Monday, November 21, 2011

Apples with apples

Okay, I have no professional training in stats.  I have had an interest in them over the years, and have done some reading, and worked with stats for three years before I retired.   All still in a fairly non-academic way.

But it really irritates me when news reports use stats to make a case without comparing apples with apples.

Let me be clear.  I suspect there is a considerable disparity between the rich and the poor in New Zealand, perhaps (and I use that word advisedly) more than in the past.  But I don't have actual figures to prove this.  And nor, it seems to me, does Lane Nichols in his article on Stuff.co. nz entitled Revealing the gap between NZ's rich and poor.  He throws a number of 'stats' around, but do they make the case he's claiming?  I'm not sure.

Firstly, anytime you quote the OECD stats, you need to remember that there are two issues with these: firstly, they're often out of date (Mr Nichols doesn't tell us what figures he's discussing here, it needs to be noted) because the OECD can be comparing stats from 2009 for one country with stats from 2002 for another country.  This was something that concerned me greatly when I did stats work over the last three years.

Secondly, the OECD stats from one country will be based on the way that particular country collects them - which isn't necessarily the way another country collects them.   One of the issues we looked at constantly in my last job was suicide.  The suicide stats are always some years behind, because they rely on coroners confirming that a person has actually committed suicide.   That's in New Zealand, and is probably the same in similar countries.  In other countries suicides aren't necessarily recorded in the same way at all - in some countries suicides can bypass the stats easily. 

This isn't to say that the OECD stats are inflated on one hand and deflated on another.   It just means that we have to be careful that we're comparing apples with apples, and unfortunately we're usually not.

Mr Nichols writes about a report from the NZ Stats: The report's 2004 data – the latest available – reveals the richest 10 per cent collectively possess $128 billion in wealth, with median individual wealth of $255,000. In contrast, the poorest 10 per cent collectively possess $17.2b, with median individual wealth of $3200. While the richest 1 per cent held 16.4 per cent of the country's net wealth, the poorest 50 per cent owned just 5.2 per cent. 

Let's look at this a little.   Firstly, the report is seven years old.   We've had a recession since then, and considerable chaos in the world markets.    That's not to invalidate the report, it just needs to be borne in mine.

Secondly it would be interesting to know what the phrase 'median individual wealth' means, especially in relation to the poorest 10 per cent.   What wealth do these poor possess, exactly?   And since this is a median figure that presumably means some of them will possess/own nothing, while some will be comparitely well-off.   Meanwhile, what about the 80 percent (I'm not sure why 'per cent' is the form for Stuff, but I normally write percent as one word) in the middle?    What's their 'median individual wealth'?   We have no idea from this article, because it's making a case relating only to two groups, pitting extremes against each other.  Furthermore it doesn't seem to quite tally up with the 99 percent slogan put out by the Occupy movement.

I'm not going to pick this article apart word by word.   One final comment about another statement: But the gap between rich and poor still ranked ninth worst in the developed world in 2008. Again we're not given any stats/information to back this up, and I'm a bit confused as to what 'ninth worst' actually means.  In relation to what, or to whom? 

Reports that use stats in this way are similar to those 'scientific breakthrough' articles that we often see.  They throw around a bit of information, but not enough.  'Cure for cancer' is touted - until we read the fine print.  The trouble is the fine print seldom appears in the news reports, as it doesn't in these stats reports.

No comments: