Posted by: Eoghan | July 3, 2008

Stat’s not right!

You know what really grinds my gears?

Well it’s me mainly since I’m learning to drive at the moment (More on that in the next post). But what’s grinding my gears in the “getting my goat” sense of the phrase is the abuse of statistics.

What with print media being mostly cut-and-paste press releases, the least we could hope for is a careful analysis of the information the journalist is being spoonfed. Instead standards just seem low across the board. Here are two examples, please feel free to join in the grumpy fun and highlight any others you come across.

1. Men Kill Women Shocka!

  • “Men ‘to blame for most female road deaths'” the Irish Times informed us on Monday.
  • “Male drivers to blame for 68pc of female road deaths” the Indo offered on Tuesday.

 

“Bold men”, you say. But here’s a headline the facts actually support.

  • “Women to blame for most female road deaths”

 

Hang on! Isn’t that just the IT headline with the first word changed? It is. It’s also true. So is this…

  • “Female passengers twice as likely to die with male driver”.

 

The actual statistics (from a joint RSA/CAWT report) relate to female passenger deaths on the road between 1997 and 2006.

In short, 345 female passengers died in road collisions, 234 (68%) of those in vehicles driven by men.

Further numbers follow to highlight the extra risk of males aged 17-24. There are also some interesting figures on female passengers’ opinion of their male drivers and what action they take/do not take as a result.

There are at least a couple of underlying assumptions*. One is that if you were the driver of a car in which someone died in a collision then you are to blame for their death**. A second is that “road death” means “car user death” although these are only 55 per cent of all road deaths for the period in question. That’s fine as long as we’re all aware what is being talked about.

2,226 car users died in collisions between 1997 and 2006, out of 4,015 road deaths. Male road deaths are three times as frequent as female ones. That implies 557 female road deaths in the period. The RSA/CAWT report tells us that 345 died as passengers. That means 212 died as drivers. That means 212 female drivers + 111 female passengers driven by other women = 323 total women died in cars driven by women. 323/557 = 58 per cent.

Hence, women were to blame for 58pc of female road deaths. Of course, men were to blame for most female car passenger deaths but that’s not what the headlines said. Or what leading paragraphs in the articles said either, viz

New research shows that more than two-thirds of women who died [as passengers – EC] in car crashes from 1997-2006 were passengers in cars driven by men. – Irish Independent

Women are more likely to die in a car crash if they are a passenger in a car driven by a man [than if they were a passenger in a car driven by a women], new statistics have revealed. – Irish Times

Of course the main statistical problem with the report is that raw road death numbers don’t really tell you how relatively dangerous different ways of doing the same thing are. No more than my telling you I had firefighters wear orange helmets for three months (during which there were 4 deaths) and yellow helmets the rest of the year (during which there were 9 deaths) tells you that orange helmets save lives. The death rate was 16 per year with orange and 12 per year with yellow… maybe yellow is safer.

The point is, if a risky activity is being tried numerous different ways over a period of time, it is important to know how long each way is tried for. In this case, how many hours do women spend driving, driving other women, being driven by men?

*In the following discussion, I used NRA/RSA statistics collected on this site to fill gaps in the data.

**Seems obvious for single car collisions but what about an innocent being hit/run off the road by some other vehicle. If there were some big difference between the sexes in the frequency or lethality of these accidents then this assumption might no be valid. I’d guess it broadly is though.

 

2. Growth in Rate of Economic Growth Fails to Grow… Shocker?

Another classy effort from the Times, “Economic growth contracts in first quarter” they assure us. The first sentence of the article reads, “Economic growth in the Republic shrank by 1.5 per cent between January and March this year…

Growth is the rate at which something becomes bigger.

To contract is to become smaller.

So “growth contracts” means that a reduction in the rate at which something is getting bigger. So, if I weighed 69kg in 2006 and 74kg in 2007 and now weigh 77kg, then I grew by 3kg this year compared to 5kg growth last year. My growth has contracted. I have not, however. I have gotten larger. I am now a svelte 12 stone where once I was a lanky eleven. What’s going on?

Well… Gross Domestic Product (GDP) for Ireland, which is a measure of the size of the economy, was 1.5 per cent smaller at the end of the first quarter of 2008 than it was at the beginning. So “growth” was MINUS 1.5 per cent. It (growth) was -1.5 per cent. It did not SHRINK by 1.5 per cent, it, itself, IS minus 1.5 per cent. The economy became 8.7 per cent larger during the first quarter of 2007 so, if you don’t mind a year-on-year comparison, economic growth has gone from 8.7 to -1.5. It has fallen by 10.2. Measured in percentage terms, 10.2/8.7 = 117 per cent, so economic growth contracted 117 per cent of itself year-on-year, as in we lost ALL growth (all = 100%) and then some. Alternatively, growth fell by 10.2 (per cent) of the economy’s size, from 8.7 to -1.5.

Bottom Line: Economic growth didn’t contract by 1.5 per cent, the economy did.

Advertisements

Responses

  1. […] Eoghan at Casa Casey Courtney has had enough of this abuse of statistics. He’s got your number. […]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Categories

%d bloggers like this: