Congress victory hands Trump sweeping powers
Congress victory hands Trump sweeping powers
The Times, November 15, 2024: 28.
BBC, US election: https://www.bbc.co.uk/news/topics/cj3ergr8209t
Biden: bring down the temperature
The Senate and the House of Representatives
An eventful year
The results for the presidency
Election step ratios: UK slump and US stumble
Incumbency: historical and international comparisons
Predictions for 2024 using past results rather than opinion polls
My predictions for 2028 and for all time
Predicting the opinion polls in 2028 and for all time
Commentary, opinion polls and election results …
… my commentary, Thursday 30th October, 2024
… my commentary, Thursday 7th November, 2024
The voting was on Tuesday 5th. It is now the 15th. Counting has continued throughout this period and is even now not complete. Trump had won quickly but the Senate took longer and the House only now is definitely past the winning post. Three wins for the Republicans - although the final percentages are not yet known (and so the error in the opinion polls is not yet known). Biden has invited Americans to “lower the temperature” and he promised a peaceful transition when Trump visited the White House. Meanwhile Trump is putting together his team. Some of his choices have proven contentious and there is speculation as to whether they will be confirmed by the Senate and the House.
Biden: bring down the temperature
“Biden called for unity after Tuesday’s fractious contest. ‘You can’t love your country only when you win,’ he said, ‘You can’t love your neighbour only when you agree. Something I hope we can do, no matter who you voted for, [is] to see each other not as adversaries, but as fellow Americans. Bring down the temperature.’ ”
“Bring down the temperature, urges Biden as he is blamed for defeat.” The Times, November 8 2024 6-7.
The Senate and the House of Representatives
“Matt Gaetz for Attorney General … It almost certainly won’t happen. The AG role requires Senate confirmation and, with a Republican majority there of 53-47, only four GOP senators could block the nomination.”
Gerard Baker. “Think Trump’s an isolationist? Think again.” The Times, November 15, 2024: 25.
The Senate
The Senate has 100 senators. In the new Senate, a third were voted for last week; a third (34) were voted for in the mid-term elections of 2022; and a third were voted for in 2020. In the old senate the Republicans just fell short of a majority, with 49 seats, the Democrats having 47 and Independents 4 seats. Now in the new Senate, the Republicans have a slender absolute majority with 52+ seats, and Democrats on 45+ seats and Independents on 2.
The details for last week are that 34 of the 100 senate seats were up for re-election.
Democrats retained 17 of their 19, losing 2.
Republicans retained all 11 of their sets, and gained 3.
Independents retained 2 of their 4, losing 2.
Thus a large proportion - 30 of the 34 seats - did not change hands (net).
https://en.wikipedia.org/wiki/2024_United_States_Senate_elections
House of Representatives
The House of Representatives has 435 seats, with 218 required for a majority. In 2022, the Republicans had a majority with 222 seats, Democrats having 213 (51% and 47.3% of the all votes, respectively). Now after last week’s vote Republicans again have a majority with 220 seats, Democrats having 212. Thus, between 2022 and 2024 there was very little change in the number of seats.
https://en.wikipedia.org/wiki/2024_United_States_House_of_Representatives_elections
An eventful year
We now quickly sketch some key events in the year leading up to the election. President Joe Biden announced he would be standing for re-election. Throughout the Republican primaries Donald Trump dominated even as he lost a number of court cases. Two old white men. In a fateful debate with Trump, Biden showed great mental confusion and this led to increasing pressure on him from his party to step down. Later the Democrats adopted Kamala Harris as their candidate. There was just one debate between Harris and Trump, with Harris to some extent doing better than Trump. Trump suffered a glancing injury in a failed assassination attempt.
In polling in the period 2021-2024 about a possible Biden v Trump contest, Biden had been about 10% ahead into 2021 but that gap closed by the end of the year and in the following three years the gap was never more than a few per cent with two Trump mini-peaks and two Biden mini-peaks, Trump being ahead of Biden from Autumn 2023 onwards; the gap first narrowing slightly and then recovering slightly.
In polling about the Harris v Trump contest, Trump was ahead in July 2024 but the gap closed in early August, with Harris increasing her lead to a maximum in late September, falling back to a small lead in early November. Looking further back to 2021 Trump led Harris by around 53% to around 47% throughout much of the period – before Harris became the Democrat candidate.
It is worth noting that nothing in the polling at any stage until the crucial debate recommended anything other than a Biden v Trump contest – in the sense that no other candidate appeared likely to do better for their party.
Into the final week the polls were fairly evenly divided between Harris and Trump. Different individual polls gave hope to each side. Five polls reported on Monday 4 November. Three gave a win for Harris by 2%, 2% and 4%; one gave a win for Trump by 1%; and one gave a tie. The average of the five polls is 1.4% for Harris [“538” put the Harris lead at +1.2%].
