named Biden the presumptive President-elect. The top 6 words in the subjects (excluding common words) are:
  • Biden, Fight, Fund, and Need: 4 occurrences
  • Asking and One: 3 occurrences
While unfortunate that Biden also created a fund to support the legal battles that Trump was waging (the Biden Fight Fund), there's no real surprises here either.

Analyzing the Bodies

Here's a word cloud of the bodies of Biden emails from October 23 to November 3:

Word Cloud from bodies of Biden emails: Oct 23 to Nov 3

The top 7 words in the bodies (excluding common words) are:
  • Final: 180 occurrences
  • Goal: 169 occurrences
  • Make and Need: 163 occurrences
  • More: 160 occurrences
  • Joe: 139 occurrences (for reference, Biden came up 82 times)
  • Election: 126 occurrences
As for names,  Donald came up 54 times, Trump came up 89 times, Kamala came up 81 times, and Harris came up 50 times. Again - a fairly banal set of words in the body which are to be expected from any campaign donation. Anecdotally, the emails also read a lot more passively compared to the Trump campaign emails (again, highlighting the relatively boring quality of the Biden run for presidency - or the "return to normalcy and decency").

Here's a word cloud of the bodies of Biden emails from November 4 to November 15:

Word Cloud from bodies of Biden emails: Nov 4 to Nov 15

The top 6 words in the bodies (excluding common words) are:
  • Biden: 73 occurrences (the dynamic flipped from the previous set of emails as Joe showed up only 31 times)
  • Fight: 64 occurrences
  • Fund: 61 occurrences
  • Trump: 43 occurrences (both as "Trump" and "Trump's")
  • People: 42 occurrences (interestingly - Kamala Harris' presidential slogan was "For the People")
  • Democrats and election: 41 occurrences
  • Country: 39 occurrences
There's a clear highlight of the legal battles that were occurring at the time (legal came up 26 times and lawsuits came up 21 times), which was the primary use of the Biden Fight Fund. Biden's presidential win was also highlighted as victory occurred 18 times and president occurred 19 times.

What can we learn?

I think one of the biggest lessons here is that the Democrats ran a very vanilla campaign. There was nothing inspiring from Biden's emails, whereas Trump's emails showed a much clearer charisma. However, the Biden team also likely realized that the "return to boring politics" would be a benefit to his image as a uniting force in the country. It also showed the unprecedented legal challenges that had occurred over the past few weeks. There are still a number of Trump emails that come through regarding the election, and a number of lawsuits. The election is being litigated to a significantly more extreme degree than prior elections and it is laying the groundwork to undermine the legitimacy of our electoral process. As I said previously, we must continue to fight the misinformation that is coming from Trump's legal team and show that the election was legitimate.

Monday, November 16, 2020

A Tale of Two Emails - Analyzing Donald Trump's Emails to Supporters

Through some tomfoolery, I somehow landed myself onto Team Trump's email list. While I largely ignored these emails prior to the election (having firmly decided to vote for Biden), I couldn't help but notice how his messaging became far more combative after the election had been decided. This comes as no surprise to those closely following US news; Trump has clearly sought to sow seeds of doubt in America's democratic process, having in all likelihood lost the election. He has done so on Twitter and in public settings. But one mode of communication that should not be lost is how he communicates directly to his base. Twitter and press conferences are almost theater for those who don't support Trump. But his emails to supporters echo the same sentiments.

Before we get in, just a couple of Editor's notes:

  • I did not donate to the Trump campaign, which may explain the number of emails received
  • I likely put in that I am an Illinois resident, which means I may not have received as many emails related to contested elections in Pennsylvania, Wisconsin, or Georgia has residents of those states
  • The Word Clouds were done using Power BI, and I am new to the Word Cloud functionality. Hopefully, I'll get better at it :)
  • I copied the data manually from my emails (321 from October 23 to November 15). So there is some inconsistency and potential human error. But the overall story tracks. Also, unless you want to review the 321 emails I received, this is the best you'll get.

Analyzing the Subject Lines

If you're lazy like me, you decide to read emails based on the subject headers. So let's take a look at those first. Here's a word cloud of the subject headers before Election Day and after the final Presidential Debate:

Word Cloud from subjects of Trump emails: Oct 23 to Nov 3

Over these 11 days, the Trump team sent out a respectable 121 emails (or 11 a day). The top 8 words in the subjects (excluding some common words and weird omissions) are:

  • Trump and win: 10 occurrences
  • Vote: 9 occurrences
  • President: 8 occurrences
  • Deadline, Election, Tomorrow, Biden: 6 occurrences
Then comes the day after Election Day. Michigan and Wisconsin, which were looking to be Trump wins on Election night, were "red mirages" where the tallies of mail-in votes helped spur Biden to victories in those states. As that happened, the narrative Trump supporters received from the President drastically changed.

