Thoughts from a neo-psephologist

Until now, I’ve never really been in favor of proportional representation in elections. But as I get older (though probably not a lot wiser) I’m coming round to the idea, and electoral reform in general (not only in the UK but elsewhere). The UK’s First Past The Post (FPTP) electoral system is no longer fit for purpose. It’s not as though we’ve never had a stab at proportional representation. Elections to the European Parliament were run in this way.

So what has brought about this Damascene experience? Well, you only have to examine the consequences of the 19 December 2019 General Election here in the UK or the recent presidential election in the USA to realise that something is rotten in the state of Denmark (Hamlet, Act I, Scene IV). The current parliamentary makeup is not serving the people adequately here.

I’m surely not the only person who feels that the current Boris Johnson-led Conservative Government is the most inept, corrupt even, of any government they have had to live under. During my lifetime (I’m 72), the UK has had fifteen Prime Ministers (Harold Wilson served twice; there was a gap of almost four years between his first administration ending in June 1970, and returning to power in March 1974), eleven were Oxford educated, one at Edinburgh, and the other three (including Sir Winston Churchill) did not go to university.

Without a shadow of a doubt, in my opinion, classics scholar (a term I deploy advisedly) Boris Johnson is in a league of his own as perhaps the worst Prime Minister of the lot. I admit that my Twitter feed is full of tweets from like-minded individuals. And cronyism is definitely on the rise during this Covid-19 pandemic, as analysis of the award of contracts, for example, to provide personal protective equipment (PPE) has clearly indicated. What I find hard to understand is why Johnson isn’t doing worse in the polls.

Opposition parties in Parliament are there to hold the government of the day to account. But with an overall majority of more than 80, this Tory government is essentially unassailable. Yes, it has had a few wobbles when Eurosceptic Tories have voted against their own party. But with Brexit [1] out of the way, so to speak, Johnson and his cohorts essentially have unlimited licence over the next four years until the next mandated General Election to do whatever they like. And we should all be worried about that.

Taking the UK out of the European Union has already eroded a number of significant rights and privileges that membership gave all citizens of the UK. I simply don’t trust Johnson to legislate for the greater good.

So, let’s look at the last General Election.

Voter turnout was greater than 67% (of a registered electorate of more than 47.5 million). I don’t claim to have access to a significant amount of data or to be anything like an expert. These are just some of my observations that reflect my concerns about electoral reform.

Of the 650 seats in the House of Commons, the Conservatives won 365, on a 43.6% share of the votes cast.

That means that more than 56% of the voting public supported parties other than the Conservatives. Labour’s share was 32.1%, giving them only 202 seats. The next biggest party (with just 3.9% of the national vote) was the Scottish National Party (SNP) with 48 seats, all in Scotland of course. Scotland is now an SNP monopoly after winning just a 45% share of the votes across the 59 Scottish constituencies. The Greens attracted 2.7% of the national vote but gained just a single seat. As for the Liberal Democrats, the situation was even more dire: 11.6% share of the votes resulting in only 11 seats in Parliament. No wonder the Lib Dems have long advocated a change to proportional representation.

If seats were allocated based on their share of the vote, the Conservatives would have just 283, Labour 208, and the Lib Dems, 75. I voted Lib Dem at the last election, but it was essentially a wasted vote, as would have been a vote for the Labour candidate in my constituency at the time, Bromsgrove in Worcestershire, that was retained by former Chancellor of the Exchequer Sajid Javid, who retained his seat with a slightly increased share of the votes cast, at 63.4%.

Now in constituencies that have long enjoyed domination by one party or another, such as Conservative Bromsgrove for example or Labour-held Knowsley on Merseyside (with an almost 40,000 majority, >80% of votes cast), proportional representation is hardly likely to change that sort of result. However, where the number of votes cast per candidate is more evenly spread, and where the FPTP winner actually has a minority share of the vote, then proportional representation is going to have a much more significant effect.

