sábado, febrero 22 de 2020

Datasketch

Octubre 27, 2016

Why Trump will win the US election contrary to nearly every forecast

What key patterns and factors throughout history have proven useful in forecasting the winner of the US presidential elections? What big data analytics are useful in measuring the success of the Clinton and Trump social network campaigns? Guillermo Valencia, from Macrowise tell us how this questions will be critical in predicting who will be the next president of the United States.

Guillermo Valencia

Guillermo Valencia

@macrowise

Hillary Clinton Incumbent, Donald Trump Challenger.  Analysis Macrowise

 
  • Many methodologies are used to try to predict the results of the elections: polls, demographic analysis, panels of analysts and/or experts on political issues, analysis of social networks that use big data, etc.

  • However, these polls and political experts are failing to capture the rise of new social conventions.

  • At Macrowise, we believe that the methodologies most successful in predicting the United States elections consistently,is the combination of two methodologies:

     1. Big data- understand if the campaigns are implementing    

         successful strategies on social networking sites.

     2. Analysis of patterns- analyze the performance of the incumbent

         party of the Presidency.

 
 

The Failures of Surveys

 
  • Nearly really all polls project Clinton as the winner, with the Real Clear Politics’ average of all major polls showing a 3.8-point current advantage (as of October 28) just as they did with the Brexit vote.

  • Beyond the end data of polls, a valid pattern of convergence is evident in both Brexit and the US election.

  • In the case of Brexit, the polls went from a peak 8 point lead of Britain remaining verses exiting the European Union, to a mere 2-point advantage in the final poll, and ultimately a complete reversal, where voters chose to exit.

  • In the case of the United States elections, Clinton had an advantage to Donald Trump's of 19 points at one time according to some polls. This difference has been deteriorating for months, as Trump as closed the gap to less than 4 points.

  • The best monthly variation for Trump has been 3.3 points per month, while the worst monthly variation for Clinton was - 3.3.

 
 

Brexit Survey.  Source: WhatUkThinks.org. Macrowise Analysis.

 

The problem with these surveys is that they fail to capture the vote against moral convention.

 
  • To exit the European Union, and similarly voting ‘No’ in the recent vote for peace in Colombia, were viewed as “morally wrong.” Biased analysis skewed the reality of the situation, as both votes ended up going against the “morally correct” and widely predicted outcomes.

  • In much the same way, voting for Donald Trump challenges the theory of Moral Convention and many people may be reluctant to share their true voting intentions in the polls and instead will wait until the actual election to exercise their seemingly unpopular decision.

 
 

US Presidential Election Survey. Source: Real Clear Politics. Macrowise Analysis.

 

Trump’s campaign on social networks appears to be more successful than Clinton's.

 
  • The investigation of Ajmani, Kashyap, Balasekaran posted on Github Research shows the structure of social networks for supporters of Trump and Clinton on Twitter.

  • Analysis of Trump’s supporters on social networks shows a distribution known as potential law, confirming a much higher level of connectivity and collaboration between diverse communities as identified by different colors.

 

Network of Trump activists on Twitter. Source: https://github.com/Rameshb-umd/Social-Network-Analysis   

 

 

In the case of Clinton’s active followers, one can clearly interpret the formation of three subgroups (cliques) that not are connected with the same fluidity as Trump’s. Clinton is supported by many disconnected individuals.

 

Network of Clinton acitivists on Twitter. Source: https://github.com/Rameshb-umd/Social-Network-Analysis  

 

 

Twitter stats indicate that Trump has approximately 10,000 tweets every 15 minutes, while Clinton only 3,000 tweets over the same period of time. One can only guess whether or not the extra chatter will prove beneficial, but Trump’s social media strategy and overall visibility are clearly more prevalent than Clinton's leading to a stronger case in favor of Trump.

 

Analysis of historical patterns and judging the power of the incumbent party.

