SNP’s claim that un­em­ploy­ment has dropped is ... MOSTLY TRUE

Sunday Herald - - FACT-CHECKING SERVICE - BY THE FERRET FACT-CHECKER SER­VICE

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 in un­em­ploy­ment then al­tered its 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.

“Good news on em­ploy­ment again to­day. Scot­tish un­em­ploy­ment down to 3.9% (UK 4.4%); em­ploy­ment rate up and now marginally higher than UK,” Ni­cola Stur­geon said.

This po­si­tion was also sup­ported in a news re­lease by the Scot­tish Gov­ern­ment which used the same sta­tis­ti­cal in­ter­pre­ta­tion.

Ferret Fact Ser­vice has as­sessed 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 Au­gust 16 this year, 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. The num­ber 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.

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 per cent 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 STV News and 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 cau­tions against com­par­i­son with the rolling three­month 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 AprilJune 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 pre­vi­ous set of fig­ures, those for March-May,” says the ONS, “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 its 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”.

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. The Ferret Fact Ser­vice (FFS) at https://the­fer­ret.scot/ is Scot­land’s first non-par­ti­san fact check­ing ser­vice. We check state­ments from politi­cians, pun­dits and prom­i­nent pub­lic fig­ures about is­sues the pub­lic are in­ter­ested in. Just launched, FFS works to the In­ter­na­tional Fact-Check­ing Net­work code of prin­ci­ples.

Ni­cola Stur­geon took to Twit­ter to claim Scot­tish un­em­ploy­ment had fallen to 3.9% com­pared to the UK’s over­all 4.4% Pho­to­graph: PA Archive

Newspapers in English

Newspapers from UK

© PressReader. All rights reserved.