How the constituencies could be re-designated to better reflect current demographics I’ll leave to others better qualified to propose. But I do believe that ‘voting areas’ should be larger than the current constituencies, say counties with each’county’ returning the same number of MPs as they do in total now. But for each there could be a slate of candidates, and the seats would be allocated by the total number of votes cast per political party (similar to how the MEP elections were held in the past). There needs to be a thorough discussion about the actual system of proportional representation, and I’m not qualified to comment on that particular aspect.

I do feel strongly that we need a House of Commons that better reflects how the UK population votes. FPTP does not do that, and given the increasing polarization in political stances and viewpoints, I think we need a more nuanced approach to policy development and implementation. Yes, I appreciate that proportional representation is likely to lead to more coalition governments. Is that such a bad thing? I personally think that the Lib Dems were right to go into coalition with the Conservatives after the 2010 General Election. I don’t think they had much choice given that the country was trying to rebuild itself following the 2008/2009 financial crash.

Northern Ireland First Minister (of the DUP) Arlene Foster and then Prime Minister Theresa May after the June 2017 election.

Coalitions do come with disadvantages, however as seen in some countries that take months to form new coalition governments. Small (and maybe even extreme) parties can hold the balance. Take the religious parties in Israel, or more recently in the UK where the Northern Ireland Democratic Unionist Party (DUP) entered into a ‘confidence and supply’ deal with the Conservatives following the 2017 General Election that saw then Prime Minister Theresa May lose her majority, and therefore needed the backing of a ‘friendly’ party to keep Corbyn’s Labour at bay.

Given the current state of politics in the UK I believe the call for electoral reform will become a clamour in the not-too-distant future. Or maybe it’s just wishful thinking on my part.


Let me turn my attention to what’s been happening on the other side of the Atlantic.

Can you imagine that American politics would ever come to this? An incumbent President defeated decisively in a general election then, even more than two months on, not accepting that defeat, and going as far as trying to subvert the outcome.

From my UK perspective, the USA seems to have a crazier electoral system than we ‘enjoy’ over here. A House of Representatives that is elected every two years (with all the financial dangers of corruption to remain in power), gerrymandering across the country (especially in Republican-held districts), the billions of dollars that are raised and spent on political campaigns, and an election of the President every four years that does not take account directly of the popular vote.

Given the role of the Electoral College, election campaigns will always focus primarily on those so-called battleground states that ultimately give the winning candidate the 270 votes needed in the Electoral College to win the race.

Let’s look at the results of the 2016 General Election in the US, won by Trump in the Electoral College by 304 votes to 227, even though Hillary Clinton won the popular vote by almost 2.9 million votes.

Here is a series of maps that show the 2016 FPTP election results for President, county by county. It’s a sea of Republican red, right across the country, but with significant Democrat concentrations on the East and West Coasts, and some parts of the Mid-West.

But does that map reflect the distribution of party allegiances? Since the USA is essentially a two party nation, Republicans and Democrats, it’s straight forward to provide a rather more nuanced visualization of how everyone voted, with shades of purple reflecting the proportion of votes for each party. (This sort of map would be harder to compile for UK election results, since there were nine parties contesting the 2019 election, albeit some were regional parties like the SNP or DUP).

Even better perhaps is the same map, county by county, that shows the votes based on population, as its author stated: ‘Land doesn’t vote. People do.’ Check the visualization here. The Republican Party is primarily rural, and in those states and counties with  rather low population densities.

It’s incredible that two months on from last November’s election, which Joe Biden won with 51.4% of the popular vote (and a margin of more than 7 million votes) that Trump is still trying to game the system. Perhaps even more incredible that Trump himself won more than 74 million votes. A country divided!

This result gave Biden 306 votes to Trump’s 232. And, since he hates losers, Trump just cannot accept that he lost the election. And keeps ranting on about it.

The Electoral College does, in the 21st century, seem an anachronism. If the votes for Arizona (11), Wisconsin (10), Michigan (16), Pennsylvania (20), and Georgia (16) are discounted, then Biden and Trump would have essentially the same number of college votes, 233 to 232. No wonder Trump is futilely trying to overturn the results from these states. If just over half of the people that voted for Biden in these five states had voted the other way, Trump would remain President. That means the election was essentially determined on just under 140,000 votes. From a popular vote of over 155 million (the highest turnout in over a century), to have an election resolved by less than 0.1% of those who voted seems a shaky basis for electing someone to ostensibly the most powerful office in the world.

Trump can cry foul at every turn, that the election was stolen from him, that the Democrats cheated, the election was a fraud. Funny how fraud only occurred in states that the Democrats won. This had crossed my mind several times. Today I saw it articulated publicly. Not sure who this is. I recognise the face but can’t put a name to it. I’m sure someone will enlighten me.

We think that Johnson and his pals have brought the UK into disrepute with their handling of Brexit and the Covid-19 pandemic. Media in the EU are openly mocking this government. In the same vein, Donald Trump has eroded respect for the USA globally. Although I’m not sure the MAGA Trumpists see it that way. Poor misguided fools . . .


[1] The 2016 Brexit referendum was won by the Leave campaign on 52% of the votes cast (but only 37% of the electorate). The FPTP system really failed us on this occasion, in my opinion. For something that had such constitutional, financial, social, and political consequences the referendum rules should have been tighter. I have long argued that not only should there be a minimum turnout (it was actually quite high at 72%), but that the winning margin needed to be 50% +1 of the persons eligible to vote, not those that actually voted. We have been forced to leave the European Union on the whims of less than 40% of the electorate, a substantial number of whom now say they regret having voted that way knowing now what they didn’t then, when they were promised ‘unicorns’ and ‘sunlit uplands’.


 

There’s beauty in numbers . . .

Now, what I want is, facts . . . Stick to the facts, sir!

Thus spoke businessman, MP, and school superintendent Thomas Gradgrind in the opening paragraph of Charles Dickens’ tenth novel, Hard Times, first published in 1854.

Increasingly however, especially on the right of the political spectrum, facts have become a debased currency. ‘Alternative facts’ and ‘fake news’ have become an ‘alternative religion’, faith-based and not susceptible to the norms of scientific scrutiny. Fake data are also be used as a ‘weapon’.

I am a scientist. I deal with facts. Hypotheses, observations, numbers, data, analysis, patterns, interpretation, conclusions: that’s what science is all about.

There really is a beauty in numbers, my stock-in-trade for the past 40 years: describing the diversity of crop plants and their wild relatives; understanding how they are adapted to different environments; how one type resists disease better than another; or how they can contribute genetically to breed higher-yielding varieties. The numbers are the building blocks, so to speak. Interpreting those blocks is another thing altogether.

Statistical analysis was part and parcel of my scientific toolbox. Actually, the application of statistics, since I do not have the mathematical skills to work my way through the various statistical methods from first principles. This is not surprising considering that I was very weak in mathematics during my high school years. Having passed the necessary examination, I intended to put maths to one side forever, but that was not to be since I’ve had to use statistics during my university education and throughout my career. And playing around with numbers, looking for patterns, and attempting to interpret those patterns was no longer a chore but something to look forward to.

So why my current obsession with numbers?

First of all, since Donald Trump took up residence in the White House (and during his campaign) numbers and ‘alternative facts’ featured prominently. Trump does not respect numbers. However, more of this later.

Second, I recently came across a scientific paper about waterlogging tolerance in lentils by a friend of mine, Willie Erskine, who is a professor at the University of Western Australia (although I first knew him through his work at ICARDA, a CGIAR center that originally had its headquarters in Aleppo, Syria). The paper was published last month in Genetic Resources and Crop Evolution. Willie and his co-authors showed that lentil lines did not respond in the same way to different waterlogging regimes, and that waterlogging tolerance was a trait that could be selected for in lentil breeding.

A personal data experience
While out on my daily walk a couple of days later, I mulling over in my mind some ideas from that lentil paper, and it reminded me of an MSc dissertation I supervised at The University of Birmingham in the 1980s. My student, Shibin Cai, came from the Institute of Food Crops, Jiangsu Academy of Agricultural Sciences, China where he worked as a wheat scientist.

Cai was interested to evaluate how wheat varieties responded to waterlogging. So, having obtained several wheat lines from the International Maize and Wheat Improvement Center (CIMMYT) in Mexico, we designed a robust experiment to evaluate how plants grew with waterlogging that was precisely applied at different critical stages in the wheat plant’s life cycle: at germination, at booting, and at flowering, as far as I remember. I won’t describe the experiment in detail, suffice to say that we used a randomized complete block design with at least five replicates per variety per treatment and control (i.e. no waterlogging whatsoever). Waterlogging was achieved by placing pots inside a larger pot lined with a polythene bag and filled with water for a definite length of time. Cai carefully measured the rate of growth of the wheat plants, as well as the final yield of grains from each.

After which we had a large database of numbers. Observations. Data. Facts!

Applying appropriate statistical tests to the data, Cai clearly showed that the varieties did indeed respond differently to waterlogging, and we interpreted this to indicate genetic variation for this trait in wheat that could be exploited to improve wheat varieties for waterlogging-prone areas. I encouraged Cai to prepare a manuscript for publication. After all, I was confident with the quality of his research.

We submitted his manuscript to the well-known agricultural research journal Euphytica. After due process, the paper was rejected—not the first time this has happened to me I should add. But I was taken aback at the comments from one of the anonymous referees, who did not accept our results—the observations, the data—claiming that there was no evidence that waterlogging was a verifiable trait in wheat, and especially in the lines we had studied. Which flew in the face of the data we had presented. We hadn’t pulled the numbers like a rabbit out of a hat. I did then wonder whether the referee was a wheat expert from CIMMYT. Not wishing to be paranoid, of course, but was the referee biased? I never did get an opportunity to take another look at the manuscript to determine if it could be revised in any way. As I said, we were confident in the experimental approach, the data were solid, the analysis sound—and confirmed by one of my geneticist colleagues who had a much better grasp of statistics than either Cai or me. Result? The paper was never published, something I have regretted for many years.

So you can see that there were several elements to our work, as in much of science. We had a hypothesis about waterlogging tolerance in wheat. We could test this hypothesis by designing an experiment to measure the response of wheat to waterlogging. But then we had to interpret the results.

Now if we had measured just one plant per variety per treatment all we could have said is that these plants were different. It’s like measuring the height say of a single plant of two wheat varieties grown in different soils. All we can state is the height we measured. We can make no inference about any varietal differences or responses. For that we need several measurements—numbers, data—that allow us to state whether if any observed differences are ‘real’ or due to chance. That’s what we do all time in science. We want to know if what we measure is a true reflection of nature. It’s not possible to measure everything, so we use a sample, and then interpret the data using appropriate statistical analyses. But we have to be careful as this interesting article on the perils of statistical interpretation highlights.

Back to The Donald
One of the most important and current data relationships is based in climate science. And this brings me back to The Donald. There is an overwhelming consensus among scientists that relationship between increased CO2 levels and increases in global temperatures is the result of human activity. The positive relationship between the two sets of data is unequivocal. But does that mean a cause and effect relationship? The majority of scientists say yes; climate deniers do not. That makes the appointment of arch-denier Scott Pruitt as head of the Environment Protection Agency in the US so worrying.

Donald Trump does not like facts. He doesn’t like numbers either unless he can misappropriate them in his favor (such as the jobs or productivity data that clearly relate to the policies under Mr 44). He certainly did not like the lack of GOP numbers to pass his repeal of the Affordable Care Act (aka Obamacare).

