AVing a bias?

The UK has a referendum on the government voting mechanism on the 5th May, to offer a choice between the existing first past the post scheme and an alternative called ‘Alternative Vote’.  I don’t often venture into Political debate, and don’t really intend to now.  But, I do get irritated when I see what I perceive as mental biases being played out in public (e.g. previous post here), and I am not sure if it rankles more when it is cynically deliberate spin, or unconscious.  At its bluntest, the choice between a clever manipulator, or a well intentioned fool?  It the extent to which these mental biases get shown isn’t even close to that black and white, but there are times when it feels quite stark.  And the AV debate seems to be there.

But, I thought I might do some mental ju jitsu and use the frustration positively in prodding the arguments as examples of mental biases or dodgy thinking, whether that is in deliberate exploitation and spin (most of it I suspect), or people fooling themselves.  I should note that I didn’t really start writing this with a strongly held view on which system is better; I suspect that it is AV, but would really rather see some robust analysis, and thus far that is thin on the ground.  But, the dreadful standard of arguments put forward by the ‘no to AV’ group basically makes me more inclined to support AV … if this is their best argument then the case feels pretty weak.

You can fine the ‘No to AV’ website here, and the ‘Yes to AV’ site here.  To save you looking up the key propositions made, on the ‘No to change’ side we have declarations that AV is costly, complex and unfair, and a politician’s fix. The key arguments ranged on the pro side are that MPs working harder to earn – and keep – our support, you have a bigger say on who your local MP is, and that it tackles the ‘jobs for life’ culture.

As represented in their arguments, I have issues with pretty much all of them

How good is the ‘AV is costly’ argument?

It seems customary to start at the beginning, so I’ll start with the AV is costly argument.  This is quite special in that it is flawed at so many levels.  There is a backup document that explains their logic in detail – very welcome since this is all too often missed out, with arguments hanging in mid-air unsupported … but, as I write this I find that they have decided to remove the link to the detailed working from the web-site.  Now, why would anyone do that?  It’s not like it is any more work to leave it there!  Good job I kept a copy – you can see it here.  I’ll come to that shortly, but first it is fun to see which elements stand out in their summary, and then see if they are supported (red italics are mine, to highlight the areas of concern) :-

AV is costly
The change to AV will cost up to an additional £250 million. Local councils would have to waste money on costly electronic vote counting machines and expensive voter education campaigns. With ordinary families facing tough times can we really afford to spend a quarter of a billion pounds of taxpayers’ money bringing in a new voting system? Schools and hospitals, or the Alternative Vote – that’s the choice in this referendum.

First, lets look at the £250m, and the associated friend of all dodgy arguments ‘up to’.  It is well known that introducing a number acts as an anchor on thinking (see here).  That’s especially true if the there is one number surrounded by text, and/or is the only number.  This is why shops state a high price and a large reduction – it makes you think you got a real bargain, and it is mentally very hard to discard that ‘information’ – you end up starting there and making corrections to get what you perceive is the real value.  The addition of ‘up to’ immediately should alert you that this is not going to be the most likely number, it is going to be an estimate that includes pretty much everything.  It signals, if a signal was required, that this is likely to be a deliberate effort to influence your thinking via an anchoring bias, so you should be suspicious.

Then there is the second stand-out part of this argument summary ‘the choice’.  A deliberate aim to create and use an attentional bias in the form of clearly stating that this is a simple choice between based on the data presented.  Even by their own logic this is not true, since they have two more headline arguments. And, manifestly it can’t be true since it does not make any allowance for whether the voting is in fact materially more representative.  One might as well say that the whole of the democratic process is money that could be better spent on hospitals etc.

There are more things wrong with this cost argument even at this summary level, but it feels more fun to wade into the more detailed analysis, especially since the no to AV campaign saw fit to remove the link from their web site (reminder – I kept a copy, and you can find it here).

The £250m builds up from four components; first at £82m is the cost of the AV referendum, second at £9m is the cost of voter education ahead of the referendum, third at £90-£130m is the cost of automatic voting that is claimed to be necessary, and fourth at £26m is voter education with AV.  In total they come to £217-£247m.

Lets start with the first two.  They are, in short utterly irrelevant, since the costs are already sunk (a clear misuse of sunk cost biases – see here).  The decision on WHETHER to have a referendum is in the past, we are having one.  So, whether a change is sensible or not is irrelevant.  So, that takes the total down by £91m to £126 – £156m.

Now, lets take the biggy – the use of technology.  Two models are given for costs.  One gives a range of £76m – £101m, the other £108m.  Onto both is added VAT.  I am mystified about why.  This is government spending, so the tax comes back to the government.  It is a zero sum, so knocking 20% off at a stroke (more accurate to say that they wilfully inflated the cost by 20%).  No specific guidance is given for the robustness or otherwise of either estimate, so a reasonable start would be to use the average, or £98m.

The next question is what proportion of the costs are variable vs. fixed.  Extrapolating from some pilot schemes, and the Scottish example which has less than 10% of the UK electoral base with absolutely no allowance for fixed costs vs. variable costs seems ludicrous.  When scale goes up, the cost falls as the fixed costs get divided amongst a larger user base.  And the allowance for this .  Nada.  Zip.  Not even a mention that it might distort their figures.

But, it is far from obvious that automated counting is required.  It is definitely fair to say that opinions vary – some close to the situation regard it as essential, though Australia has managed for many years without.  But, there is some tortuous logic used.  The clearest attempt goes along the lines of ‘there are multiple different schemes used, so automated counting is the only practical way of doing the count’.  This is clearly bogus.  If you accept that there are other more complex schemes used at the same time, then the introduction of AV vs. FPTP is just a choice between schemes – neither is particularly harder than the other.  You might choose to introduce automated counting, but is it almost completely unlinked to the choice between FPTP and AV.  You might easily extend this and ask why, if there is a fairer system, all elections don’t use that and simplify things back to the point that manual voting can be used and save money?

There is a backup argument here that manual voting would take longer, and would be ‘incompatible with British electoral traditions’.  These would, presumably, be the same traditions that fought against secret ballots, and women voting?  I mean, really, when has ‘tradition’ been a serious argument for anything?  I don’t think that a few hours or days longer in counts would make a deal of difference to the world … and if it did, then we would have gone electronic a long time ago, and solved the current challenges it has.

I spent some years consulting.  If any of my teams had seriously proposed the analysis described here, even without the errors on VAT and taking the high figure not the mean, I would have sent them away to try again.  And, I would have been very suspicious of what they came back with the second time if they though the first was worthy of merit.  It is very very poor.  I’d also note that there is a paucity of analysis on the pro-AV side on whether these costs could be justified, which is unimpressive.  But, this effort is woeful.

Finally there is the cost of voter education. This is given as £25,917,712, based on the cost of the Vote Scotland 2007 campaign.  I have some issues with the methodology since once again it assumes that all costs are variable, and can be scaled simply, which doesn’t seem wholly defensible.  But, as a first approximation, since many of the costs will have been media spend or leafleting, it will do.  I would note that leaflets were found to be particularly effective, and leaflets do not cost 50p/unit.   But, I will try and avoid my own hostile media bias, and just accept it.  It is 50p per voter.  I think I can live with that as a cost.

So, what are we left with on the cost argument?  There might well be a cost of about £25m. About 10% of what was claimed, and though it seems a large number, when set against the UK total expenditure that Governments are responsible for it is minuscule.  It wouldn’t buy a Mars bar for each voter.  For something important.


Right, this post is quite long enough, so I will stop here, and maybe return to the other arguments later.  I noted earlier that I didn’t have a strongly held view on the the choice between AV and first past the post.  On reflection, that is a little disingenuous.  It would be more accurate to say that whilst the quality of political positioning, commentary and discussion is of the quality shown in the ‘no to AV’ positioning, the actual voting mechanism seems like a second order issue at best.

A few others links of interest.  On the ‘for’ side, see here.  For the against side see this in telegraph (pretending to be impartial)

This entry was posted in Cognitive biases, politics and tagged , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s