«Investigating lives, celebrating creativity, honouring ingenuity.»

When Numbers Lie
 

Photo: SXC

MIND MAGAZINE |Every time you open a newspaper, a magazine or a book, you'll see some numbers and stats. They come from somewhere, may it be a national statistics office, an organization or an individual that has done research. These numbers can give a brief overview of the world surrounding ourselves. The way experts collect all the raw data to come up with all these clear-cut numbers is an important part of the process. How we see numbers through our social filter and education plays the most important role when seeing facts. It is the hardest thing to change in one's mind.

Data handling is done by clerks or other people and so can lead to errors (we're all humans, right?). Can you imagine yourself in front of a computer inputting thousands of numbers all day long? In 1999, in Saskatoon (Canada), the fire department was heavily criticized by the city officials because of a slow response time using a new 911 dispatching system. But a month's data in an urban area fire station was compiled in just three days and errors happened in data inputs. The average response time seemed 1 to 2 minutes longer than before. In fact, the real figures were far more better, by a number of seconds( that can save a house from completely burning down).

In 1992, in New York, some molecular biologists deducted that every family tree led to Eve (remember the oldest woman body we found in Africa some time ago). That surprised many people, after all we can't all have our origins in a single person. That strange outcome results from a flaw in the design of the software used to compile the data. The application focused on finding the family tree that had the least mutation in order to track down the same genes. Such an assumption coming from the part of the researcher led to indeed strange results. In fact, genetic codes do mutate slightly from generation to generation, like Mr. Wolpoff from the University of Michigan declared in reaction to the surprising news. Of course, most biases have been corrected since then, but always look behind the numbers.
 

Advertisement

A very frequent error is related to the interpretation we give to some numbers. In his very impressive book The Skeptical Environmentalist, Bjorn Lomborg talks about that. For example, in an impressive 500-page report, an Oxford Scholar (Peter Gleick) assumed that the percentage of people not having access to water would rise in-between 1990 and 2000. In fact, it went down by 10-12%, according to various estimations. When projecting numbers for the 1990s, he simply assumed that no one would newly get access to water in the decade, resulting in very odd numbers. In fact, many more people did gain access to water. That wouldn't be so bad if the number would not have been repeated in various publications and used so widely by organizations to justify all their actions. It is true that we must do something, water is vital after all. But things are getting better, not worse.

You've probably seen polls on the internet. Often, you'll see news anchor asking a question on TV and then telling the viewers to participate. The information is then used on the following day's editorial to supposedly describe people's opinions. This is the best example of biased statistics. Each TV channel and each web site targets a specific audience, whether it be segmented regionally or socio-economically. We have absolutely no background information about the people answering. Does participation really reflect society's thoughts as a whole? These numbers are not reliable because they are not based on anything tangible. That's what I mean when I say that the way data is collected is very important.

Many of us will form opinions when seeing some statistics. But don't forget that some stats are collected on specific objects only, not the accompanying factors. For example, when seeing that thousands and thousands of seals are killed every year, one would quickly oppose himself to these horrible and unnecessary deaths. But we have eliminated the seals' natural predators, and their population is growing without control. It has played a significant role in fish stock reductions throughout the Saint-Laurent Gulf, although pollution is also to blame here.

The worst error is also human, but this time it's related to judgment. Never forget this: the person that handles out the numbers also has political and moral opinions. It is an individual with emotions seeking to convince you of something, just like right now. There's a lot of data out there, and journalists select what they want to. For example, many studies were made on the relationship between silicone implants and breast cancer in the 1990s. On the other hand, studies showing negative relationships between the two were overlooked, such as the 1999 National Academy of Sciences research. What we see in the media is not always true. There are 6 billion opinions on everything from the way rice should be cultivated to whether or not we should drop nuclear power. And out of these, how many are those of the people who choose the information you're looking at on TV.

Don't be afraid to trust numbers when you see them in the newspapers, they're not necessarily flawed or wrong. They could be, however, and you must be suspicious when seeing numbers that look too strange or that look over-exaggerated. One of the threats to the veracity of stats is slowly disappearing: errors in the collection of raw data. Computers and more rigorous research protocols have greatly helped. The other threat, human this one, is not about to disappear! People will always have opinions and feelings on specific things. The good aspect is that indicators like GDP or the poverty rate will exist for a long time and will always give us of an idea of the world surrounding ourselves. But manipulators are always there, somewhere, trying to use supposedly facts at their advantage. Beware...


© MindMagazine.com

MAGAZINE E-MAIL SUBSCRIPTION
-----------------------------
Details | Privacy Policy