He regularly dismisses the verifiable information in front of his eyes, preferring ‘alternative facts’ and often inflated numbers to boot, instead. Just remember his sensitivity and his absurd claims that the 20 January National Mall crowds were largest for any presidential inauguration. The photographic evidence does not support this Trumpian claim; maybe fantasy would be a better description.

Time magazine has just published an excellent article, Is Truth Dead? based on an interview with The Donald, and to back it up, Time also published a transcript of the interview. This not only proves what Mr 45 said, but once again demonstrates his complete lack of ability to string more than a couple of coherent words together. Just take a look for yourselves.

Part of Trump’s rhetoric (or slow death by Tweet) is often based on assertions that can be verified: the biggest, the longest, the most, etc. Things can measured accurately, the very thing he seems to abhor. His aim to Make America Great Again cannot be measured in the same way. What is great? Compared to what or when? It’s an interpretation which can be easily contradicted or at the very least debated.

That’s what so disconcerting about the Trump Administration. The USA is a scientific powerhouse, but for how much longer if the proposed agency budget cuts that The Donald has promised really bite (unless related to the military, of course). There’s an increasing and worrying disdain for science among Republican politicians (and here in the UK as well); the focus on climate change data is the prime expression of that right now.

 

Lies, damned lies, and statistics

Lies, damned lies, and statistics – a saying popularised by Mark Twain who attributed it to Victorian Prime Minister Benjamin Disraeli; but there are others. It generally refers to the bolstering of weak arguments with statistics. And you only have the watch the news each day to see how sloppily statistics are used, often by politicians.

I was crap at mathematics at school – only just scraping a pass in my GCE ‘O Level’ examination at the age of 15, whereupon I dropped the subject completely afterwards. But I love playing with numbers – data, especially data I have generated myself through my own research. There’s just so much information to mine in data sets, looking for patterns that throw up lots of different questions, hypotheses even.

Not everyone sees it that way, however. I remember having an argument with one of my students – I can’t remember if it was an undergraduate who had completed a final year honours project with me, or one of my postgraduates. But we disagreed about how best to present data in tabular format. When this student handed me a draft, I asked some pertinent questions about the data and what she thought  they indicated. The student was not able to answer with any conviction. I suggested she should reorganize the data in several ways to see if any patterns emerged. ‘You can’t do that’, she retorted, ‘you are placing a bias on the data’. ‘Humor me’, I asked her, and she duly made the adjustments I had suggested. Lo and behold, she was able to detect a number of patterns, relationships even, that had not been apparent in her ‘random’ tabulation of data. Now, it was still necessary to undertake appropriate statistical tests to see if the relationships she observed were cause and effect, so to speak, or had occurred merely by chance.

But I think this example just highlights how much information can be ‘hidden’ in data sets.

And one man, who is passionate about statistics, is on a mission to make statistics meaningful for everyone. He’s Professor Hans Rosling, a global health expert and Professor of International Health at the Karolinska Institute in Sweden. Last night I watched a highly entertaining – and illuminating – one hour program on BBC4 titled The Joy of Stats. So enthusiastic is he to uncover the hidden messages in data sets, he’s set up an organization called Gapminder Foundation that aims to visualise data in a way that teases out lots of the underlying detail.

In the video below (taken from the BBC4 program, both of which are freely available on the Gapminder web site), and using some 120,000 data points, Rosling tracks the relationship of life expectancy and income in 200 countries over 200 years.

Rosling is now a highly acclaimed speaker at conferences around the world, especially at TED meetings (Technology, Entertainment & Design). He emphasises the need to access publicly funded databases, to link them, search them, and bring the underlying data messages to the surface, often in the face of those who curate the databases and tell him it’s not possible.

In my own field of agricultural research, Rosling has, in Gapminder Agriculture, taken 700 indicators on production of crops, livestock, etc. from the Food and Agriculture Organization.

As I said, Rosling is a man on a mission – but a highly worthwhile and innovative one. Do take time to watch the videos. Your patience will be rewarded.