The results for the presidency
And then the day itself … “early exit polling bodes ill for Harris” … and this was confirmed as the results started coming in. In the end Trump won 312 electoral college votes and Harris just 226.
Turning to the popular vote, The Times on Thursday morning put Trump on 50.98% and Harris on 47.52%, a lead of 3.46% for Trump. At that stage not all the votes had been counted. A week later, Wikipedia was reporting, based on 95% of the vote, that Trump had 50.2% and Harris had 48.2%, a lead of 2.0% for Trump. With that change in the available results, there is a corresponding change in the perceived error in the polls.
Last week’s result can be looked at in a variety of ways. Trump was ahead by just 2% in the popular vote but won 12 more states than Harris and was well ahead in the electoral college. Compared with Biden in 2020, Harris was down by 3.1% and down by six states. On the other hand, compared with Clinton in 2016, the vote share of 48.2% was the same. In terms of social groups, Trump had greater support amongst those with attributes of military service, no college education, white and male – much as he had in 2016 and in 2020. There was a common general fall in support amongst all groups – such as men and women – but also a specific extra fall for certain groups such as low-income groups and Latin groups. In terms of issues, Trump supporters were concerned about the economy and about immigration whereas Harris supporters were concerned about democracy and about abortion.
Commentaries have raised a variety of points. Perhaps there is a tendency for incumbent parties to lose in these difficult times – see the next two sections. Perhaps the Democrats have been too derisive of their opponents – or fail to understand them.
So.A small Trump lead in percentage vote but a large lead in number of states and number of college votes. A small change in the percentage vote and in the number of states gained but a large change in the number of college votes. A small general change but substantial changes for specific groups.
Table The results for 2024
2024
Trump 50.2%; Harris 48.2%; Other 1.6% … Trump lead 2% [still counting]
Trump 31+ states; Harris 19+ … Trump lead 12
Trump 312 college votes; Harris 226 … Trump lead 86.
turnout ?
Table The results for 2024, 2020 and 2016. The changes
2024
Trump 50.2%; Harris 48.2%; Other 1.6% … Trump lead 2% [still counting]
Trump 31+ states; Harris 19+ … Trump lead 12
Trump 312 college votes; Harris 226 … Trump lead 86.
turnout ?
2020
Trump 46.8%; Biden 51.3%; Other 1.9% … Biden lead 4.5%
Trump 25+ states; Biden 25+ … tie
Trump 232 college votes; Biden 306 … Biden lead 86.
turnout 66.6%
2016
Trump 46.1%; Clinton 48.2%; Other 5.7% … Clinton lead 2.1%
Trump 30+ states; Clinton 20+ … Trump lead 10
Trump 304* college votes; Clinton 227* … Trump lead 77. *also faithless votes
turnout 60.1%
Change 2020-2024 … Trump; Biden-Harris
Trump +3.4%; Harris -3.1%%; Other -0.31% … Trump lead +6.5% [swing 3.25%]
Trump +6 states; Harris -6 … Trump lead +12 [+6]
Trump +80 college votes; Harris -80 … Trump lead +160 [+80].
Change 2016-2020 … Trump; Clinton-Biden
Trump +0.7%; Biden +3.1%%; Other -3.8% … B&C lead +2.4% [swing 1.2%]
Trump -5 states; Biden +5 … B&C lead +10 [+5]
Trump -72 college votes; Biden +79 … B&C lead +151 [+75.5].
Election step ratios: UK slump and US stumble
Whereas in the UK it had been a slump, in the US it was a stumble. In July the step ratio for the UK Conservatives was 0.54 (=23.7%/43.6%). Last week the step ratio for the US Democrats was 0.94. For the Democrats, Biden in 2020 had 51.3% of the votes and Harris in 2024 had 48.1% of the votes. It was a step difference of -3.2% and a step ratio of 0.94. The UK step ratio of 0.54 was a slump – the USA step ratio of 0.94 was a stumble.
Incumbency: historical and international comparisons
In difficult circumstances, governments find it difficult to govern successfully and so lose popularity and elections. Robert Ford discussed it as did Matthew Parris, citing George Eaton in the New Statesman – who in turn cited Robert Ford … Tim Bale joins in … a piece by John Burn-Murdoch in the FT refers to “the steepest drop in Democratic support since 1980. His graph shows that “every governing party facing election in a developed country this year lost vote share, the first time this has ever happened”. On the other hand the USA was the country that has fallen least this year.
