Geopolitics Review - 23rd September 2024
Geopolitics Review - 23rd September 2024
State-Level Hypothesis Testing
Contents
Introduction
Harris 2024 vs Biden 2020
Urbanisation By State
Problems With Polling and Forecasting
Concluding Remarks
Bitesize Edition
If we focused on how every event could affect the outcome of the U.S. election, we’d drive ourselves crazy. Many events have little impact on the overall result, and some events aren’t even seen in changes in the polls. There is a lot of noise when it comes to elections.
In the United States, the presidency is won through the electoral college which I’ve discussed previously. Biden beat Trump in 2020, and so I questioned how Harris is currently polling today in comparison to Biden in 2020. If Harris is polling better, is victory in November nailed on? If Harris is polling worse than Biden, does she have a tougher path to the White House, or is her path completely blocked, paving the way for a Trump victory?
In the buildup to elections, there are no certainties. We can look back with hindsight after the election has occurred and reflect on our predictions, but when approaching the election date, we can’t assume anything with certainty. What we can do prior to an event, is see if the probabilities and odds are on our side with any prediction we make. Does the data support the trends and evidence we believe exist? That’s what I’ll explore today on a state level. Is Harris polling better or worse than Biden did in 2020, and how will this affect her campaign from here?
Introduction
In what will be my final piece for now covering the US election, I’ll explore how Harris is polling in 2024 in comparison to Biden in 2020. Biden eventually won the Electoral College by a margin of a few states in 2020. Will Harris command such a victory, will her battle be tighter than Biden’s, or is she set for a defeat?
I’ll also explore a longer trend unfolding in the United States, with that being urbanisation and its effects on politics in the United States.
Harris 2024 vs Biden 2020
Null Hypothesis (H(0))= Harris's polling in 2024 is the same as Biden’s polling in 2020.
Alternative Hypothesis (H(1)) = Harris is polling worse in 2024 than Biden was in 2020.
Data (Figures Are Percentages):
Note: Figures are rounded and so gaps between candidates sometimes appear miscalculated.
From the data, Harris is polling worse than Biden in 14 of the 17 states I’m analysing. I chose these states due to their narrow margins in the 2020 election results, or in the current 2024 election polling. The Harris underperformance is seen in the bottom row, with Harris polling better than Biden only in Alaska, Maine, and Colorado as represented by a positive figure.
Of course, the important aspect is which states fall to which candidate, and not if Harris or Biden is polling better.. Harris could perform worse than Biden and still win, of course. Its here that were have to once again explore the election on a state level.
Of the states explored in the dataset, Biden won 13 of the 17 in 2020.
By Harris’ polling, she is predicted in 11 of the 17 states. The losing states for the Democrats from Biden 2020 to Harris 2024 are Georgia and Arizona. All other states are currently predicted to fall as they did in 2020 in 2024.
Before November, this of course can shift, but Georgia and Arizona along with Pennsylvania are, in my view, shaping up to be part of the most important states in this election.
Let’s continue with the hypothesis test.
Mean Difference Between Biden and Harris = -2.85%
Standard Deviation = 3.5%.
The standard deviation shows the potential spread of results when comparing Biden and Harris. Hence, in taking the mean of -2.85% and adding and taking away one standard deviation, Harris could outperform Biden by 0.65% in all states or underperform by as much as -6.35%.
We will be using the Student T-Test for this hypothesis test. This is due to the small sample size of less than 30, and we had an unknown standard deviation which we had to predict using the data as seen above.
Test Statistic
Degrees Of Freedom = N - 1 = 16
P-Value = 0.002.
Decision = The P-value of 0.002 is less than our 5% significance level of 0.05. Hence we reject the null hypothesis. This means Harris is polling worse than Biden did. The mean of -2.85% when comparing Biden and Harris supports that Harris is polling worse than Biden. Of course, we could have determined this without the hypothesis test, but I wanted to highlight the process that is performed.
With this conclusion in mind, where do we look next?
By current polling, the Democrats would win 292 electoral votes versus 246 for the Republicans. The finest margin states are North Carolina (0.2%), Nevada (0.8%), Pennsylvania (0.8%), and Arizona (-0.8%).
What does leaving these four states as undetermined lead to?
With all these states as tie-ups, the Democrats have three pathways to victory, versus two for the Republicans.
One of three such paths to Democrat victory requires they win Pennsylvania only (19 Votes).
The two Republican victories in this scenario must see the Republicans win Pennsylvania.
If the Democrats don’t win Pennsylvania, they need North Carolina (16 Votes) or they have no pathway to victory.
Again, regardless of how many scenarios are considered, it seems all roads lead to Pennsylvania. This is likely an indication as to why the Presidential Debate was likely held there. Note, that this is one such scenario. As I’ve discussed before, many other states can change before November.
Many questions could be asked, such as will Tim Walz as VP pay off for the Democrats? As many thought before the pick, would Josh Shapiro, the Governor of Pennsylvania, have put the race for the electoral college to bed?
Biden eventually won victory in the Electoral College in 2020. Harris polling worse than Biden means she’ll face a greater challenge than Biden did in winning the electoral college and hence the presidency.
One thing is clear. The race is far from over, especially in these key battleground states.
I’ve now explored this election in depth. However, there are long-term trends where politics and demographics intersect. One such area is urban voter support for Democrats.
Urbanisation By State
Last week, I discussed party systems. This refers to periods that often see one party dominate US politics for several election cycles.
In my analysis of demographic trends in the United States, one aspect that pointed towards strong and growing Democrat support was amongst Urban citizens. With urbanisation on the rise, I wanted to explore if this hypothesis is supported by the data, and hence could indicate a long-term potential trend in US politics.
Null Hypothesis (H(0)) = Urban voters do not align more with Democrats compared to suburban and rural voters.
Alternative Hypothesis (H(1)) = Urban voters align more with Democrats compared to suburban and rural voters.
For this test, we will use a z-test. This is because we will take an assumed sample size of 1000 for each group, 3000 in total. This is much larger than the n<30 that is required for the t-test used above. We are also comparing the proportions of voters, and the distribution of these sample proportions will approach a normal distribution or a bell-shaped curve. The normal distribution uses the z-test.
In 2016, our probability of urban voters choosing Clinton was 70%. The average of the suburban and rural voters (45% and 34%) is 39.5%. Below, we calculate P* as an overall probability across all groups, those being urban, suburban, and rural. The standard error refers to the amount of variability in our prediction. The Z-score is used to determine the p-values, through which we can complete our hypothesis test.
In 2020, our probability of urban voters choosing Biden was 66%. The average of the suburban and rural voters (54% and 33%) was 43.5%..
Decision = The p-value from the Z-scores in both the 2016 and 2020 data is less than 0.0001. This can be calculated through online calculators, or via Excel. Mathematically, we calculate this by utilising the cumulative distribution function for the standard normal distribution. This small p-value of less than 0.05 indicates there is strong evidence against the null hypothesis. Hence we reject the null in favour of the alternate hypothesis. According to the data evidenced, in 2016 and 2020, urban voters aligned more with Democrats than suburban and urban voters did.
In 2023, the rate of urbanisation growth was 0.7%. Will continued urban population growth lead to a stronger raw number of voter support for the Democrats? Demographic trends such as this could contribute to long-term party systems that have emerged in the history of the United States. Is this the start of another? It’s certainly a topic to keep an eye on.
Problems With Polling and Forecasting
In polling and forecasting, as I’ve discussed before, there are many problems that arise. There is always an amount of random error in any model, and we can sabotage ourselves in our polling or forecasting process. Here are some potential pitfalls we can be susceptible to that can’t be forgotten:
Sampling Error - If an overall sample doesn’t represent that of the total population, there exists bias. In this election cycle, polls that lean heavily in one direction, whether Democrat or Republican, risk overestimating this party in the sample, and hence also their probability of success. Also, small sample sizes have a greater margin of error, so polls must be sufficient in sample size.
