Pink and White Terraces Christmas – the Annual Ordeal Alternative Medicine ‘Journal’ Water Treatment Woo CUSP Podcast Turns Four
Total Page：16
File Type：pdf, Size：1020Kb
How lumpy is random? Pink and White Terraces Christmas – the annual ordeal Alternative medicine ‘journal’ Water treatment woo CUSP podcast turns four number 110 – summer 2014 content editorial How lumpy is random? 3 The Pink and White Terraces: Forwards and still lost? 7 Newsfront 10 backwards Christmas: the annual ordeal 12 ND so another year begins, and as I write this on New Year’s A journal, but not as Day 2014 there is the opportunity, as with every new year, to we know it 15 A reflect on past years and consider the prospects for the future. 2014 Anti-ageing water from will no doubt be an especially busy year for recollections and com- your tap 16 memorations, marking as it does the centenary of the start of World On the CUSP 18 War I. Few could have had any idea, on that New Year’s Day of a century ago, of what the next few years would bring. Forum 20 Book Review: It’s always fun at this time to look back at what predictions the Salt Sugar Fat 21 psychics have made for the previous year. RelativelyInteresting. com has a good summary of their efforts for 2013. Among the In which we encounter a very more entertaining were that Prince William and Kate would have a strange idea about water 22 daughter “whom many will believe is the reincarnation of Princess Diana”, experimental monkeys would escape from a lab and cause a pandemic, and meditation would be proved to be the gateway to contact loved ones on the other side. As the Danish proverb says, ISSN - 1172-062X predictions are difficult, especially about the future. Contributions Looking a bit further back, 10 years ago (2 January 2004) I took part in a successful hunt for the supposedly extinct New Zealand Contributions are welcome and should be sent to: storm petrel, not reported for more than a century but reliably sighted a couple of months previously less than 100 km from downtown David Riddell Auckland – a truly unexpected development. In the intervening 122 Woodlands Rd decade live birds have been captured, their DNA analysed, and just RD1 Hamilton Email: [email protected] last year their breeding grounds were discovered on Little Barrier Island. Deadline for next issue: March 10 2014 The Summer 2004 NZ Skeptic editorial (by Annette Taylor, also on the trip) commented that the storm petrel’s story was very different Letters for the Forum may be edited from those of other elusive creatures such as moa, lake monsters, as space requires - up to 250 words is preferred. Please indicate the or Bigfoot – or, for that matter, the supposed ‘panthers’ in the South publication and date of all clippings Island. The NZ Herald website had another article about this mystery for the Newsfront. beast on 22 October 2013, photographed walking across the ice of Lake Clearwater in August. The photo looks quite dramatic, with Material supplied by email or CD is appreciated. a distant feline silhouette sharply defined against the icy landscape and no obvious visual cues to help assess size, but when the animal Permission is given to other non- is magnified and examined in isolation it’s clearly a feral domestic profit skeptical organisations to cat. reprint material from this publication provided the author and NZ Skeptic are acknowledged. Opinions expressed in the New Zealand Skeptic are those of the individual authors and do not necessarily represent the views of It’s a fairly safe prediction that 2014 will see more sightings of NZ Skeptics (Inc.) or its officers. South Island panthers, but no actual specimens. What else the year may hold is anyone’s guess. number 110 – summer 2014 main feature How Lumpy is Random? and other burning questions in quantitative reasoning David Bulger Most people are very bad at distinguishing genuine patterns from random noise, but fortunately there are statistical methods that can help. This article is adapted from a talk at the NZ Skeptics Conference, in Wellington, 7 September 2013. TATISTICS is one of those the whole theory is absurd if we we read a number of anecdotes Sthings, like health, politics and don’t appreciate the Earth’s im- about vaccinated children de- home maintenance, that everyone mense age and size (especially veloping autism, or see a string should know a little about. It when compared to the speed of of record-breaking storms in is important in the life of an in- molecular biochemical processes the news, or see a cloud formed modern citizen bearing a striking resem- and a major part of blance to (Renaissance scientific literacy. painters’ depictions of) Jesus. Lastly, a qualita- A narrow definition tive overview of statis- of numeracy implies tics tells us what kinds proficiency at special- of questions statistics ised calculations; a can and cannot answer, broader