GDP, or gross domestic product, is arguably the most important of all economic statistics. It is a measure of all the activity in the UK economy in a particular period, and is published on a quarterly basis.
If the quarterly GDP measure has increased compared with the previous three months, the economy is growing. If the figure is negative, the economy is contracting.
The UK's GDP figures are produced on a rolling monthly basis for each quarter.
Revisions, not mistakes
A first - or preliminary - estimate is produced in the first month after a quarter, a second in month two, and a further revision in the Quarterly National Accounts.
There are later revisions when we do the annual publication of the UK National Accounts - the so-called Blue Book.
Revisions to the GDP figures do not usually mean that we have found a mistake - mistakes are actually rare - but rather that we have received new information.
GDP is compiled from thousands of survey returns completed by UK businesses. The main sources used for the preliminary estimate are ONS surveys of manufacturing and service industries.
Information on sales is collected from 6,000 companies in manufacturing, 25,000 service sector firms, 5,000 retailers and 10,000 companies in the construction sector.
Data are also collected from government departments covering activities such as agriculture, energy, health and education. For the second and third estimates, increasing amounts of data become available from other surveys and administrative sources.
The preliminary estimate of GDP comes only 25 days after the end of the quarter, it's the quickest publication of a preliminary estimate in any major economy.
Government and businesses want information fast, so ONS produces its early estimates of GDP based on partial information before all the data that will eventually be available is in.
Does that mean that the preliminary estimate of GDP is really - as some have, indeed, claimed - no more than a 'guess'?
Well, no it doesn't. In fact, the preliminary estimates have proven to be remarkably accurate.
The figure produced in the third month after the end of a quarter - in the Quarterly National Accounts - uses more of the available data. Yet the revision between the first and third monthly estimates is typically only one or two tenths of one per cent.
Nor is it true, as some commentators allege, that the first estimate is biased downwards - that they generally underestimate the strength of the economy and will be revised upwards as later estimates emerge.
In fact, the average revision since the start of 2006 between the first figure and the third month has been minus 0.01%. So revisions have generally been very slightly downwards, albeit by a figure so close to zero it makes no difference.
This means that there would be no benefit in terms of quality, if, as some suggest, we were to delay the production of the preliminary estimate. We'd be losing timeliness but for no gain in quality.
It's like being able to accurately predict the outcome of an election after only the first 20 or so seats declared.
One of the reasons the preliminary GDP estimates, and any later revisions, attract so much attention is the urge to define the economy as in or out of recession: especially when the media is speculating about the possibility of an unprecedented so-called triple-dip recession.
Recession is often defined as two consecutive quarters of negative growth - or economic contraction. Yet when growth is close to flat, the difference between, say, estimated growth of 0.1% (no recession on the popular convention) and -0.1% (one quarter contributing to a recession on the same convention) is actually within the known statistical margin of error.
In fact, ONS does not attempt to characterise the economy as in or out of recession. We see no statistical or economic basis for this.
We recommend looking at the path of the economy over a period of time. Overall, in recent years, the economy appears to have been on a bumpy plateau, with an upward trend but at well below historic growth rates.
The latest quarterly estimate adds to the story but movements in one quarter, upwards or downwards, need to be seen in the longer context.
The opinions expressed are those of the author and are not held by the BBC unless specifically stated. The material is for general information only and does not constitute investment, tax, legal or other form of advice.