Lies, damned lies, and statistics

Gabrielle asked:

A recent investigation into universities in Northville reveals that the percentage of philosophy prefessors who are female was 11.2486102% in 2012, 10.9399783% in 2013 and 10.6400161% in 2014. This is very alarming because it tells us that in only 37 years , 6 months and 15 days, female philosophy professors will constitute less than 0.33333333% of Northville teaching force. Identify two main problems.

The CIA World Factbook is one of the most reliable sources for worldwide statistics. According to it only 99% of germans are literate — with a population of 80.716 million people, this means a straggeering 807,160 germans cant read! But working from the same source , we see that only 351,100 canadians have trouble reading, clearly Canada’s education system is superiror at teaching is citizens to read? Is there a problem with this question?

Answer by David Robjant

The abuse of statistics is a political art-form practiced by Governments and campaigners alike. It is all the more effective for the general ignorance of the population. It is possibly a worthwhile research project to find out whether we philosophers are also among the ignorant. You might suspect so, because philosophy isn’t primarily about numbers, but on the other hand the abuse of statistics is mainly about the careful handling of definitions and deductions, which is right up our street.

1. The Canada v. Germany question is pretty easy, once you realise that the total population of Canada is less than half that of Germany. It’s not at all surprising, therefore, that Canada has less than half the number of illiterate adults. Nothing here suggests that Canada’s ‘education system is superior at teaching it’s citizens to read’.

There are more complex kinds of statistical swindle. Like:

2. Pretending that a conclusion or hypothesis is proved by the data, when it is merely consistent with it. Firstly, you can’t deduce anything about gender representation in 10 years time from three years of data. If 2010 and 2011 are lower figures, the 2012 figure could be purely anomalous and no evidence of a trend. Secondly, you need to know how many universities and university professors there are in Northville, before you could work out what would be statistically significant here. In connection with that, you need to work out if there are temporary factors. If the sample size is small enough, the retirement of one staff member could be enough to suggest a trend in a three year run of figures.

It’s not really the case that there are lies, damned lies, and statistics. Typically the statistics are correct, but the problem is that it’s too easy to fool the ignorant by pretending that statistics imply something that they don’t.


Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.