notion of it and how statistical ar- encompasses intuition guments can be used to about quantity and inform or mislead. significance, and a general understanding The online version of of how statistics works this article includes some and what statistical interactive content to claims mean. These augment the exposition. concepts are essential Please consider access- for a critical engage- ing it on the NZ Skeptics ment with journalism James Kerr, left, introduces David Bulger at the 2013 NZ website (www.skeptics. and public discussion. Skeptics Conference. org.nz) in an HTML5- enabled browser. For instance, some sense of the scale of very large and the size of the relevant mol- Pareidolia and small numbers is required ecules and cells). Intuition for to evaluate the plausibility of significance helps us draw the We don’t have fangs, exoskel- abiogenesis and the evolution of line between mere coincidence etons, camouflage or venom; diverse species by natural selection; and meaningful patterns when humans rely on intelligence to page statistics survive. A big part of intelli- Hypothesis testing producing observations, we can gence is pattern recognition, and use it to produce a large sample Hypothesis testing is one of we’re very good at it. Maybe too of points on the graph; this is the central methods of statistics. good. The human brain does not shown in Figure 1a as a cloud of Loosely, it is used to quantify the take an ‘innocent until proven grey dots representing hypotheti- amazingness of a coincidence. guilty’ approach to looking for cal observations. When a pattern is perceived, patterns; we imagine trends and hypothesis testing can be used connections on the slightest Now suppose we make an to distinguish between pareidolia whiff of evidence. This kind of actual observation, and plot and real systematic trends. over-interpretation of random it in black, superimposed on data is called pareidolia. the cloud of hypotheti- The human brain does not take cal observations. If it Sometimes we know falls comfortably in the the perceived patterns an ‘innocent until proven guilty’ ap- middle of the cloud (eg, are imaginary, and they proach to looking for patterns; we Point A), then it accords can even be useful. For imagine trends and connections on perfectly with the null example, grouping the the slightest whiff of evidence. hypothesis: the grey dots stars into constellations and the observed point sketching out fanciful im- may have the same dis- The world is already awash ages aids in learning the arrange- tribution. On the other hand, with quantitative, nitty-gritty, ment of the night sky – whether if it falls well outside the cloud formulaic descriptions of the or not we believe that Poseidon (Point C), then it is not the kind mechanics of hypothesis testing, placed Cassiopeia in the sky as a of point produced by the hypoth- and I’m not going to add to them punishment. But when we don’t esis; we have strong evidence (not here, anyway). Instead, know a priori whether perceived against the hypothesis. let’s focus on the conceptual patterns are systematic, intuition framework. is a poor guide. One of the main But where do we draw the line? If the observed point falls aims of statistics is to provide Hypothesis testing is all about just on the outskirts of the grey objective quantitative measures measuring the consistency be- cloud (Point B), it can be difficult to determine whether apparent tween observed data and some to judge how consistent it is with patterns are too strong to appear hypothesis (called the null hy- the hypothesis. Mathematics by chance. pothesis) about where the data (or, often nowadays, computer came from. This is done by Part of the difficulty with simulation) can be used to cal- gauging how extreme the data intuition is that truly independ- culate how extreme the observed would be, if the null hypothesis ent random data tends to cluster point is with respect to the grey were true. ‘How extreme’ re- more than people expect. (Ran- distribution. Conventionally, if ally means ‘How improbable dom is lumpy!) For instance, the observed point falls in the would this data be if the null the chance of any two people outer five percent of the grey hypothesis were true?’ or more sharing a birthday is about 1/365, cloud, the null hypothesis is accurately ‘How likely would we or 0.27 percent. However, this rejected. (This threshold, five be to see something as extreme seemingly rather unlikely event percent, is called the significance as – or more extreme than – the probably will happen (with a of the test. It describes how observed data, under the null 50.73 percent chance) in some much evidence against our null hypothesis?’ pair, given a group of only 23 hypothesis we will need to see people – but it will still seem like Suppose we make an observa- to persuade ourselves to reject a surprising coincidence.