Last night I stayed up too late because I was paying attention to British election results. If it is not odd enough that I am a political junkie of my own national politics, and institutional politics, I must freely own that I am a political junkie in general, even about areas that have no jurisdiction over me and in which I have no personal involvement [1]. Such is the case with British politics, and I refreshed the BBC live election coverage every minute or two to see the map of Great Britain fill up in an odd selection of colors. One could see the yellow of the Scottish Nationalist Party sweep nearly the entire country, and see the tendrils of Conservative dominance fill large parts of Wales and England, with isolated redoubts of orange from a Liberal Democratic party that got crushed, losing 2/3 of its electoral support and even more of its seats. About midnight or so, it became clear that the Conservatives were likely to gain a majority, and there were only about 50 seats or so left that had not declared a winner, and so I went to bed having satiated, for at least one night, my love of pondering and examining political choices.
There is a striking aspect to this election, and that is that the pundits nearly universally declared that it would be a hung parliament that may not even be able to cobble together a ruling coalition. Yet the conservatives won a majority with several seats to spare, not even needing to join with the unionist parties of Northern Ireland (which won 10 seats between them) or join up with the Liberal Democrats who were their previous coalition partners. Such a majority would seem unthinkable looking at the polling, which showed a Labour lead for most of the campaign and a tie vote for the past few weeks. Instead, the Tories won the election by 6.5% and increased their number of seats by 25. How did this happen? This is all the more notable because this is a regular pattern. The exit polls underrepresented the Tory win by about 15 seats (about half of their margin of victory), only doing slightly better than the majority of polls that were way off. Yet this is a regular mistake. The under-representation of the Conservative margin by 6% was not too far from the 1992 election’s under-representation of the Conservative vote haul by 8%. When polls show a systemic bias, the conclusions made by people on that bogus data will be consistently wrong.
As someone who works and cares a great deal about the integrity of data, this is a big deal. The systematic under-representation of the Conservative vote in England is called the “Shy Tory” effect and even if it is recognized, every election the talking heads on television and the commentators for websites, including a few Americans, make the same broad conclusions about hung Parliaments and the possibility of a second election only to be endlessly surprised by a larger than expected Conservative vote. When something happens so often, it ought not to be a surprise when it happens. Rather, it is evidence that there is a problem in the underlying validity of polling statistics that merits high levels of concern. Admittedly, the arcane nature of British politics is not an area of personal expertise, for although I am a political junkie, I am not someone whose life is immersed in the particular structure and order of the British political system.
That said, though, there are at least several issues that even an untrained and unskilled observer like myself can notice immediately. For one, the British electoral system (like that of the United States) is not a single election but rather a series of 650 campaigns, one in each constituency. The overall percentage vote matters less than the results of those constituencies aggregated together. Such a system tends to reward regional parties, like Northern Ireland’s Unionist and Sinn Fein parties, the Welsh Plaid Cymru, or the Scottish Nationalist Party, none of whom are considered in the polling questions, despite the fact that all of them have a strong regional base. When you consider that polls tend to be taken with a supposedly representative sample but also routinely ignore these regional parties, and ignore the fact that Parliamentary elections are 650 simultaneous small elections rather than one big one, it is clear that the polling is going to be subject to some gross systematic errors because of a failure in structure. In stark contrast, it should be noted, the American political system has become skilled at the micropolling in at least most of the marginal seats so that within a fairly tolerable boundary, the general trend of most elections is fairly stable, at least on a constituency by constituency level. It should be noted though, that in the last two elections, if not longer, there have been serious discrepancies between the overall polling nationwide and the actual result, signifying that here too it is the results that must be aggregated, and not the overall numbers. Yes, this requires more detailed work, but accuracy in the detailed work is the difference between being right and being embarrassed for predicting a Labour victory or draw in an election where Labor lost by 99 seats.
Why would a Tory be shy in the first place, though? The Conservatives have generally done well in the polls, and except for the period of Tony Blair’s success in charge of a rather Clintonesque “Third Way” Labour in the 1990’s and 2000’s where Labour abandoned doctrinaire socialism, the Tories have been the dominant party since 1980 or so. A comparison with America’s political and cultural system may be notable here. After all, there is a marked left-wing bias in the American media, one which continually under-represents the popularity of conservative positions and candidates. Even the perception of such a bias, where it does not exist, would be enough for people to not mention their actual beliefs, for fear of ridicule or disapproval. In ironic ways, the success of the left wing in dominating so-called “mainstream” journalism has led to their inability to see reality for what it is, since those who oppose them and their cultural agenda often simply refuse to show themselves except in the privacy of the voting booth, leading to endless weeping and gnashing of teeth when the results of the cold hard data of votes does not align with the world of polls and models that are based on systemically flawed data, data that is flawed at least in part because of a lack of trust in the respect that the media has for the ordinary and somewhat shy Tory (or Republican) voter. How are institutions to build trust, so that they may be able to see the world as it is, rather than see it through their own biased perspective, a bias that is noted by others, even if silently and with a stiff upper lip of reserve and privacy. Most people, myself included, know when our opinions and thoughts are not valued and respected, and the general response is to be silent, except where we know we may find support, which undermines those who would presume to speak for everyone without having first shown that they are willing to listen to what others really have to say.
[1] See, for example:
https://edgeinducedcohesion.wordpress.com/2011/07/04/landslide/
https://edgeinducedcohesion.wordpress.com/2015/04/27/book-review-all-things-through-christ/
https://edgeinducedcohesion.wordpress.com/2015/02/18/you-made-me-promises-promises/
https://edgeinducedcohesion.wordpress.com/2014/08/23/book-review-the-greatest-comeback/
Edge Induced Cohesion 
Pingback: Some Minutes Of The SQL Brown Bag Lunch #3 | Edge Induced Cohesion