In this Friday's Extra comes the story that your average Thanet citizen earns a whopping £30,000 a year (or so says the Halifax). ("Typical Thanet Salary £30,000!", Martin Jefferies, The Isle of Thanet Extra (KM Group), Friday December 5. 2008, (DE) page 3).

While this might seem outlandish and is clearly unrepresentative it requires interpretation to be understood. It seems that Martin Jefferies did at least get some one to talk to but that Dave Kinnear of the Margate Town Partnership was clearly not experienced with interpreting data.

So it leaves Thanet Star to explain exactly what the numbers mean while attempting not to send you to sleep. Read on to find out just how badly you are being misled by Halifax and others who cite these figures and why they want to mislead us. It's an old slight-of-hand and today it gets exposed.

(hint: it's all about getting you money from you).

Lies, damn lies and statistics

Typically of the easily led press this figure of £30,000 is quoted with little context and no interpretation. Statistics require both context and explanation to mean anything and the Halifax should know better than to release contextless data to the press. I imagine they are after the headline space and they seem to have got it too.

I was unable to find a single news source willing to cite the survey in question and just as few (none) willing to link to it so I am working blind with third hand data. It's not the best but I can still paint the broad strokes for you.

The best source was fool.co.uk which at least published some data.

First lets look at the figures that have gotten everyone so worked up.

Thanet tops the table for 2003 to 2008 increase (£18,769 to £29,956) with an apparent 60% jump. Presented like this the figures are meaningless. They are meaningless because a pound in 2003 is not the same as a pound in 2008 due to our friend inflation.

What we must do is apply some formula to account for the change in value (some of you might be familiar with projecting "real terms" value of next years money using "Net Present Value" and other techniques that show year on year you are working for less).

Using a toy on the world wide web and assuming an average inflation rate of 4% the 2003 figure would actually be £22,835 in today's money which is only a 31% increase. If you don't believe me ask an accountant or a local maths teacher.

This 31% is still inflation busting but is hardly as sensational as the first number. However, it is a lot more indicative of the rate of change.

Next to consider is something called population sample. This sounds a little bit scary but is actually dead simple.

The results will only consider those people in employment. So the population sample is already selecting against the unemployed. As a result the apparent household income is greatly increased on paper while still being nearer £7,000 in reality.

Oh yes, time to crack out the old one about "lies, damn lies and statistics".

Next let us find the context. Context has a special meaning in statistics. This meaning is pretty much what you would expect it to be but it relates specifically to the way the data was gathered and the quality of the techniques used to gather this data.

As I do not have access to the Halifax's report I can not tell you if they used stratified random samples (a good way to get a fairly accurate picture) or if they just stopped a bunch of people going to lunch in expensive places. Without context the data is meaningless.

To give it meaning I'm going to assume that they used some form of representative sample and got a good accurate picture of the population. It's pushing things a bit but without doing so we have no way of guessing how and where the data was fudged.

So how can a poor area have a higher average income than the rich areas. This has two causes and they overlap.

The first can be explained fairly simply the second one is going to make me resort to diagrams.

The average used here will have been "the mean" which is worked out by adding all the values up and then dividing by the number of values. For example if Bob, Harry and Jane have £10, £20 and £0.02 about their person the average would be £10.01. Such a number is meaningless with such a small group but it should demonstrate the method and it's weaknesses.

If you remove some values from the sample the average will change. For example if Jane goes home the average money in the pocket becomes £15. This is what has happened in the Halifax survey I think.

When job losses happen, as they are at the moment, it tends to be uneven. For example most managers, directors, systems analysts, over-paid political types, footballers, soldiers (not including new grunts), surgeons, architects and most of the high brow lot are still earning a small fortune.

The people getting the boot are labourers, factory workers, drivers, shelf fillers and other low paid people. What is more when a company folds you might loose 8 or 9 board members and another 6 or 7 high grade managers but you will loose hundreds (or even thousands) of lower paid workers.

This is equivalent to Jane going home. The average shoots up but actually nothing has changed for those with the big incomes.

What the data actually shows is that the working class guy is in big trouble and could be facing very long queues at the job centre.

