The Problem With Statistics

The popular media presentation of statistics can be misleading.

How can politicians be expected to come up with the right answers when the information they are given is not honest?

Consider the following “evidence”:

Golf World Top 5 June 2018

1

Justin Thomas

United States

2

Dustin Johnson

United States

3

Justin Rose

England

4

Jon Rahm

Spain

5

Jordan Spieth

United States

This could be sensationalized by claims that you have to have a J as an initial to be a good golfer, and the figures show this claim to be accurate. But it is simply not true. You need many skills both mental and physical to be a world-class golfer but these combined skills do not include your name.

Numbers can help quantify a problem and they can help focus on solutions but they can also be manipulated to distort clarity and to mislead people. One simple way to mislead is to use percentages rather than actual numbers. A simple example. You wish to show that hospitals are overworked and working at maximum. The figures you have: show that in year 1 there were 100 beds and 70 of these were occupied. In Year 2 you had 50 beds and all were occupied. If you publish the actual figures it is clear that the reduction in the number of beds is the cause of any problems. So you say that in year 1, bed occupancy was 70 percent but in year 2 this had grown to 100 percent and you need more resources to deal with this increase in demand.

Road safety can also be “adjusted” in this way; all you need is a slight, but unpublished, change in the definition of a road traffic accident. Change reported incidents to accidents where someone is treated in hospital, and you have an immediate improvement in road safety. Express figures in terms of accidents per thousand vehicle miles gives a different impression from accidents per mile of roadway. This is especially useful since the actual number of vehicle miles is always a guess that cannot be challenged. The usual estimate is based on the total number of registered vehicles multiplied by ten thousand miles, but there are no actual figures that justify this estimate. So adjustments can be made to suit your side of a debate.

A recent example of manipulation of headlines was a well-publicised claim over the number of “shootings” in American schools. Any shooting is bad and should not occur, but including incidents in places that were closed and no longer schools and including drug-related crimes as if they were attacks on pupils is not an honest use of numbers.

Official statistics over the recycling of waste is a minefield. Are the figures those of volume or weight? A ton of empty plastic bottles is a large volume, while a cubic meter of used newspapers is very heavy. Pick your presentation to enhance your personal objective.

Climate change is yet another forest of distortion. Yes, the global climate is changing; it is always changing and always has been changing. The rate at which change is occurring and the direction of that change could be important information for the governance of the whole world but so much adjustment and interpretation or extrapolation is applied to what little actual data we have means the chances of good governance are reduced. To get meaningful rates of change we would need data going back 100,000 years and we do not have that on a worldwide basis. Core samples from the ice at a pole are indicative, as are the geological surveys that say during the previous ages, before humans, areas now deserts were under water. All these show that climate change continuously takes place but it is very hard to grasp the rate of change and even harder to point to naturally occurring events, volcanic eruptions, major earthquakes etc. as affecting the rates of change. Expecting sensible political anticipation of future difficulties is impossible while the data is getting so distorted by the publication of very biased statistics.

Even the media presentations of reasonably accurate data can lead to the misleading of public understanding. Consider an electorate of 1000 voters. An “opinion poll” asks 10 of these whether they support increasing taxes to improve social benefits to the poorest. The poll tries to cover all the social economic groups and ages. They ask two young people with low incomes. These do not pay taxes so they say, "Yes, support the increase of taxes." They ask two wealthy pensioners, who also say yes; because they do not wish to appear mean and lacking in awareness of the social issues. They ask two who are self-employed under 30-years-old and they say no. The poll then asks two from the 50-year-old managerial section of society and these say yes, again because they are very aware of their public “image.” The last two are political activists, one left wing who votes yes and one right wing who says no. The opinion poll then claims that 70 percent of the people will support the raising of taxes. If one political party takes this as a leading aspect of their election campaign they will get a rude awakening. 4 of the 7 who told the opinion poll they supported increasing taxes will not cast secret votes to actually do so. The lower paid and younger sections of the 1000 voters are historically less inclined to vote. If the increase tax party is complacent, because of the 70 percent opinion poll, they will find only half of their supporters bother to vote. At the election vote count, suddenly a large majority reject increasing taxes.