Word Cloud from subjects of Trump emails: Nov 4 - Nov 15

In these 12 days, the Trump team sent 200 emails, an astounding 16+ per day. The top 9 words in the subject lines were:

  • Election: 32 times
  • Defend: 14 times
  • Need: 12 times
  • News, Defense, Update: 7 times
  • Democrats, Help, President: 6 times
Trump used the word "Election" in his subject headers in a whopping 16% of his emails to his supporters. He also brought in words like "Defend" and "Defense", both in relation to "defending the election". It's clear that the message shifted away from casting a vote for President Trump to defending the sanctity of an election that, for all intents and purposes, did not need any defending.

Analyzing the Body of the Email

There's also a lot of messaging beyond the subject (obviously). Here's what the bodies of the emails look like between the final debate and Election Day. Editor's note: I copied these all from the emails in a fairly inconsistent manner, so terms like "Contribute" will be ignored:

Word Cloud from bodies of Trump emails: Oct 23 to Nov 3


The top 5 words in the bodies (excluding some common words and weird omissions) are:

  • President: 206 occurrences
  • Trump: 156 occurrences
  • Election: 143 occurrences
  • Day: 118 occurrences
  • America: 93 occurrences
Additionally, the bodies of the emails mentioned Biden 59 times and Kamala 26 times, though interestingly the word cloud does not include Joe or Harris (likely due to Trump's penchant for nicknames).

Now let's look at the bodies of the emails between the day after Election Day and November 15 (again with the analysis excluding terms like "Contribute"):

Word Cloud from bodies of Trump emails: Nov 4 to Nov 15

The top 6 words in the bodies (excluding some common words and weird omissions) are:

  • Election: 413 occurrences
  • Trump: 369 occurrences
  • President: 278 occurrences
  • Official: 159 occurrences
  • Defense: 155 occurrences
  • Fund: 148 occurrences
More troubling, the following words showed up an inordinate number of times:

  • Fight: 100 occurrences
  • Steal: 45 occurrences
  • Force: 36 occurrences
  • Undermine: 22 occurrences
The analysis of the body shows two things:
  1. A frightening increase in aggressive rhetoric. While some may argue that "fights" or "force" can refer to upcoming legal battles, this language has arguably spurred violent actions well beyond the courts, especially when you look at some of the potential plots stopped in Philadelphia (link). Terms like "Steal" or "Undermine" attack the heart of democracy and the faith in the election system.
  2. A clear fundraising effort. Donald Trump has established an Election Defense Fund, and for those of you who don't know, a lot of this money is not going to the legal battles related to the election, but rather the RNC and Trump's PAC (link). Trump appears to be actively undermining faith in our democracy to raise money for the Republicans and his PAC.
What can we learn?

Obviously, Trump's rhetoric to the public has been troubling, but it is reverberating even further in the echo chambers that his supporters are in. Beyond conservative media and conservative Twitter heads, Trump's direct messaging to his supporters is increasing in aggression and turnaround. Even on November 15, 12 days after the election, Trump was sending emails to his supporters at a pace of one every 90 minutes. And it is clear that the rhetoric will not simmer down as many hope, especially with recent news regarding Floyd County (link) where arguments of a "stolen election" will only be amplified. This is despite the fact that (1) this was only 2600 total votes resulting a shift of 800 votes in Trump's favor and (2) this is likely an isolated incident of human error (especially when considering the fact that Fulton county is already done with their hand count). We cannot expect Trump to tamper down his rhetoric, but we must continue to have diligence in reading the news, listening to our election experts, and trusting our democratic infrastructure. And more importantly, we must fight the misinformation that will only continue to increase.

I'll be working on a companion piece where I analyze Joe Biden's emails during the same timeframe. So please stay tuned for it!


Tuesday, November 10, 2020

Impact of Progressive Policies for Re-election

For those following the elections (not just the Biden-Trump one), you'll see a common theme. The Democrats may have won the Presidency, but they didn't have the same success down-ballot. Representative Abigail Spanberger VA-7 was purported to have vehemently stated her concerns about the "socialist" tilt of the Democrat party. Representative Alexandra Ocasio-Cortez NY-14 (AOC for her supporters, Rep. Ocasio-Cortez for her haters) tweeted the following:

While this is factually true (the seven Democrat representatives in GOP-leaning districts who co-sponsored Medicare for All or HR 1384 won re-election), I wanted to see if there were additional measurables that captured the impact of M4A and the other big-ticket progressive bill (the Green New Deal) for the House of Representatives.

