How to use and abuse statistics; a lesson from both LNP and ALP

My previous posts on the levels of youth unemployment in the Queensland regions has certainly created some interest. In a story in today’s Courier Mail both sides of the political divide have attempted to use the original, unadjusted ABS data on youth jobs to score points. As is too often the case the reality is rather different.

Firstly let’s take a look at the LNP claim that “about 11,000 young people have lost their jobs since the last election”. Given the last election was held on Jan 31st 2015 we shall use the Feb 2015 data as our starting point. The original ABS youth employment data for Queensland shows 400,700 young employed for that month. For May 2016 (latest available data) that number is 390,100; a loss of 10,600 which is certainly “about 11,000″…so far so good. However, what theses unadjusted numbers do not tell us (and the reason we must prefer our own Conus Trend series) is the effect of seasonal variation in this data set. Youth employment numbers are historically and consistently much higher in Dec, Jan and Feb than at other times of the year. If the LNP has used Mar 2015 as their starting point the difference would have been PLUS 11,800! The fact is that using “since the election” is a classic case of cherry-picking data to make a point. When we consider the Conus Trend data (which accounts for this kind of seasonality) we see youth employment since Feb 2015 has fallen just 300.

As if not to be outdone with the manipulation of data the Government’s Employment Minister Grace Grace claims that “youth unemployment has fallen in nine out of 12 statistical divisions in southeast Queensland”.This claim is tougher to address given that Grace has not specified a time frame as to since when “youth unemployment has fallen”. Also, does she mean actual unemployment numbers or the unemployment rate? If we assume she means exactly what she says (i.e. youth unemployment) rather than what we might think (i.e. youth unemployment rate) then of the 12 SEQ SA4 regions (those in Greater Brisbane) plus Gold Coast, Toowoomba and Darling Downs-Maranoa over the past month 8 have fallen….this is the closest we can get to what Grace is claiming. However, if we consider the original data from a year before then only 5 have fallen with the other 7 all up.

If Grace is referring to the youth unemployment rate then the original ABS data shows 7 fell over the past month but only 4 were down over the year. So even allowing for the use of the ABS original data, and looking at actual unemployment (rather than the rate) Grace’s statement is hard to back up. If we consider the Conus Trend Youth Unemployment Rates for these 12 regions we see 5 have rates which have fallen both from last month and last year.

The moral here is that using the ABS original data is fraught with danger and wide-open to cherry picking abuse. When we consider the Conus Trend data for youth unemployment we see that jobs have remained broadly stable over the course of the year. The Conus Trend Youth Unemployment Rate for the Greater Brisbane area is unchanged at 12.0% while for the Rest of Queensland it has actually fallen to 14% (from 15.6%); the issue of youth unemployment in Queensland is regionally very patchy and there is clearly no state-wide disaster at hand.

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *