[r-t] Survey #4: Naming of rotations. Results.

[John Camp]

At 00:23 on 27 October 2014, Matthew Frye wrote:

> On 26 Oct 2014, at 14:06, John Camp <camp at bellringers.org> wrote:

>> How statistically significant is a poll of potentially over 200
>> people with 33 respondents, self-selected and with a particular
>> interest in (and firm views on) the topic?  The list itself is
>> self-selecting to start with.

> Presumably the 200 figure is an over-estimate for various reasons
> (duplicates, inactivity).

The actual number is 235.  There are very few duplicates.  Inactivity
doesn't mean lack of interest.  There are plenty of interested lurkers
around.

> As for a bare mathematical statistical significance, I confess I
> know little on the subject of statistics. I will, however, stick my
> neck out and say that a result of 16-17 is not statistically
> significant, whereas the previous 28-7 almost certainly was. I don't
> know where between those two you could draw a line.

Would you say "I know little on the subject of ringing, but I will
stick my neck out and say that a course of Plain Bob Minimus has 20
changes"?

The maths don't matter if the design of a survey is unsound.  This
"survey" tells you nothing except what 33 ringers think.  The sample
is anything but random, even within the list membership.  On the
contrary, it is composed of people with an expressed interest in the
topic.  The findings may be of interest, but it would be unwise to
draw generalised conclusions from them.

JEC