Scotland’s latest statistics on employment have been released, and generated immediate controversy.
Major news outlets covered the story in different ways, with some reporting an increase in unemployment and others interpreting the data as a drop in those out of work. BBC Scotland online initially reported an increase then altered their story to indicate a drop in unemployment.
The First Minister took to Twitter to publicise the figures, saying they showed an improvement in employment and a decrease in unemployment numbers.
This was also supported in a news release by the Scottish Government which used the same statistical interpretation.
Ferret Fact Service has assessed Nicola Sturgeon’s claim and found it to be Mostly True.
Evidence
The statistics in question are published by the Office for National Statistics (ONS), which reports on data from the Labour Force Survey (LFS).
The latest data was published on 16 August 2017, and included quarterly figures and changes on employment and unemployment in Scotland.
It covers the period April to June 2017, and shows a slight increase in the level of employment (1.1 per cent), which now sits at 2,650,000 (74.2 per cent of working age population).
ONS also reported a small decrease in Scotland’s unemployment level (-0.5 per cent) when compared to the previous quarter of January to March 2017. The number of unemployed people fell 12,000 to 107,000 on that measure, accounting for 3.9 per cent of the potential workforce.
The figures also show UK employment rose by 125,000 to 32,073,000, which equates to 75.1 per cent of the working age population. Scotland’s latest rate of employment is therefore higher than that of the UK, and unemployment remains under the 4.4 per cent UK-wide rate.
if(“undefined”==typeof window.datawrapper)window.datawrapper={};window.datawrapper={},window.datawrapper.embedDeltas={“100″:591,”200″:407,”300″:365,”400″:365,”500″:349,”600″:323,”700″:323,”800″:323,”900″:323,”1000”:323},window.datawrapper.iframe=document.getElementById(“datawrapper-chart-aAlp2”),window.datawrapper.iframe.style.height=window.datawrapper.embedDeltas.iframe.offsetWidth/100),100))]+”px”,window.addEventListener(“message”,function(a){if(“undefined”!=typeof a.data)for(var b in a.data)if(“aAlp2″==b)window.datawrapper.iframe.style.height=a.data+”px”});
if(“undefined”==typeof window.datawrapper)window.datawrapper={};window.datawrapper={},window.datawrapper.embedDeltas={“100″:424,”200″:373,”300″:348,”400″:348,”500″:322,”600″:322,”700″:322,”800″:322,”900″:322,”1000”:322},window.datawrapper.iframe=document.getElementById(“datawrapper-chart-DEYSs”),window.datawrapper.iframe.style.height=window.datawrapper.embedDeltas.iframe.offsetWidth/100),100))]+”px”,window.addEventListener(“message”,function(a){if(“undefined”!=typeof a.data)for(var b in a.data)if(“DEYSs”==b)window.datawrapper.iframe.style.height=a.data+”px”});
However, the controversy stems from the fact that labour statistics are actually published every month using a rolling three-month average. This allows for comparison between overlapping three-month periods, which can lead to differing statistical interpretations.
The previous three-month release, covering March to May, put Scottish unemployment at a 25-year low of 3.8 per cent, with 104,000 Scots considered unemployed.
So, when compared to this release, Scotland’s unemployment rate looks to have actually increased by 0.1 of a percentage point to 3.9 percent of Scots over 16. This amounts to a rise of approximately 3000 people.
This is how the latest figures were reported on a number of news websites, including the BBC, which later changed its article.
The employment rate did increase both compared to the previous three months and the rolling three-month average published last month.
There has been media debate over which of these statistics is the most accurate, but the ONS caution against comparison with the rolling three-month statistics.
An ONS spokesman told Ferret Fact Service: “Our advice has always been not to compare with an overlapping period, but rather a non-overlapping one – thus, for example, on the April-June figures we have just published, the quarterly comparison period would be January to March 2017.”
Extrapolations of increases or decreases between overlapping quarterly sets are not statistically secure, given it would mean that the comparisons would include two months which are common to both sets.
“The reason we advise against comparing with the previously set of figures, those for March-May, is that the April and May data are common to both, so in effect you are comparing the change between the single months of March and June.
“But the sample design of the Labour Force Survey means that while we can be confident that it is nationally representative over a three-month period, we cannot be confident that this is true for a single month.”
The statistics body also states in their employment statistics release that several non-overlapping quarters should be used when looking for trends in the data, as “changes on quarter at regional level are particularly subject to sampling variability”.
if(“undefined”==typeof window.datawrapper)window.datawrapper={};window.datawrapper={},window.datawrapper.embedDeltas={“100″:368,”200″:351,”300″:310,”400″:284,”500″:284,”600″:284,”700″:284,”800″:284,”900″:284,”1000”:284},window.datawrapper.iframe=document.getElementById(“datawrapper-chart-BY8Rl”),window.datawrapper.iframe.style.height=window.datawrapper.embedDeltas.iframe.offsetWidth/100),100))]+”px”,window.addEventListener(“message”,function(a){if(“undefined”!=typeof a.data)for(var b in a.data)if(“BY8Rl”==b)window.datawrapper.iframe.style.height=a.data+”px”});
Data in the reports are only estimates of the true figures, and the changes referenced in Sturgeon’s tweet were within the margin of error set out by the ONS. So caution should be exercised when making statements based on only two non-overlapping quarters.
Neither approach is without potential for misinterpreting or overstating statistical trends, however the First Minister is comparing non-overlapping periods which is recommended by statisticians at the ONS.
Ferret Fact Service verdict: Mostly True
The First Minister’s claims on employment are correct when compared to the previous non-overlapping quarter, although the changes are within the margin of error. Media reports cited an apparent increase in Scotland’s unemployment when compared to the last rolling three-month statistics, for March to May, which showed a record low of 3.8 per cent. However, the Office for National Statistics warns against direct comparison of overlapping month-to-month data as it is not possible to accurately compare the two sets.
In response to a Ferret Fact Service evidence request, The Scottish Government provided a link to its press release announcing the Scottish results.
Photo thanks to J J Ellison, CC BY-SA 3.0
https://www.sharethefacts.co/share/176314ac-d1a7-410c-a1d8-1c8bcd74cc1f
