Saturday, July 14, 2012
Alarming abuse of statistics
My wife and I tried to determine what the survey actually studied and concluded. There weren’t much details. But we did find out that they surveyed 1,290 students from 5 secondary schools and 1 primary school; the students ranging from 6 to 29 years old,. A 29-year-old student? That’s rather odd to start with, isn’t it?
So what exactly is the deal about the 400 dollar pocket money? According to their honorary advisor, a pharmacist, those who get fewer than 400 dollars a month has a 5% higher risk of being depressed, compared with those who gets more than 400 dollars?
Wait a minute. Is a 5% difference statistically significant? There was no answer to that question.
And, what does it mean to have 400 dollars for pocket money? Does it cover only snacks? How about lunch and transportation? What about stationery and other supplies? It is quite a bit of money for most primary school children, particularly primary 1 and 2. But certainly not as much for secondary 5 and 6. Have these been taken into consideration? We couldn’t find any hint from the reports. Neither could we find the actual report itself, even from the organization’s web site. It does not seem to have been considered. Given all these uncertainties, is the 5% difference still a valid observation?
A doctor at the press release did say that more studies were needed on the relationship between pocket money and depression and that a smaller amount of pocket money did not necessarily lead to depression, presumably because there had not been sufficient data so far. But this was exactly what the headlines were screaming. So who was to blame for such an inflammatory finding? The researchers for a poorly designed study and a falsely made conclusion? Reporters and editors for generating unsubstantiated but inflammatory headlines?
What I know for sure is that parents who are already hard-pressed do not need further pressure to give their children more money, particularly for those parents who cannot afford it. What they need is better information and support.
This is just more evidence that probability and statistics should be a compulsory subject, particularly for all “professional” disciplines.