In the brief time that remains before the election in December, the media is likely to be full of all kinds of statistics, charts and graphs showing us survey results, voting trends and potential outcomes, and exploring various “what-if” scenarios.
We all know that statistics can be rolled out to demonstrate whatever perspective best suits the politicians’ purposes, or to contradict their opponents; but we don’t always realise how charts and graphics can also be designed to enlighten or to mislead. And, of course, that’s not just the case with political information, but with business information, too.
So let’s just look at three of the most common types of chart, how they can best be used, and some things to look out for to see whether you’re being manipulated.
Perhaps the simplest type of chart is the classic bar graph, which features a series of rectangles or columns of varying lengths, each representing the value corresponding to the categories being compared. The length of each of the columns is proportionate to the value it represents.
These graphs are ideal for comparing any sort of numerical value, including group sizes, inventories, ratings and survey responses, and we often use this type of graph to present financial forecasts and outcomes.
The bars of this type of graph can be displayed horizontally or horizontally, but the standard would be as vertical columns. A typical example of this use would be to show expenditure in different time periods, where the vertical (Y) axis would show the expenditure, while the columns arranged along the horizontal (X) axis would each represent a different time period.
The X axis would be labelled to show whether we’re comparing months, quarters, years etc., while the Y axis would be scaled according to the type of figures we’re showing – a unit representing a penny, a pound, a hundred pounds or a million, depending on what we’re talking about.
The advantage of a simple bar graph is that with one quick glance, audiences can see exactly how the various items size up against one another. Even so, there are a few things to watch out for.
As we’ve seen, the X and Y axes are chosen according to the values we’re looking at, so if there are two or more charts on the same page, they may not be showing the same scales: comparing monthly expenditure for one item with quarterly expenditure for another is like comparing apples and oranges, so it’s important to check the labels on the axes to see precisely what information is on display.
If voter opinion is polled on four political parties, for example, and then only three of the four results are graphed, the resulting chart can be very misleading. By omitting the information about the top scoring party, the one that has polled second can be shown to have a lead even if they are trailing the real leaders by a huge margin. Voters who are considering tactical voting may be persuaded to change their vote without realising that the information they have been given is incomplete.
If the values are to be mapped are all very close – e.g. 104, 110 and 112 – the difference in height of the bars won’t be very noticeable. In this case, the starting point of the Y axis may be moved so the differences can be highlighted. But if the Y axis is set to start at 100 rather than the standard zero, the figures we’ll be looking at on the graph are actually 4,10 and 12. While the difference between the actual values is negligible, the graph shows hugely significant differences.
Line graphs are useful to illustrate trends in data over a period of time and the horizontal (X) axis often displays a timeline. The value – e.g. how many units are sold in each time period – is plotted against the Y axis, and the points connected, resulting in a single line from left to right showing the trend. If we do this for Dataset 1 and then repeat the process for Dataset 2 and produce a line in a different colour, we can compare the two.
The comparison can help determine whether sales of two products are interdependent; for example, if sales for one decrease when sales for the other increase, it suggests that one may be a substitute for the other.
Again, it’s easy to only tell half the story. Omitting weather data on a chart that maps sales of umbrellas and bikinis might make you think that the more umbrellas are sold, the fewer bikinis will be wanted. While such an omission may be intentional, it may also be because the information is unavailable, or simply because nobody has realised that other factors should be considered.
Perhaps the biggest drawback of correlations is assuming cause and effect. There are some marvellous examples of this on Tyler Vigen’s Spurious Correlations website, including the fact that in the ten years to 2009, US spending on science, space and technology correlated almost exactly to the number of suicides by hanging, strangulation and suffocation.
Both these charts consist of circles divided into segments, with the arc of each segment showing the proportional value of each piece of data. They are the simplest and most efficient visual tool for comparing parts of a whole, for example, the market share of the big corporations in a sector.
These charts are frequently used to report market-research question responses: e.g. when offered a choice of A, B or C, 30% chose A, 20% chose B and 50% chose C.
The segments need to add up to 100% – the full circle – so it may be necessary to add in a segment corresponding to “other”, in order to deal with missed questions, spoiled answers etc.
Alternatively, if respondents are allowed to choose more than one answer, the sum of the segments will be more than 100%, so a different form of chart will need to be used for reporting the results.
Because we tend to understand visual data so much more quickly than text or figures, it’s possible to manipulate a pie chart by using inaccurate drawings: if one block in the picture is bigger than another, very few people will look at the actual figures to check it’s drawn to scale.
If there are two or more charts on a page, comparing the answers given to different questions by men and women, for example, it is important to make consistent use of colours to show the men’s answers and the women’s answers. If the colours are swapped, it’s easy for the reader to get confused.
For most of us, visual information is much more quickly apprehended by the brain than text or figures, and proportions and relative sizes make more sense than decontextualised numbers. That means that imagery and graphics are a fundamental way of getting people’s attention and displaying and communicating information about our businesses, about politics, and about the world we live in.
Bar graphs, line graphs, pies and donuts are just a few of the common chart types used. And we’re so familiar with them that we don’t always look more closely to see whether they are being used well or whether we are being lied to – whether that’s unintentionally or with malice aforethought.
Clarity is a vital aspect of communication, but, particularly in the world of politics, it isn’t always in the interests of the person producing the information. It’s important, then, to take responsibility yourself and look more closely at what – and how – data is being presented in order to ensure that you are as well-informed as possible.
And when it comes to communicating information about your own business, it’s a good idea to get the help of an expert to make sure the message is put across clearly and correctly.
If you’ve got a tricky message or complicated information that you need to get across to your stakeholders – clients, staff, investors… – at Tantamount we specialise in simplifying the complex and communicating unfamiliar concepts clearly through well-chosen words and careful design. Why not give us a call on 0798 661 3437?