Wednesday, August 20, 2008


Paul Luna found this old book on graphology in a second-hand bookshop. The cover is interestingly and appropriately worn – appropriately, because at first glance I took it to say... well, read the first four letters for yourself. Graphology, if you recall, is the 'science' of analysing handwriting, which some organisations apparently take seriously when considering job applications. This particular book claims to tell you how to judge someone's confidence, altruism, degree of introversion, and many other things, but it makes little mention of 'was writing this on a train', 'was using a rubbish biro', 'is obviously French' or 'was trained as a graphic designer'.

Wednesday, August 13, 2008

You will only remember 10% of this, apparently

Every now and again you see a claim that we remember 10% of what we read, 20% of what we see, 30% of what we do... I forgot how it goes, and so I should since it is entirely spurious. In one place, I found it reported with much more plausible precision: 9%, 17.5%, 31%, etc. But it was still rubbish, it turns out. Some years ago I saw this stated authoritatively on a BBC web page, even attributing it to 'recent research', so I wrote to ask for the citation. They replied that they had got it from the British Dyslexia Association, so I wrote to them. They in turn replied that they had read it somewhere, but they hadn't got a source. 

Hunting for the source, I posted a query on the Infodesign Cafe, which put me in touch with Michael Molenda of Indiana University, who was on a similar hunt. He eventually published a short paper with his findings (you can find it here). He traced it to Edgar Dale's 'cone of experience', published in the late 40s. Dale used a schematic diagram (below) to illustrate his view that increasing richness of experience would lead to greater learning. 


Somewhere along the way, someone has added the figures, and these have been repeated endlessly ever since, deeply embedded in the teacher training curriculum. Tony Betrus and Al Januszewski of the State University of New York have published a collection of bad cones

It seems unlikely that a 'quotation' like this would have survived for so long unless there is some truth in it - in other words, it chimes with people's experience in some way, just as a saying such as 'a picture is worth a thousand words' does. Perhaps this is just a modern version of a proverb – it's just that these days we need statistics. 

The Education department at Cisco Systems have looked into the evidence that actually does exist, and produced a useful metareview .

Diagrams and irony

There are various websites around that collect data graphic interpretations of pop songs - they're good for a chuckle until you tire of them. One that's been doing the rounds is this nice graphic from Evita (credited to brianmn). 
To be literal minded for a moment, I could point out that the song doesn't actually suggest that anyone should 'cry for me' – because, after all, 'I never left you'. Or perhaps it is ironic and suggests that Argentina should actually be crying.

But of course, diagrams don't do irony very well. This next one (credited to sftekbear) shows another limitation of its chosen format. There are in fact fifty ways to leave your lover, only a few of which are specified in the song, and they are not given comparative frequencies as implied by this chart. 

However, a professor writes: 
In fact, although Simon (1975) is often quoted as identifying ‘50 ways to leave your lover’, we must treat this figure with caution. Reviewing the primary source, we find that Simon speculates that there ‘must be’ 50 ways, but does not present supporting data, nor does he claim 50 as an exact number. Only four ways are detailed:
  • Just slip out the back
  • Make a new plan
  • Just drop off the key
  • Hop on the bus.
  • Simon makes 2 additional proposals concerning the manner of departure
  • You don’t need to be coy
  • You don’t need to discuss much.
  • A major theoretical problem arises from the lack of a clear categorial distinction between the 4 ways. An alternative view is that these are simply 4 stages of a process model: that is, in combination they describe only one way to leave your lover: 
    1. Make a new plan; 
    2. Drop off the key; 
    3. Slip out the back; 
    4. Hop on the bus. 
    However, this view is easily countered by further reference to the original data: Way 1 (slip out the back) specifically applies to a named individual (viz. Jack), whereas Way 2 (make a new plan) is specific to people named Stan. Since the principle underlying the allocation of method to individuals appears to be rhyming, we may reasonably speculate that Way 1 would also be appropriate for persons named Mac, or Zak, while Way 2 is also appropriate for persons named Dan. 
    On this basis we may proceed to a more accurate calculation of the different ways to leave your lover – that is, it must correlate with the number of available names within the population, with allowances made for duplication resulting from homophonic terminal phonemes. We may, then, posit a direct relationship between available forenames within a particular language, culture, or discourse community and available options for terminating amatory relationships. 
    This leads to the conclusion that the number of available amor-terminatory strategies is directly proportional to the number of available personal nomenclature allocation options. 
    Some cultures (eg, the UK) permit an infinite range of options, with no rules for spelling (viz, Agnes, Agyness), while others (such as Portugal) require parents to choose from a prescribed list. In Sweden, there is no prescribed list, but parents can be prevented from choosing unusual names. It is therefore tempting to hypothesise that divorce rates in regulated countries should be lower than unregulated countries, since there will be correspondingly fewer ways to leave your lover. 
    This is indeed confirmed by the statistics: UK – 2.7 divorces per 1000 population; Sweden – 2.4; Portugal – 1.9. Of course, this figure should only be properly calculated using data adjusted for the frequency of matching terminal phonemes (which reduce the lover-leaving options within some language groups). And further we may speculate that in unregulated societies, parents may opt for names that, having no suitable rhymes effectively insulate their progeny from the risks of divorce: this may have been the motive of the Swedish parents naming their children Lego, Metallica or Brfxxccxxmnpcccclllmmnprxvclmnckssqlbb11116 (see Daily Telegraph, 7June 2008). We may conclude that further research is necessary. etc, etc.