An article in the Gothenburg local news paper, GP (full article, Google translation) , indicates that young people are becoming increasingly comfortable with faking sickness to get away from work. However, it is likely that the journalist is wrong. Here’s why.
The conservative issue with young people and labor
Based on a statistical survey carried out by the Swedish opinion-measuring institute SIFO on behalf of the news paper (Göteborgsposten, GP, independent liberal) the journalist Anna Holmqvist draws the conclusion that young people don’t want to work, but rather stays at home pretending to be sick. However, she is obviously not very familiar with how statistics works, nor does she hide her political agenda. Let’s go through the article together while I point out some of the most obvious misunderstandings in it.
According to the article, six percent of the Swedish people thinks it is okay to tell authorities that you’re sick, when in fact you are not. Already here the problems arise. First, the method of the survey is not disclosed, so we’re left to guessing the importance of these results. Not publishing such numbers is often not a good sign, so let’s suppose that SIFO is not that confident in these results itself (and trust me, these people know their statistics). Probably the answering frequency was not brilliant, to say the least. Second, the survey says nothing about for how long these people think it is okay to stay home for “non-sickness” reasons, nor does it define what is meant with “faking disease”. This is also a serious problem for this survey.
The which-hunt for the young
In the next sentence, the article states that 15 percent in the ages 15 to 29 years thinks it is acceptable to fake sickness. It then refers to a similar survey made five years ago where only 12 percent in this group said this behaviour would be acceptable. Then Anna Holmqvist makes the claim that this is a significant increase, and thus inclines that the youth increasingly disobey the rules. While a populist statement, there is no significant data to support a such statement. In fact, going from 12 to 15 percent is a 3 percent increase. Given that this survey probably reached less than a thousand people (many people are unreachable when SIFO calls etc.) three percent represents 1000*0,03 ≈ 30 persons. Then, note that we’re only considering the group between 15 and 29 years old here, which statistically can’t make up more than a third of the people in the survey (probably less, since these surveys have problems reaching young people and thus are biased from the beginning). This leads to the conclusion that these three percents consist of less than 10 people! Less than ten! Hello uncertainty. When working with such small numbers of individuals, conclusions will be drawn from random events rather than statistical trends. Everyone working with statistics knows this. Anna Holmqvist either doesn’t, or hides this knowledge well to support her populist political agenda. I don’t know which, but obviously it’s not good journalism.
The next sentence supports that Anna does not know statistics, however, since it comes with a rather hilarious claim: “Something seems to happen with our morale when we turn 30.” Hopefully, this is a joke. Of course it could look like this to an untrained eye, but this is an issue with how the age groups are divided. Draw the line between the groups at 28 instead (15-27, 28-45, etc.) and you’ll suddenly see that something happens with our morale when we turn 28… This part is just bullshit.
Who is less educated?
Then the article goes on drawing similar conclusions related to education, saying that lower education leads to higher a degree of cheating. Once again without caring about statistical issues. Holmqvist does not even point out that less educated people are most likely over-represented in the 15-29 age group, and that these numbers thus influences each other in some way. Instead, she goes on to conclude that older people often are less educated (which is true, of course), and that this makes the results even more puzzling. Please, make the connection between young age and low education here…
At the end, the article points out that people living in the Gothenburg area are more likely to cheat with disease faking than other people in Sweden. However, once again we can use math to disprove the significance of this: 1000 people in the survey, the Gothenburg area holds about a million inhabitants, while Sweden have about 9 millions. Thus, Gothenburg have about 100 people answering to this survey, comparable to Stockholm. Smaller towns in Sweden will have too low answering frequency to be significant. Now the article claims that 14 percent of the people in Gothenburg accepts cheating. 14 percent – that’s about 14 people. Wow! What a significant number to compare. And the article don’t even give Stockholm’s numbers for us to compare.
Percents are not facts
Personally, I think that GP should send Anna Holmqvist on a 15 hec statistics course. To everyone else, please don’t accept when journalists throw percentage number around themselves. Percents are not facts. Percents are statistical measurements, and as such they should be taken with a huge grain of salt.