David Slater November 4, 2020
Oh no, not another abysmal failure to predict accurately, the result of a crucially important vote, by our self-promoting polling “experts”? Oh yes! It’s a pattern, it’s the norm, polling by these clipboard data manipulators doesn’t work! And the reason should be obvious. The basis of these surveys is at best naïve and at worst can be less than objective – (what numbers / answers did you have in mind sir?). From the analysts’ point of view, all they need are sufficient returns (expressed as numbers) from different categories of people, selected at random from overall statistics. To them is put a set of precooked, (designed) set of standard questions. These are fired at their “representative samples”, to obtain “statistically meaningful” pictures of our intentions; to predict how we (like sheep?), will behave.
But although we’re all created equal, we pride ourselves on being individuals. Ask Bob Pointer, if he was “effectively communicating” with another individual, would he doorstep them and demand answers to an intensely personal set of questions, to which the real answers may be personal, private, even embarrassing; perhaps reflecting a defensive, or contrived persona we desire to project. In general, in this situation, people tend to be polite and give the answers that they perceive are required and / or are politically and socially correct. In the security and anonymity of the voting booth, we may well vote with our heart and not our scheming minds and do precisely the opposite. Perhaps because we can. This is one of the reasons the UK Parliament insists on MP’s voting activity being exposed, public and open to sanction (is this democratic?).
For example, in private, one could feel embarrassed about voting for posing, populist demagogues, or for personally advantageous laws (could be!), and publicly tell the pollsters (or “whips”), otherwise. Similar arguments apply to the methods used by naïve economists publicly polling consumer preferences – of course I always buy the best! How much would I pay? Oh, lots of money, I’m very successful!
As long as the only people out of pocket are the self-interested political parties, or commercial companies that fund them, there is little damage; except to their credibility and possible career prospects. Again, offering compensation for the imposition and intrusion, merely adds pressure to provide the “correct” or popular response.
But this “lazy” reduction of people to statistical categories and academic hypotheses, is, I suspect, one of the major reasons we are not coping well with the COVID 19 pandemic. The virus remains fairly constant as a biohazard (although constantly mutating); in chemical composition and mode of infectivity and toxicity. But even though the populations affected are composed of millions of individuals, no two are identical and averages for age, etc. are at best “opinion polls” as to expected effects.
Thus, managing infection spreading, by using lagging, lumped indicators such as nationally averaged R numbers, is proving less than effective; especially if addressed using blunt, one size fits all, national measures. We have got to be more intelligent and more effective. We must be smarter, more targeted, more focussed on understanding the patterns of individual vulnerabilities and social situations. Just as we now, at last, have recognised that the wearing of face masks to try and stop the virus spreading at source, is a more effective and better precaution, than dismissing their effectiveness by academically consulting global statistics on past observations on average viruses, in average conditions, using ancient datasets whose applicability at best is questionable.
People matter; and people are individuals not statistical samples. The sooner we get to grips with understanding what makes us individual and different, the sooner we may get more intelligent predictions of how we may react in different scenarios. There are no short cuts and statistics are only a guide. We need to understand the uncertainties inherent in diversity and individuality.