From the college of policing about that 15% study: "This finding was not statistically significant." A study by the same researcher looking at cameras worn by railway staff showed a 45% decrease
https://journals.sagepub.com/doi/full/10.1177/0734016818814889
...which is why their spokesman works for the College of Policing and not in Mathematics. A 15% increase represents a substantial increase. If there were 1000 assaults in a week, this would increase the number to 1150. I'm quite sure every one of those extra 150 assaulted personnel would feel quite strongly about the matter.
I'm not particularly arguing one way or the other regarding bodycams, haven't read the paper in question in detail either, and am certainly not intending to minimise the 'significance' of referee assault generally or for the individuals concerned.
However, it is important not to dismiss both papers purely on a (very common) misunderstanding of the meaning of the term 'statistically significant'.
Also, although Brian is probably quite correct that the spokesman isn't a mathematician, I'd be surprised if the authors of the paper have not had input from a professional statistician in both the design of the study and the analysis of the results.
Essentially:
- In the scientific literature, something is deemed a 'statistically significant' change if it is very unlikely to have occurred by chance [precisely
how unlikely mathematically should be made clear in the paper].
- 'Substantial increase' is not a precisely-defined term, but I think people generally intend it to mean 'a big increase', or 'a big percentage increase'.
It is perfectly possible for a 'substantial' increase to not be 'statistically significant'.
It is perfectly possible for a 'tiny' increase to be 'highly statistically significant'.
To give an example of the former (not the best analogy, but hopefully clear enough):
I toss a coin twice and it comes up heads once and tails once.
I toss a
different coin twice and it comes up heads
both times.
So we have seen 100% more heads with the second coin than with the first coin.
I think it would be difficult to argue that 100% is not a '
substantial' (percentage) increase.
However, I think that it would be equally difficult to argue that the difference is '
statistically significant', since this outcome is quite likely to occur just by chance (it would actually be expected to occur 25% of the time - whereas the threshold for 'statistical significance' is usually specified at <5% or <1% or even lower for some areas of science).
And to give an example of the latter:
If instead I tossed each coin 1,000,000 times & had 500,000 heads for the first coin and 501,000 heads for the second coin, that is now only a fraction of a percent more heads with the second coin than with the first coin, so a
much less substantial difference than before, but it is also now a
much more '
statistically significant' result, as it is extremely unlikely to have happened by chance.
~~~~~~~~~~~~~~~~~
On another point, GDPR is not my area of expertise, but I'd be interested to know whether if the only change made by IFAB / FA was to
allow (rather than to
mandate [or even organise / facilitate]) the use of bodycams then
perhaps the GDPR requirements would be somewhat different. i.e. if a referee
chose to film whilst officiating, that
might be seen differently under GDPR to an organising body
requiring them to film. Obviously once the 'organisation'
receives any footage from a referee who has chosen to send it to them then they have to handle it appropriately. But it could perhaps be argued that is no different to how they would have to handle any footage I choose to send them as a parent if I happen to capture a significant incident when videoing my son or daughter playing in a match, or my son refereeing.