Medicare for All

There were seven Democrats that co-sponsored Medicare for All that were in GOP-leaning or toss-up districts (the Cook Partisan Voting Index in parentheses):

  • Rep. Ann Kirkpatrick in AZ-02 (R+1)
  • Rep. Josh Harder in CA-10 (EVEN)
  • Rep. Katie Porter in CA-45 (R+3)
  • Rep. Mike Levin in CA-49 (R+1)
  • Rep. Jared Golden in ME-02 (R+2)
  • Rep. Peter DeFazio in OR-04 (EVEN)
  • Rep. Matt Cartwright in PA-08 (R+1)
All seven representatives won re-election and as-of now, no co-sponsor for Medicare for All has lost their re-election bid. 

The eight seats that have been flipped so far on the Democrat side were all representatives who did not co-sponsor Medicare for All. While some seats were in Democrat-leaning districts (such as FL-26 and FL-27), the seats in GOP-leaning districts were much more heavily GOP-leaning. The districts as-of now were:
  • MN-07 (R+12)
  • NM-06 (R+6)
  • OK-05 (R+10)
  • SC-01 (R+10)
This leads to the question: did Medicare for All really help the candidates in these districts or were there other factors that supported these candidates' re-election campaigns?

To analyze this, I focused on the margins of victory for those who supported Medicare for All vs. those who did not, looking primarily at candidates in districts with a PVI of 0 or R+. I plotted the margin between the next closest competitor in 2018 vs. 2020 on a scatter plot. Everything to the left of the dotted line represented an improvement from 2018 to 2020:


The orange diamonds represent those that supported M4A. Four of the seven representatives (57%) that supported M4A saw an improvement in their margin between 2018 and 2020 compared to four of the thirty-four representatives (12%) saw an improvement in the margin. 

I also looked at the change in margin between 2018 and 2020. The average change in margin between 2018 and 2020 for these representatives was -3.1 points (6.4 points in 2018 to 3.3 points in 2020). For those that supported M4A, the average change in margin was -0.3 points (. For those that did not support M4A, the average change in margin dropped to -3.7 points. 

When weighting for the % of vote reporting per the New York Times, the weighted change in margin (change in margin * % of reported vote) is -2.8 points. For those that supported M4A, the weighted average change in margin was -0.6 points. For those that did not support M4A, the weighted average change in margin was -3.3 points.

Note that those that supported M4A did have a higher starting margin in 2018 by approximately 2 points, but overall, those that supported M4A performed significantly better than those that did not:


So what does this mean for the Green New Deal?

The Green New Deal is a significantly different matter. For one thing, two of the big pieces of the GOP plan coming into the primary revolved around the updated tax plan and, more relevantly, the attempted repeal of Obamacare. Many of the representatives in the blue wave may have been expected to deliver landmark healthcare for their constituents and the support, or perceived lack thereof, for nationalized healthcare may have buoyed or cratered the representatives' campaigns. 

Additionally, polling for the two ideas have different levels. Per Figure 7 of the following link from the Kaiser Family Foundation, 56% of people surveyed have a favorable view to Medicare for All. The Kaiser Family Foundation also published a study on the Green New Deal (albeit in 2019), and for those who have heard of the Green New Deal, only 36% supported it. While this is only a single, unvetted data point, there is discrepancy on how popular each of these proposals are.

Practically, one co-sponsor of the Green New Deal (Rep. Debbi Murcarsel-Powell of FL-26) has already lost re-elction and another co-sponsor (Rep. Thomas Suozzi of NY-03) is facing a challenge from his GOP opponent. So the Green New Deal has not had the same effect as Medicare for All has had to-date. That does not mean the more progressive idea is inherently bad politically; an argument can be made that Sen. Ed Markey was able to fight off a primary challenge from Rep. Joe Kennedy in part for his sponsorship of the Green New Deal. But there may need to be more concrete benefits for constituents before the Green New Deal becomes a viable option up and down the Democratic party.

What can Democrats learn?

It is clear that the Democrats are at an impasse. With a loss of seats in the House and a Sen. McConnell-led Senate (unless both GA senate candidates win their runoff elections), Democrats will be hard-pressed to pass meaningful legislation even with a President-Elect Joe Biden at the helm. Making this even harder is the wide political spectrum in the Democratic party. In the Senate, Sen. Bernie Sanders (I-VT) represents the leftists, while senators like Sen. Joe Manchin (D-WV) and Sen. Joe Tester (D-MT) who are both in GOP-leaning states represent centrist views. And in the House, you have Rep. Ocasio-Cortez and Rep. Spanberger on two differing sides of the argument. A strictly progressive agenda may be a pipe dream, but there are clearly progressive ideas that have support from the American people. If there's anything to take from this, it is this: Medicare for All can and in my opinion should be a centerpiece for the 2022 primaries, even if President-Elect Biden is able to get the public option passed.