Now it's time for the charts.

This first chart shows data "normally distributed" we see that the biggest group are those near the average while as we get further away (up and down) there are less members of the population. If this was IQ most people would be "average" while a few would have astoundingly high IQ and a few would need special help.



If you don't get it yet just nod your head and keep reading. It should become clear shortly.

What we probably are looking at is Skewed Data. Which means that the bulk of the population are no where near the average. This could happen if after Jane went home, Mary shows up with £100 on her - the average is now £43.33 even though it is not representative.

What we should be doing here is identified extreme values (such as Mary) and discounting them so as to get a less distorted set of values. I imagine that Halifax chose not to do this because (a) they don't know how or (b) the figures would be depressing.

At the very least they could have provided an arithmetic mean (multiply all the values together and take the nth root where n is the number of values) which would give us £27.14 as the average value of money in people's pocket. But that is a heck of a lot harder to do unless you are happy using logs and antilogs to get to the answer (or you have a nice pocket calculator).

Conclusions: The Halifax figures hint that the low paid are getting laid off first.

Why the lack of clarity?

So what does it all mean? Why would the Halifax and every newspaper give us a false impression like that?

Let's look at that shall we.

The Halifax exists to make money. It needs three things to do this (a) a confident market, (b) breathing room from political action; and (c) public attention.

Skipping a and b for a moment let's look at the need for attention. We call it advertising in it's traditional form but that is just part of marketing.

On an average year the Halifax gets a few schollarly mentions that are probably worth quite a lot of money in free advertising and brand building.

On a good year the Halifax gets more than just a few mentions it gets a massive wave of publicity as the figures (which, allegedly, cannot lie, apparently) stun everyone by saying something that wasn't expected. In this case that poverty ridden Thanet is rich, rich, rich!

Then lots of reporters start talking about it without the slightest understanding of what they are saying. They get quotes from people not qualified to comment on such topics and they re-enforce the illusion that unemployed people are clearly just too lazy to get a job because the job market appears to be blooming. After all "the average salary" is up, up, up!

This witless reporting hides the fact that in places like Thanet, Havering, Shepway, Hammersmith and Fulham, Westminster and Ryedale the working classes are finding themselves out of work at alarming rates while the top earners are quite safe (just currently).

Not even Myra Butterworth, Personal Finance Correspondent for the Telegraph seems to understand what these numbers mean incorrectly assuming "growth in earnings" when this is not proven by the figures. (That said moneynews.co.uk do seem to be giving a more honest report than most).

What should really scare you though is that the "high-street spending" upon which our economic system rests is powered by the spending habits of these working classes who currently are in position to spend money. (That's why we are in a huge economic down turn).

But Halifax does not care about this they get endless free mentions and appear to know something that everyone else does not. More than that they get talked about. The only thing worse than being talked about is not being talked about, don't forget.

The end result is that Halifax get's more business. Rich people and top earners feel less like freaks, feel less guilty as a result and so feel more like banking with the Halifax. While us less wealthy folk feel reassures by the name that is connected to the apparent good news.

Furthermore such figures stimulate the housing and high priced items markets. People feel more confident to buy and they might just do it with the only lender to understand that the market is "good" (that'd be the Halifax then). (Need A - confident market)

In addition to that they also show the political leaders that they need not worry and so they back off a bit and leave them to do what they want with people's money. (That's what got us in this mess by the way).

The only side effect is that a few of us are left scratching our heads and wondering what's going on. (the public being well and truly had, is what's going on).

In short we are deceived by the numbers because it's in the best interest of everyone perpetrating the story to join in the deception and the remainder are blind to the fact that they are following the wrong crowd in the wrong direction. Which is why the headline writer for the extra has no idea what they are saying when they take average and substitute "typical".

The "typical" value would require the use of such things as the median or the mode (middle value and most common value) and a great deal more investigative reporting. To really give the full picture we should be talking household income and disposable income but that would benefit far fewer big businesses and paint a much grimmer picture.

I would say ignore the KM group (and most other news sources) on this one because, in my opinion, they don't know what they are talking about.