Science: tell me about... distorting facts with graphs

Click to follow
The Independent Online
SHOULD you believe everything you see? The graph (right) appeared recently in a newspaper purporting to be examining the problem of rising waiting lists - especially in the light of Labour's election promise to reduce them by 100,000 between May 1997 and the next election.

On first sight it really looks like Labour, and particularly Frank Dobson, the Health Secretary, have an insuperable task. Look: there is the gradual rise in numbers on the lists; and there on the right is the target, at the bottom of an amazingly steep ramp. Impossible? It certainly looks it.

But look closer, for this is a supreme example of a graph that misinforms. Without knowing if it was intended to tell a lie, a scientist would prefer to say that it distorts the truth. That, it certainly does.

First, look at the lower axis. The solid points - the real data points - are at three-monthly intervals, when the Department of Health released its figures. But after the latest ones, which cover just one year, we suddenly fast-forward, so that the same space that previously covered three months now covers four years - a compression factor of 12 times. No wonder that ramp looks steep.

Now check the left-hand axis. Does it start at 0, so that you can see the true relative difference between present, past and future values? No. To emphasise the changes (both real and promised), a false axis, starting at 1 million, has been used. Again, that makes the upward ramp look steeper than it is in true proportion, and makes the downward one (already radically distorted) look even wilder.

The effect: Labour's election promise to reduce waiting lists by 100,000 (slightly less than 9 per cent) over five years has become a 5 in 1 slope - that is, overstated by a factor of more than 50.

Such tactics are common in propaganda wars. Because we are so used to processing visual information without effort (hence "windows" and "icons" are the currency of computer interfaces), it is always sensible to be suspicious when you are offered a graph as evidence of something.

Do both axes start at sensible points? If different graphs are being shown together, are they really comparable? It is said there are lies, damned lies, and statistics. Don't forget, graphs are statistics, too.

This doesn't, of course, help Mr Dobson, who has a mighty task on his hands. But at least his task isn't as impossible as it's been made to look. Just very, very difficult.

Comments