However, there is a clue in the table and it’s use of the word ‘experimental’. ONS publishes the ‘unemployment by age’ data (table X02) separately from the headline statistics because although the estimates are ‘derived from the same data source as the headline figures [the Labour Force Survey or LFS]…due to the relatively small samples sizes and subsequent sampling variability, the figures should be used with caution and are designated as Experimental Statistics’.
- My only purpose in drawing attention to Scotland's very poor performance on 16 & 17 year olds on the experimental LFS data was to highlight the dubious nature of the series as a whole;
- it's not really a matter of 'the Scottish Government taking the credit when things go well etc' - it's about using statistics in a consistent fashion. If the experimental data are good enough to justify a claim that youth unemployment has fallen by a third, then we should be very worried about what they're telling us about 16 and 17 year olds. However, as should be clear to anyone who's read the blog, I think we should treat these data with extreme scepticism;
- I happily accept that the APS is a more reliable series. Again, one of my purposes in writing this blog was to make such a distinction between the experimental LFS data and the APS! What isn't credible is for the Scottish Government to use the experimental data to justify the one-third claim then argue that the sample size for 16 and 17 year olds is too small. Are we mean to accept that the sample size for 16-24 year olds yields a perfectly accurate result? That the experimental data are fine for 16-24s but we should turn to APS for the 16 and 17 year old cohort? Of course the LFS sample sizes are too small - this is why the data should not be used as the basis for such bold assertions.