 

The methodology most successful in predicting the United States elections consistently is the analysis of patterns. This analysis helps one understand the events that create tectonic movements within the party in charge of the Presidency of the United States.

 

This analysis consists of asking key questions that go beyond of the push-pull of the debates and the news. This methodology was developed in a paper by the Russian professor V. I. Keilis-Borok, specialist in prediction of earthquakes at UCLA, and the prestigious professor and historian Allan Lichtman of American University.  

 

They ran an analysis from 1860 until 1980 and identified 13 key questions that have predictive power in the United States elections. Since 1984, they have applied this methodology and correctly predicted every election. Their current prediction is that Donald Trump will be the next President of the United States.

 

The concept of Lichtman is based on the elections are the accountability of the political party in charge of the Presidency. This methodology divides the candidates in two categories: incumbent and challenger. Then a series of 13 true/false statements and answered, known as the "Keys to the White House" to determine the predicted winner.

 

For example: strength of party manager, leadership within the party, re-election, third party, economic outlook for short and long term, reforms of national interest, social dissatisfaction, scandals, military failures, foreign policy, the incumbent candidate, and the charisma of the challenger. The rule specifies, that a true statement always favors the incumbent party, and if there are 6 or more “falses” the winning candidate will be the challenger.

 

Below are the key statements:

 

ID

Question

Category

Y/N

1

After the midterm elections, the incumbent party holds more seats in the U.S. House of Representatives than after the previous midterm elections.

Party Mandate

No

2

There is no serious contest for the incumbent party nomination.

Contest

Yes

3

The incumbent party candidate is the sitting president.

Incumbency

No

4

There is no significant third party or independent campaign.

Third party

No

5

The economy is not in recession during the election campaign.

Short-term economy

Yes

6

Real per capita economic growth during the term equals or exceeds mean growth during the previous two terms.

 

Long-term economy

Yes

7

The incumbent administration effects major changes in national policy.

Policy Change

No

8

There is no sustained social unrest during the term.

Social Unrest

Yes

9

The incumbent administration is untainted by major scandal.

Scandal

No

10

The incumbent administration suffers no major failure in foreign or military affairs.

Foreign/ military failure

Si

11

The incumbent administration achieves a major success in foreign or military affairs.

Foreign/ military success

No

12

The incumbent party candidate is charismatic or a national hero.

Incumbent charisma

No

13

The challenging party candidate is not charismatic or a national hero.

Challenger charisma

Si

 

Statement 1 is NO, as the Republicans control 57% of the House of representatives. The second no comes for key question #3 as Obama is not eligible to be the candidate for the Presidency. The third no is achieved from statement 4. Gary Johnson has more than 6% of popularity and is supported moreso by Democrats than Republicans, taking more votes away from Clinton. The 4th no in the analysis is triggered by statement 7, there was no major policy change in Obama's second term like the Affordable Care Act in the past election. The fifth no pertains to Clinton’s email revealed by Wikileaks. This could cost Clinton as much a 6 points. The sixth is no is keyed in statement 11. Obama has not had a military triumph during his second term. The seventh and last no is from statement 12. Hillary Clinton is not a national heroine, nor does she have the charisma of Obama.

 

Donald Trump is a candidate with very little reference in the history of United States, threats to politicians, and has identified with Putin as a model to follow. He has been involved in sexual scandals, speaks disparagingly of immigrants and women, and does not have experience in politics. However, he is business man with a great capacity for leadership and action and has the support of many powerful people and organizations as well as many terrified citizens.

 

In conclusion, Macrowise has considered that there are two main factors playing a decisive role in the upcoming Presidential Election of the United States: The success of Trump’s social media campaign and the poor performance of the Democratic Party in the current term.

 

Our prediction is that Donald Trump will be the next President of the United States.

compartir

Guillermo Valencia

Guillermo Valencia

@macrowise

Co-fundador de Macrowise, una firma de investigación macroeconómica con énfasis en análisis geopolítico y disrupciones técnológicas. Columnista de Dinero.com