Shibboleth, Mixolydian, Heteroscedasticity – and Kipling

Interpreting Scatterplots
25 February 2013
Twitter for educators
11 March 2013

All areas of human endeavour have specific language. Cricket commentators, art critics and wines buff make this very obvious.


My son, who is blind, autistic and plays the piano like an angel, is studying Jazz, and I’m helping him. You can see him on his YouTube channel . There is a specific language around Jazz, and I’m not talking about ‘scat’. (Hmm just realised the other meaning for that word!) In the Jazz course they use words like Mixolydian, Chromatisism, Quartal Harmony…  I nod and smile. This language expresses ideas clearly and uniquely and is outside my comprehension. (Mixolydian is based on the Major scale, but with a flat 7. – clearer now?)

Trumpetty yellow, Daffodils, Narcissus

This week there was a statistics list discussion about the meaning of the term “multivariate”. As part of the ongoing discussion, someone suggested that using exact terminology exactly avoids a situation such as saying “I have yellow flowers in my garden with trumpetty bits, that come out in spring and have oniony looking bits in the ground.” This can also be said as “I have daffodils in my garden”.  However it can also be said as “I have Narcissus pseudonarcissus  in my garden”. Each of those phrases expresses the same idea, but with differing clarity or exclusiveness depending on the audience.
Hagley Park Daffodils


Language can be used to exclude, as well as to inform or communicate. The term “shibboleth” comes from the book of Judges in the Old Testament of the Bible. When the Gileadites wished to find out if people crossing the river were Ephraimites, they would ask them to say the word “shibboleth”. If they said it as sibboleth, they killed them. The Old Testament can be a bit like that. The word “shibboleth” is now used to mean a code word, or knowledge that only a certain culture or group will know. Sometimes it can seem that statistical terms are used so only the initiated will be able to understand.

Virtue and Common Touch

As statisticians, operations researchers and teachers of statisticians and operations researchers we have many different opportunities to select the language we use. We must always be aware of our audience. In the poem, “If”, Kipling encourages people to be able to “…talk with crowds and keep your virtue, Or walk with Kings – nor lose the common touch,” Academics “walk with kings” when they write academic papers, using highly specialised and exclusive language. We need to make sure we do not lose the common touch. At the same time we should “keep our virtue”, and use the correct statistical term when the circumstances arise, making sure that we retain the common touch so that all understand.


When I use the term heteroscedasticity I am usually doing so for one of two reasons. First, that the data in question has non-constant variance, and I am explaining the concept and technical term to a client, student or colleague. Second, because I really like the word. “Heteroscedasticity” is eight syllables of tongue-twisting goodness! But, really, “non-constant variance” says exactly the same thing, has only six syllables and is easier to understand. I suspect a degree of linguistic snobbery appearing.

Communicating Statistics

Greenfield wrote a paper in 1993, which is still disappointingly relevant today. In “Communicating Statistics” ( he suggests that statisticians have a great deal to offer the world, and that we aren’t doing a good job of making people aware of that. He was damning of the type of language used in academic publications, which ensure that any potentially useful results are obscured by “prolix and pseudo-objective style”.
This flows over into our consulting endeavours, where the aim should be to communicate rather than exclude. Greenfield gives the example fictionalised in this comic:

Depiction of true event.

Depiction of almost true event. Click to view.

Greenfield’s parting provocative statement was to suggest that statisticians produce more cookery-books and more easy-to-use programs, and encourage their use by everybody who can benefit. These books and programs can carry the message that if they want to do better they should study more and seek the guidance of statisticians.
In closing he says “Our audience, our customers are out there. They need us, even if they do not realise it. We must change our culture, our philosophy, our public relations and our use of language to reach them.”

Choose our words

When we use very specific technical terms we need to make sure that they are really necessary. Is there a simpler, and just as accurate way of saying the same thing? If our audience is statisticians, then really we can indulge in specific technical language. But if the audience includes students, non-statisticians and the general public, then we should probably use simpler terms, or at least “gloss”, or say what the word means along with its use. (There was an example of glossing right there!)
I have written earlier about the minefield of statistical terminology, particularly when the statistical word also has an everyday meaning which is not quite the same. Examples of this are “significant”, “random” and “relationship”. The post includes some suggestions for teaching statistical language.
But as well as teachers, we are also communicators, and need to get our message across in the best way possible. It is vital to determine our audience, and make sure we bring them along with us.
I contemplate the new New Zealand curriculum with excitement. Through the efforts of a group of statisticians we are able to inculcate a greater understanding of the essentials of statistics from an early age to much of the population. The role of the statisticians is to help the teachers feel at home in the world of statistics, so that they can invite their students along. These are exciting times. The rest of the world is watching.


  1. Patrick says:

    Even the term non-constant variance has a degree of snobbery. It assumes that you understand the term variance!

    • Richard Brown says:

      But you could say something like “the variability is not the same in all the groups” or “the variability appears to change with time” or something else which ties what you are trying to say back to the data you are looking at.

  2. The first, i would like to thanks i recevie date from yoor organizaation
    i feel very good to recevied date frome Learn and Teach Statistics and Operations Research.
    i have aquestion . Would you advice to me please . Where can i search information environment statistics? and about scholarships environment statistics subject

Leave a Reply

Your email address will not be published. Required fields are marked *