https://www.newstatesman.com/international-politics/2024/11/can-labour-escape-the-incumbency-curse
https://morningcall.substack.com/p/morning-call-the-incumbency-curse
https://x.com/robfordmancs/status/1854214807047860418
https://x.com/robfordmancs/status/1854449264371306714
https://x.com/ProfTimBale/status/1854425142174056949/photo/1
https://www.ft.com/content/e8ac09ea-c300-4249-af7d-109003afb893
.graph: https://x.com/ProfTimBale/status/1854425142174056949/photo/1 [from JBM]
Predictions for 2024 using past results rather than opinion polls
I am interested in ‘predicting’ (retrospectively) the results for this year’s US presidential elections. Rather than using opinion polls, I use past results. It is likely that my predictions will be wrong but it is also likely that the errors in my predictions will be small. The table below gives the following predictions based on simple notions: for all time; no change; combining for all time and no change; one-stage flip; and two-stage flip. An explanation is given in the following section.
All the predictions are within 5% of the 2024 results. The All-time prediction is the best and the 1-stage flip is quite close.
Table Five simple-notion predictions for the 2024 election
All time Republican 49% Democrat 49% Other 2% [1]
No change Republican 46.8% Democrat 51.3% Other 1.5% [2024]
combined Republican 47.7% Democrat 50.2% Other 1.7% [3]
1-stage flip Republican 51.3% Democrat 46.8% Other 1.9% [5]
2-stage flip Republican % Democrat % Other % [4]
Result 2024* Republican 50.6% Democrat 47.9 % Other 1.5% [5]
*as at Friday 15 November 2024
My predictions for 2028 and for all time!
I am interested in predicting the results for future US presidential elections. Rather than using opinion polls, I use past results. It is likely that my predictions will be wrong but it is also likely that the errors in my predictions will be small. The table below gives the following predictions based on simple notions: for all time; no change; combining for all time and no change; one-stage flip; and two-stage flip. An explanation is given below the table.
Table Five simple notion predictions for the 2028 election
All time Republican 49% Democrat 49% Other 2% [1]
No change Republican 50.6% Democrat 47.9% Other 1.5% [2024]
combined Republican 49.8% Democrat 48.5% Other 1.7% [3]
1-stage flip Republican 47.9% Democrat 50.6% Other 1.5% [5]
2-stage flip Republican 51.3% Democrat 46.8% Other 1.9% [4]
For all time
My first prediction is ‘for all time’. It is based on the notion that there is no reason to suppose that one of the two main parties will do better than the other and that other ‘third’ parties will secure only a small percentage of the vote. So I predict:
Republican 49% Democrat 49% Other 2% [1]
Note that the result this year was as given in [2024] below. Using [1] to ‘predict’ [2024] gives the errors indicated in [E]. The highest error was for the Republicans and was quite small at 1.6%.
Republican 50.6% Democrat 47.9% Other 1.5% [2024]
R Error 1.6% D Error 1.1% O Error 0.5% [E]
Note that applying [1] to predict the 2016 result gives a highest error of 1.9% (for the Republicans); and applying [1] to predict the 2020 result gives a highest error of 3.3% (for the Democrats).
No change
My second prediction is ‘no change’ from one election to the next. So my prediction for 2028 is simply that it will be the same as the present 2024 result – see [2024] above.
How well would the ‘no change’ prediction have worked in the past? Clinton in 2016 had 48.2% and Biden in 2020 had 51.3%, a change of 3.1% - and so this amount of error in a ‘no change’ prediction. (The change in Trump 2016 to 2020 was smaller).
: Trump in 2020 had 46.8% and Trump in 2024 had 50.6%, a change of 3.8% - and so this amount of error in a ‘no change’ prediction. (The change in Biden 2016 to Harris 2020 was slightly smaller).
Combining ‘all time’ and ‘no change’
My third prediction combines the all-time prediction with no change to form a weighted average. In the following I take an equal weighting of a half, combining [1] and [2024] to give [3], a prediction for 2028.
Republican 49.8% Democrat 48.5% Other 1.7% [3]
Note that using a weighted average model to predict the 2020 and 2024 results gives highest errors of 3.2% (for the Democrats) and 3.2% (for the Republicans), respectively.
Two-stage flip
Historically it is common for presidents or their party to have two terms of office (and no more). There is a corresponding two-stage flip in the percentage from one party to the other party. My prediction for 2028 is based on the 2020 result when Biden won. Flipping this result for 2028 gives
Republican 51.3% Democrat 46.8% Other 1.9% [4]
Using the 2016 result and flipping it to give the 2024 prediction gives a highest error of 2.4% (for Trump).
One-stage flip
However since Obama’s two terms in office there have been two one-stage flips. Using the 2024 result and predicting a one-stage flip for 2028 we predict the following result for 2028:
Republican 47.9% Democrat 50.6% Other 1.5% [5]
Using a one-stage flip to predict 2020 and 2024 gives highest errors of 5.2% (for Trump to Biden) and 1.1% (for Trump to Harris), respectively.
