False Correlations and Family Values

I paid my weekly visit to Quadrant Online today. Quadrant bills itself as "the leading general intellectual journal of ideas ... published in Australia" but, under the stewardship of current editor Keith Windschuttle, it remains the leading generally anti-intellectual journal of ideas published in Australia. Plus many of the ideas that Quadrant publishes are only ideas in the same way that astrology, cartomancy and dianetics are ideas.

One of Quadrant Online's frequent writers is Bill Muehlenberg, a staunch defender of family values; Bill's very keen that every child's right to a biological father be respected, though he's a bit vague on the details of how this will be achieved. Under the title "Our war against children" Bill's regurgitated a lot of the arguments he put forward in "The perils of fatherlessness" in February this year. He's also regurgitated a fair bit of his undigested evidence:

Strong connections between crime and family breakdown have been made by the Centre for Independent Studies in Sydney, which compared crime rates with out-of-wedlock birth rates from 1903 to 1993. It found that the “percentage of ex-nuptial births correlates significantly with both serious and violent crime at both one and two decades time lapse”. (February)

 The Centre for Independent Studies in Sydney compared crime rates with out-of-wedlock birth rates. It found that the “percentage of ex-nuptial births correlates significantly with both serious and violent crime at both one and two decades time lapse”. (October)

 So does this really mean that between 1903 and 1993 single mothers were spawning a lot of feral children who, when they reached the age of ten later went on to slipping shivs into their peers or, even worse, their elders and betters? Well, no it doesn't. These "significant correlations" are good examples of false correlations - underlying all of them is another potentially significant driver: population growth. As the population grows you get both more women capable of spawning children "out of wedlock" and more people capable of becoming serious violent criminals. As for those feral ten year olds, you might like to consider Figure 1, which shows, generally, how a plot of population in any given year versus the population ten years later would look for a growing population.

Figure 1: idealised plot of population vs population 10 years later for a growing population.

When I did a Google search for this unnamed paper at the CIS web-site it didn't turn up. I didn't give the search much more than ten minutes - if I'd published a study with such an obvious and egregious error, I wouldn't want anyone to know about it either.