If it looks too good to be true, it probably is

Be wary of analysis where all the indicators seem to point in the same direction: “If it looks too good to be true, it probably is”.

Again

306 – 232

Is the number everyone is focused on.

Yes Biden/Harris seems to have a clear mandate, with almost 5 million more votes than Trump/Pence.

The numbers who voted in this election are unprecedented – Trump/Pence lost with the highest number of votes for any incumbent presidential ticket.

The efforts of the democrats to push people, especially minorities and disenfranchised people to vote paid off.

Well…

Not in the way you may think. According to CNN exit polls (1):

Trump/Pence did better in 2020 among all non-white-races (vs 2016), 

and 

Biden/Harris (2020) worse among all non-white races compared with Clinton/Kaine (2016).

Some of you would have noticed that the changes do not add up to zero (loss of democrats is smaller than gains to Trump/Pence). This is because people are more candid about their choices this round:

Fewer people voted for a third party or refused to reveal their preference.

So, how did Biden/Harris apparently win more votes then?

The Democrat campaign was focused on re-capturing the people who voted for Obama/Biden in 2012 and who switched to Trump/Pence in 2016, especially in the rust-belt (2).

And this strategy seems to have delivered the white house (3)(4).

Data can show much more if you have an open mind and look deeper.

 

 

1 https://edition.cnn.com/election/2020/exit-polls/president/national-results

2 https://www.reuters.com/article/us-usa-election-biden-insight/bidens-winning-strategy-flip-rust-belt-trump-states-and-hold-on-tight-idUSKBN27N0OC

3 https://slate.com/news-and-politics/2020/11/midwest-rust-belt-georgia-will-decide-presidential-election.html

4 https://dyn.realclearpolitics.com/elections/live_results/2020/president/

 

Data Literacy is more important than ever

Data has been sitting in servers and databases for the last 50 years.

Artificial Intelligence has been around for at least that long too.

But data, on its own, does not inform you, does not help you make better decisions, it needs to be interpreted, modelled…

A simple example:

290 -214

This is the current number of votes from the electoral college that Biden-Harris and Trump-Pence are projected to receive. Since a minimum of 270 votes are the minimum to be declared the winner, Mr Biden finds himself called president-elect.

But there is more to this:

The fact remains that close to 8 million people more voted from Trump-Pence in 2020 (as of now) compared to 2016. The Trump-Pence ticket has grown in votes.

Which number you choose to focus on depends on the story you want to tell. A truer picture should use both sets of numbers.

Analytics is not about the data; interpreting, modelling the data, being aware of the weaknesses of modelling techniques is critical.

Data literacy is more important than ever.