Note: Counting was still continuing while people were criticising the opinion polls for their error. The last few countings did not change the picture much but they did very slightly reduce the size of the error.
Predicting the opinion polls in 2028 and for all time!
Paradoxically we can use the predictions of the results to predict the opinion polls! Basically the polls are equal to the results plus some error. Since the 1960s the error has varied with a median of somewhat below 2% biased in favour of the Democrats.* Using this an all-time prediction for the opinion polls is:
All time Republican 47% Democrat 49% Other 4% [6]
Commentary, opinion polls and election results
What were the election results? What were the opinion polls and how did they relate to the election results? What was the commentary and how did it relate to the results and to the opinion polls?
The election results. Writing this on Sunday 10th, the final results are not yet known. Wikipedia is reporting, based on on 95% of the vote, that Trump has 50.3% and Harris has 48.1%, a lead of 2.2% for Trump.
Opinion polls. Five polls reported on Monday 4 November. Three gave a win for Harris by 2%, 2% and 4%; one gave a win for Trump by 1%; and one gave a tie. The average of the five polls is 1.4% for Harris [“538” put the Harris lead at +1.2%] … so the five polls are in error by 3.6% in Harris’ favour – in terms of the results so far.
Inference. Going back before the election, what inference might we make on the basis of the opinion polls?
Links
Polls: https://www.realclearpolitics.com/epolls/latest_polls/national_president/index.html
https://projects.fivethirtyeight.com/polls/president-general/2024/national/
… Wikipedia:
graph July 2024 on
graph 2020 on
Election results:
https://en.wikipedia.org/wiki/2016_United_States_presidential_election
https://en.wikipedia.org/wiki/2020_United_States_presidential_election
https://en.wikipedia.org/wiki/2024_United_States_presidential_election
My commentary, Thursday 30th October, 2024
[The median of the seventeen polls is a tie. Trump’s actual lead was 2.2%. So the error was 2.2%. Six of the seventeen polls gave Trump a lead of +2% or +3%.]
“Trump v Harris 2024; opinion and conflict at home and abroad
GO TO Trump: Mass, Space and Time in USA Elections, 1789-2020
GO TO Trump: the Abstract Structure of Public Opinion, USA 2020
Negative discourse: Trump [supporter] says ‘garbage’ then Biden says ‘garbage’. (BBC)
Saturday October 19th to Tuesday October 29th, 2024:
17 polls, Trump ahead in 6; Harris ahead in 8; tie in 3:
T / - / H / H H / H T / T T / = = / = / H H / T / H H T
+3 /-/ -1 / -4 -3 / -3 +2 / +3 +2 / 0 0 / 0 / -1 -4 / +2 / -3 -1 +2”
https://www.realclearpolitics.com/epolls/latest_polls/national_president/index.html
USA Today, the seven swing states:
BBC: https://www.bbc.co.uk/news/topics/cj3ergr8209t
My commentary, Thursday 7th November, 2024
2 A brief note on the results: elections - the small difference hypothesis
A pattern which is sometimes observed in elections is that there is a “landslide” for the winning party which gives them a “mandate”. In the UK the Labour Party enjoyed a landslide in July. This week Donald Trump enjoyed a landslide in the USA. In the UK the landslide was in terms of seats in parliament. In the USA the landslide was in terms of college votes. We might refer to these as election system landslides.
However election system landslides do not always correspond to percentage individual votes – and may indeed produce a winner different from the candidate with the highest percentage vote. This week Trump gained 50.8% of the vote and Harris 47.5%, Trump ahead by 3.3% [these are the earlier figures, not the finalresults]. This rather small difference in the individual votes was associated with a large difference in the college votes: 312? for Trump and 226? for Harris, a difference of 86. The small percentage difference has been magnified to become a large college vote difference.
A common question is how well the opinion polls do. Critics, I feel, sometimes do not conceptualise the situation appropriately:
“How did the pollsters get it so wrong again? … the polls had “predicted the closest race in modern times – instead it was a rout.” The Times, November 7, 2004, 8-9.
Hugh Tomlinson
“The average polling error seen in these swing states was not actually that big. Still, in tight campaigns, small changes can make a big difference.”
https://www.bbc.co.uk/news/articles/cj4ve004llxo
To judge Tomlinson’s observation we need to attach numbers to words such as “close” and “rout”. The race was 3.3% “close” in terms of percentage vote and an 86 “rout” in terms of college votes. Mostly the polls dealt with the percentage vote. Different polls gave different leads, ranging in one case between +3% for Trump to +4% for Harris with a median of 0% - see below. So the median had an error of 3.3%. The range of poll percentages included the actual result percentage 3.3%.
THE END