Wording - If a poll uses a leading question, such as implying agreement with one party, then respondents are more likely to agree with them. Also, complex questions can lead to confusion or answers that don’t align with the question asked.
Response Bias - People answer with what they believe is socially acceptable, not what they truly believe.
Poll Timing - As I’ve reiterated in this series, there is still a long way to go in this election cycle. Polling today will differ from that of a month ago. It will certainly differ with polling from multiple months ago since the Democrats have a new candidate.
Polling Method - The use of technology can sometimes exclude certain demographics who lack access to the means to poll.
Adjustments - Over or underestimated demographic groups can be adjusted to correct for the imbalance. The process of adjusting the weights can lead to further bias.
Margin Of Error - This represents the range within which the true result is likely to fall. This can lead to small differences between candidates being interpreted as significant when the margin of error can determine the results are too close to call. This is especially prevalent in this election cycle.
Over-Reliance - 2016 is the perfect example of this. Clinton was expected to win, but Trump was underestimated in the polls and went on to win. Lessons must be learnt to ensure this doesn’t occur again.
Concluding Remarks
This concludes my work on the election in the United States in November. I’ll return to the topic in the weeks before the election date and we’ll explore it in the aftermath of the results. One thing is clear, a small handful of states will be the most important in determining which way this election falls. Neither side is out of the race at all, but regardless of my work or anybody’s work, any predictions remain uncertain and dependent on polling which can possess multiple flaws. The only certainties will be formed after the election.
In exploring longer-term trends, and its clear demographics can play a huge role in how elections unfold. As demographics change, so do the potential outcomes in elections. Do you have any long-term predictions regarding US politics?
Other News
Chinese Jet Tails US Spy Plane Making First Pass Over Taiwan Strait In 5 Months
Israel War Cabinet Greenlight Offensive War Against Hezbollah
JD Vance Stands By Springfield Pet Eating, Stating Firsthand Accounts As Source
Macron: France Stands By Lebanon And We Are Saddened By Civilian Casualties
Nasrallah: Residents Of The North Won’t Return Unless War In Gaza Stops
Nasrallah: Retaliation For Attacks Will Come - In Expected and Unexpected Ways
Netanyahu Adds New Goal To War: Return Northern Residents To Areas Safely
Romanian Defence Minister: Response To Russia’s Violation of NATO Airspace Necessary
UN General Assembly Pass Resolution For Israel To Pull Out of Palestine Within 12 Months
Thanks for reading! I’d greatly appreciate it if you were to like or share this post with others! If you want more then subscribe on Substack for these posts directly to your email inbox. I research history, geopolitics, and financial markets to understand the world and the people around us. If any of my work helps you be more prepared and ease your mind, that’s great. If you like what you read please share with others.
Key Links
The Geopolitics Explained Podcast
If you want to see daily updates and discover other newsletters that suit you, download the Substack App.
You can become a paid subscriber to support my work. There are long-form monthly articles in my global questions series exclusively for paid subscribers. Read Geopolitics Explained for 20p per day or start a free trial below to find out if my work is for you! I appreciate your support!
Sources
https://en.wikipedia.org/wiki/Urbanization_in_the_United_States
https://projects.fivethirtyeight.com/polls/president-general/2020/national/
https://poll.qu.edu/poll-release?releaseid=3901
https://www.270towin.com/2020-polls-biden-trump/
https://www.270towin.com/2024-presidential-election-polls/
https://en.wikipedia.org/wiki/Josh_Shapiro
https://data.worldbank.org/indicator/SP.URB.GROW?locations=US
If Trump wins in November the question of future voting trends due to the likelihood of increasing urbanization will become moot. He has already promised there will be no more elections “needed”if he regains the presidency.