There has been a lot of talk in the UNISON blogosphere about the forthcoming General Secretary election. Much of the commentary comes with ideological baggage attached. This post will attempt to apply a little scientific rationality and objectivity to one aspect of the debate and draw an objective conclusion about who it is best for UNISON members to vote for.
In his book, How to Lie with Statistics, Darrell Huff explains how statistics can be used to mislead. Much is being made of a statistic from one of the (ultra left) candidates and Huff's book is of use. Paul Holmes says that he leads a branch with 86 per cent density. The inference we are being asked to draw is that this impressive figure is some form of “outlier” and due to Holmes' leadership and the union would benefit from him being the next General Secretary.
Given the importance of who leads our union, this statistic and the inference we are asked to draw from it deserves some careful and rational appraisal.
Check the data
Firstly, any credible statistician or scientist would want to verify the raw data themselves to ascertain that this 86% figure is correct. Too many mistakes, in all areas of life, have been made without going back to the source data. It would be foolish for UNISON members to take such an important decision and to be influenced by non verified data (NB widely suspected and soon to be proven as a hubristic lie).
Even if we do accept the 86% figure as correct, the inference that this impressive figure is an “outlier” as a result of the (ultra left and messianic) leadership shown by Holmes is still problematic.
This figure is a snapshot in time
The 86% figure is admittedly impressive (albeit fictional) but it only gives a point in time and does not give the whole story.
How would the figure look if it had been at the 86% density for the past 12 years? Maintaining a good density is not as impressive as building it up.
How would the figure look if in 2005 it was 96% and it had now actually fallen? Not so impressive.
So, without the past figures it is not possible to say if the things have got better, worse or stayed the same.
This is also a snapshot in space
We know nothing of the densities of other similar branches in the locality (widely suspected and soon to be proven as higher). Anyone with a cursory knowledge of UK labour history would know that areas such as West Yorkshire (where Holmes' branch is), the North East and the North West have strong traditions of trade union activity.
How would the figure look if other branches in the locality had similar or better densities? It would not be so noteworthy would it? A branch that is similar to others is not really an outlier worthy of note.
Correlation does not imply causation
Even if we discount the issues of a temporal and spatial nature outlined above, we still find problems.
Holmes (and his controllers/supporters) would have UNISON members believe that the density is caused by his ultra left leadership. However, all we have is a correlation between the density and Holmes.
As any statistician will tell you, correlation does not imply causation. Indeed many spurious relationships have been predicted in many areas of life based upon the fallacy that correlation implies causation.
An example of a spurious relationship is that as ice cream sales increase, the rate of drowning deaths increases sharply. Therefore, ice cream causes drowning. Clearly this is nonsense. The increased drowning deaths are simply caused by more exposure to water based activities, not ice cream. The Wikipedia link here gives more info.
Indeed, looking at only one branch is not sufficient evidence on which to say there actually is a correlation between high density levels and a ultra left leadership. So a look at other ultra left led branches is in order.
We know there are many ultra left branches in the London region, two of them led by NEC members. Both of these branches have densities of less than 50 per cent. Indeed the same is the case for the majority of ultra left led branches in the union.
So not only does correlation not imply causation – there is no correlation!
Failure to look for contrary examples
It is a well reported that people tend to look for evidence to support their own preconceptions or hypothesis. This is called confirmation bias. So one thing that any good scientist will do is search for evidence contrary to their own hypothesis to test if the hypothesis is robust.
The hypothesis here, is that the high density figures are due to the ultra left leadership of the branch shown by Holmes.
A contrary example would be to find branches with high density that are not led by ultra left activists. There are many examples of branches with more mainstream activists that have good densities, but some branches that immediately spring to mind are our police branches. They are universally acknowledged in UNISON as having high density and good recruitment rates.
And not even Paul Holmes, nor his NEC blogging controllers/supporters would claim our UNISON police branches are led by ultra left activists!
Conclusion
You might ask why I've bothered to write what might appear to be a rather dry and nerdy article. Well, the union faces some of the toughest times we've seen for a long time and who leads us is vitally important. If candidates are trying persuade members to vote for them on the basis of some quoted numbers then these numbers deserve to be examined.
I am sure that Darrell Huff would agree that I have shown readers of this blog is that the numbers quoted by Holmes and his boosters are at best irrelevant, and at worst misleading. Personally speaking I do not want my General Secretary to be quoting figures that are either irrelevant or misleading.
Given that I have gone further than just looking at the density quoted by Holmes, and I have objectively dismissed the hypothesis that a hard left leadership leads to improved density, I think it is clear that a ultra left General Secretary would not help the union. So that excludes fellow ultra left candidate Roger Bannister
Of the currently declared candidates that just leaves Dave Prentis and someone unheard of by most people and is unlikely to get sufficient nominations to make it to the ballot paper.
Mmmm, not really a difficult choice is it
Mark Rayner