Lies, Damned Lies and Statistics
It is a sad fact of modern life that anyone with an idea can support that idea with statistics. The less the public knows about the source of the statistics, the more possible it is to have lies posing as scientific results. Recent opinion polling in the Czech Republic offers a case in point.
On 17th March of this year, the Czech News Agency (ČTK) announced the results of an opinion poll which claimed to show that 59% of citizens wanted a referendum on NATO entry and that 70% of the people would take part in such a referendum. In contrast to their political leaders who were generally rejecting a referendum, the general public seemed to be in favour of putting this crucial question to a plebiscite.
On 6th April, however, CTK reported that only 36% of the electorate wanted a referendum on the issue of the Czech Republic's entry into NATO. Opponents of the referendum thus had their ammunition.
What could explain such vastly different polling results? Could there really have been such a dramatic swing in public opinion in a matter of just a few weeks?
Unfortunately, because of the way these polls were reported, it is nearly impossible to say what is causing the differing results between the first poll, conducted by the polling agency Stredisko empirickych vyzkumu (STEM), and the second by Sofres-Factum. When releasing this information, CTK did not feel it important to report what the actual questions were in each case so proper comparisons cannot be easily made. Because the major daily newspapers take so much of their information directly from CTK, this confusing picture was repeated throughout the press with the result that even the most astute newspaper reader will not be able to understand the discrepancy.
This is the unsettling aspect of this episode which paints a disturbing picture of the Czech media. Usually, only a minimum of the facts are made readily available, and while this provides supporters of each side with superficial numbers to be used as ammunition in the public debate, the open-minded citizen is left in limbo. In this case, without the actual wording of the questions, one cannot really say what people were agreeing with and what they were rejecting. The reader gets only a handful of contradictory statistics and no extra unbiased information to help him or her make sense of those statistics and the world they supposedly portray.
When reporting the results of opinion polls, news agencies and the media ought to describe the polls more thoroughly. Readers should be able to see the questions as well as the answers so that they can understand more fully the views of their fellow citizens and cut through any unsubstantiated rhetoric in the public debate. Without the full picture, the reader can only assume the numbers presented fall into that darkest category of lies: bare statistics.
